AUTOMATIC ATLAS-BASED BUILDING OF POINT DISTRIBUTION MODEL FOR SEGMENTATION OF ANATOMICAL S

合集下载

第四讲(1)ArcHydro

Template Data Model Approaches Water Resources
Floodplain(河漫滩)
WQ
Arc Hydro
Permitting
Arc Hydro Database Definition
A geospatial and temporal data model for water resources
Semi-Generic
Parameterization(参数化)
Model pre and post-processing(模型预处理与后处
理)
Model Specific
Arc Hydro Template Data Model
Template Data Models(模板数据模型)
Arc Hydro Data Model
Arc Hydro Data Model Details
Detailed representation at the end of corresponding
chapters in the book (e.g. p 51)(详细的描述见书上的对应章节)
5 应用范例一
6 应用范例二(时间序列管理)
7 应用范例三(H&H Modeling)
1 ArcHydro
Arc Hydro--Template Data Model(模板数据
模型) Arc Hydro--Tools Key Concepts
Review of geodatabase issues
that operates within ArcGIS(在ArcGIS中可操作的用于水资源的地球空间和时间数据模型) Five key conceptual areas(五大主要的概念领域)

基于Cesium的多尺度城市三维建筑模型可视化

第35卷第3期2021年3月北京测绘Beijing Surveying and MappingVol.35No.3March2021引文格式：陈浩，艾廷华.基于Cesium的多尺度城市三维建筑模型可视化北京测绘，2021,35(3)312316. DOI：10.19580/ki.1007-3000.2021.03.006基于Cesium的多尺度城市三维建筑模型可视化陈浩艾廷华(武汉大学资源与环境科学学院，湖北武汉430079)［摘要］城市空间信息平台的建设既要考虑宏观上的大区域空间格局分析，也要有微观层上的细节化信息分析，因此产生多尺度空间表达的需要。

三维空间数据的多尺度表达,不同于二维数据,不仅要顾及输入数据的粒度特征,还要顾及可视化渲染软件平台的技术特征。

鉴于Cesium成为三维空间数据可视化表达的重要平台，本研究基于Cesium系统平台及其三维可视化空间数据渲染基础功能，提出一种城市建筑物三维地图综合及多尺度可视化的方法，基于空间邻近、形状细节特征构建多层次二维数据基础，然后在多分辨率三维视觉感知条件下，实现三维建筑物的多粒度可视化渲染。

［关键词］三维建筑模型；三维制图综合；Cesium［中图分类号］P208［文献标识码］A 0引言随着我国城市化进程不断推进，交通拥堵、环境恶化、公共基础设施分布不均衡等一系列社会问题日益严峻。

为解决这些城市化问题，智慧城市在云计算、物联网技术、传感器技术的发展下逐步形成，这些互联网技术领域的支撑技术为智慧城市的建设提供了硬件条件,但要更好地发挥其在城市规划、建设、管理上的智慧功能，还需要城市空间数据支持，作为信息系统运行的“血液”，城市空间数据应满足多尺度表达的需要，既要考虑宏观上的大区域空间格局分析，也有微观层上的细节化信息分析。

同时，为服务城市竖向规划、多视角景观分析及全方位的空间体验，智慧城市的空间数据表达应当为多维特征。

在智慧城市的建设当中，三维地图作为表达、传输和研究地理信息的方式和载体［1］,是所有智慧城市管理的基础，更是关键环节。

英汉海洋科学名词

abiological removal 非生物转移abiotic zone 无生命带abrasion platform 海蚀台地absolute salinity 绝对盐度abundance 丰度abyssal circulation 深渊环流abyssal clay 深海粘土abyssal fauna 深渊动物abyssal hill 深海丘陵abyssal plain 深海平原abyssal zone 深渊带abyssopelagic organism 大洋深渊水层生物abyssopelagic plankton 深渊浮游生物abyssopelagic zone 深渊层accessory mark 副轮accretionary prism 增生楔accumulation 堆积作用acoustic remote sensing 声遥感acoustical oceanography 声学海洋学active continental margin 主动大陆边缘aerial remote sensing observation 航空遥感观测African Plate 非洲板块afternoon effect 午后效应Agassiz trawl 阿氏拖网age composition 年龄组成aggregated distribution 集聚分布ahermatypic coral 非造礁珊瑚air gun 气枪air lifting 气举air-born substances 气源物质airborne infrared radiometer 机载红外辐射计air-sea boundary process 海-气边界过程air-sea interaction 海-气相互作用air-sea interface 海-气界面air-tight 气密albedo of sea 海洋反照率"algal chemistry, phycochemistry " 藻类化学algal reef 藻礁alkalinity 碱度allochthonous population 外来种群allopatry 异域分布"alternating current, rectilinear current " 往复流ambient sea noise 海洋环境噪声amphi-boreal distribution 北方两洋分布amphidromic point 无潮点"amphidromic system, amphidrome " 旋转潮波系统amphi-Pacific distribution 太平洋两岸分布anadromic fish 溯河鱼anaerobic zone 厌氧带anaerobiosis 厌氧生活analytical chemistry of sea water 海水分析化学"anchor ice, ground ice " 锚冰anchorage area 锚泊地anchored structure 锚泊结构anomalous sea level 异常水位anoxic basin 缺氧海盆anoxic event 缺氧事件anoxic water 缺氧水"Antarctic Bottom Water, AABW " 南极底层水Antarctic Circumpolar Current 南极绕极流Antarctic Circumpolar Water Mass 南极绕极水团Antarctic Plate 南极洲板块anthropogenic hydrocarbon 人源烃anthropogenic input 人源输入antifouling 防污着aphotic zone 无光带"apparent oxygen utilization, AOU " 表观耗氧量aquaculture 水产养殖aquaculture 水产栽培aquafarm 水产养殖场aquanaut work 潜水作业aquaranch 水中牧场aquatic community 水生群落aquatic ecosystem 水生生态系archipelago 群岛Arctic Ocean 北冰洋"Arctic Water, North Polar Water " 北极水arc-trench-basin system 沟弧盆系armor block 护面块体armored diving 铠装潜水artificial island 人工岛artificial sea water 人工海水aseismic ridge 无震海岭assemblage 组合assimilation efficiency 同化效率assimilation number 同化数association 群聚astronomical tide 天文潮"Atlantic Equatorial Undercurrent, Lomonosov Current " 大西洋赤道潜流Atlantic Ocean 大西洋Atlantic-type coastline 大西洋型岸线Atlantic-type continental margin 大西洋型大陆边缘atmospheric input 大气输入atmospheric sea salt 大气海盐atmospheric transport 大气输送atoll 环礁auricularia larva 耳状幼体Australia-Antarctic Rise 澳大利亚-南极海隆autecology 个体生态学authigenic sediment 自生沉积autoinhibitory substance 自体抑制物质autotroph 自养生物auxotroph 营养缺陷生物average heavy swell 中狂涌average height of the heighest one-tenth wave 1/10 [大波平均]波高average height of the heighest one-third wave 1/3 [大波平均]波高average moderate swell 中中涌axially symmetric marine gravimeter 轴对称式海洋重力仪azimuth correction 方位改正back-arc 弧后back-arc basin 弧后盆地back-arc spreading 弧后扩张backshore 后滨bacterial film 细菌膜bacterial slime 细菌粘膜bacterioneuston 漂游细菌barbor boat 港作船baroclinic ocean 斜压海洋barophilic bacteria 喜压细菌barotropic ocean 正压海洋barrier 沙坝barrier island 沙坝岛barrier reef 堡礁baseline study 基线研究batch culture 一次性培养bathyal fauna 深海动物bathyal zone 深海带bathymetry 水深测量bathypelagic organism 大洋深层生物bathypelagic plankton 深层浮游生物bathypelagic zone 深层beach 海滩beach berm 滩肩beach cusp 滩角beach cycle 海滩旋回beach face 滩面beach nourishment 人工育滩beach profile 海滩剖面beach ridge 滩脊beach rock 海滩岩beam trawl 桁拖网bench 岩滩Benioff zone 贝尼奥夫带benthic community 底栖生物群落benthic division 海底区benthic-pelagic coupling 海底-水层耦合benthology 底栖生物学benthos 底栖生物berth 泊位bioadhesion 生物粘着bioassay 生物测试"biochemical oxygen demand, BOD " 生化需氧量biodegradation 生物降解biodeterioration 生物污染bioerosion 生物侵蚀biofacy 生物相biofouling 生物污着biogenic sediment 生物沉积biogenous hydrocarbon 生源烃biogenous silica 生源硅石biological detritus 生物碎屑biological input 生物输入biological noise 生物噪声biological oceanography 生物海洋学biological purification 生物净化biological removal 生物转移biological scavenging 生物清除bioluminescence 生物发光biomass 生物量bionics 仿生学biosphere 生物圈biota 生物区系biotope 生活小区bioturbation 生物扰动biozone 生物带bipinnaria larva 羽腕幼体bipolarity 两极同源bird-foot delta 鸟足[形]三角洲Bohai Coastal Current 渤海沿岸流Bohai Sea 渤海boomerang sediment corer 自返式沉积物取芯器borate alkalinity 硼酸[盐]碱度"borer, boring organism " 钻孔生物bottom current 底层流bottom friction layer 底摩擦层bottom grab 表层取样器bottom reflection 海底声反射bottom reverberation 海底混响bottom scattering 海底散射bottom water 底层水bottom wave 底波bottom-supported platform 坐底式平台boundary flux 界面通量box corer 箱式取样器box model 箱式模型brackish water species 半咸水种brash ice 碎冰"breaker, surf " 碎波breakwater 防波堤brine 卤水"brown clay, red clay " 褐粘土bubble effect 气泡效应buoyant mat 浮力沉垫burrowing organism 穴居生物caballing [混合]增密caisson 沉箱calcareous ooze 钙质软泥"calcite compensation depth, CCD " 方解石补偿深度calcite dissolution index 方解石溶解指数calm sea 无浪capillary wave 毛细波carbon assimilation 碳同化作用carbon cycle 碳循环carbon dioxide system in sea water 海水二氧化碳系统carbonate alkalinity 碳酸[盐]碱度"carbonate critical depth, CCRD " 碳酸盐极限深度carbonate cycle 碳酸盐旋回carbonate system in sea water 海水碳酸盐系统carcinology 甲壳动物学carnivore 食肉动物catastrophe 灾变catch 渔获量catchability coefficient 可捕系数cathodic protection 阴极防护cellar connection 井口装置Central Indian Ridge 印度洋中脊central rift 中央裂谷central water 中央水chain of volcanoes 火山链"Changjiang Diluted Water, Changjiang River Plume " 长江冲淡水characteristic species 特征种chemical diagenesis 化学成岩作用chemical form 化学形态chemical oceanography 化学海洋学"chemical oxygen demand, COD " 化学需氧量chemical scavenging 化学清除chemical speciation 化学形态分析chemical speciation models 化学形态模型chemical species 化学形式chemical weathering 化学风化作用chemo-autotroph 化能自养生物chemostatic culture 恒化培养"chemotaxis, chemotaxy " 趋化性chemotrophy 化能营养"China Classification Society, ZC " 中国船级社chlorinity 氯度chlorinity ratio 氯度比值chlorosity 氯量chronostratigraphy 年代地层学ciguatoxic fish 西加毒鱼类circumpacific volcanic belt 环太平洋火山带clay 粘土"closed season, prohibited season " 禁渔期cnoidal wave 椭圆余弦波coast of emergence 上升海岸coast of submergence 下沉海岸"coastal current, littoral current " 沿岸流coastal dune 海岸沙丘coastal engineering 海岸工程coastal terrace 海岸阶地coastal water 沿岸水coastal zone 海岸带coastline 海岸线coastline effect 海岸效应coccolith ooze 颗石软泥cofferdam 围堰cold current 寒流cold eddy 冷涡cold water species 冷水种cold water sphere 冷水圈cold water tongue 冷水舌collision zone 碰撞带commensalism 共栖commensalism 偏利共生common species 习见种community 群落community ecology 群落生态学compensation current 补偿流compensation depth 补偿深度compliant structure 顺应式结构composite breakwater 混合式防波堤compound shoreline 复合滨线compound tide 复合潮conchology 贝类学"conductivity-temperature-depth system, CTD " 温盐深仪confused sea 暴涛confused swell 暴涌conservative constituents of sea water 海水保守成分constancy of composition of sea water 海水成分恒定性constituent day 分潮日constituent hour 分潮时constructive boundary 建设性板块边界consumer 消费者continental accretion 大陆增生continental drift 大陆漂移continental margin 大陆边缘continental rise 大陆隆continental shelf 大陆架continental shelf break 大陆架坡折continental slope 大陆坡continental terrace 大陆阶地"continuous cultivation, continuous culture " 连续培养continuous model 连续模型contour current 等深流contourite 等深流沉积[岩]contrast in water 水中对比度contrast transmission in water 水中对比度传输controlled ecosystem experiment 控制生态系实验convective mixing 对流混合conventional diving 常规潜水convergent boundary 会聚边界conversion efficiency 转换效率"copepodite, copepodid larva " 桡足幼体coprophagy 食粪动物coral reef 珊瑚礁coral reef coast 珊湖礁海岸corrosion in sea water 海水腐蚀cosmogenous sediment 宇宙沉积cosmopolitan 世界[广布]种cotidal chart 同潮图countercurrent 逆流crane barge 起重船critical depth 临界深度crop 收获cross-coupling effect 交叉耦合效应current meter 海流计current pattern 流型cuspate bar 尖角坝cuspate delta 尖[形]三角洲cyphonautes larva 苔藓虫幼体cypris larva 腺介幼体Dalmatian coastline 达尔马提亚岸线datum of chart 海图基准面day-night observation 连续观测deck unit 甲板装置deep current 深层流"deep scattering layer, DSL " 深海散射层deep sea fan 深海扇deep sea propagation 深海传播deep sea sand 深海砂deep sea sediment 深海沉积deep sea sound channel 深海声道deep water 深层水deep water wave 深水波delta 三角洲demersal fish 底层鱼类density current 密度流density current 异重流density-dependent mortality 密度制约死亡率deposit feeder 食底泥动物descriptive oceanography 描述海洋学destructive boundary 破坏性板块边界detached breakwater 岛式防波堤detached wharf 岛式码头detritus feeder 食碎屑动物diagonal wave 斜向浪diatom ooze 硅藻软泥"dicycle, dicycly " 双周期"diel vertical migration, diurnal vertical migration " 昼夜垂直移动dilution cycle 稀释旋回directional wave spectrum 方向波谱dissolution cycle 溶解旋回"dissolved inorganic carbon, DIC " 溶解无机碳"dissolved organic carbon, DOC " 溶解有机碳"dissolved organic matter, DOM " 溶解有机物"dissolved organic nitrogen, DON " 溶解有机氮"dissolved organic phosphorus, DOP " 溶解有机磷dissolved oxygen 溶解氧disturbing acceleration 干扰加速度diurnal inequality 日不等[现象]diurnal tide 全日潮diver 潜水员divergent boundary 离散边界diversity 多样性diving suit 潜水服dock 船坞dominant species 优势种"Donghai Coastal Current, East China Sea Coastal Current " 东海沿岸流"Donghai Sea, East China Sea " 东海Doppler current meter 多普勒海流计double diffusion 双扩散double ebb 双低潮double flood 双高潮downwelling 下降流dredge 底栖生物刮底网dredger 挖泥船dredging engineering 疏浚工程drift current 漂流drift ice 流冰drifting buoy 漂流浮标drill conductor 隔水套管drilling vessel 钻探船dry diving 干式潜水duration-limited spectrum 有限风时谱dynamic method 动力方法dynamic positioning 动力定位dynamical oceanography 动力海洋学East African Rift Zone 东非裂谷带East Pacific Rise 东太平洋海隆"ebb, ebb tide " 落潮echinopluteus larva 海胆幼体echo ranging 回声测距echosounder 测深仪ecological barrier 生态障碍ecosystem 生态系edge wave 边缘波efflux 输出通量Ekman depth 埃克曼深度Ekman layer 埃克曼层Ekman pumping 埃克曼抽吸Ekman spiral 埃克曼螺旋Ekman transport 埃克曼输送El Nino ( 西) 厄尔尼诺electrodialysis 电渗析electromagnetic vibration exciter 电磁振荡震源elliptical trochoidal wave 椭圆余摆线波embayed coast 港湾海岸endemic population 地方种群endemic species 地方种endolithion 石内生物endopelos 泥内生物endopsammon 沙内生物energy flow 能流"engineering oceanology, engineering oceanography " 海洋工程水文enhancement 增殖entrainment 卷吸environmental load 环境荷载Eotvos effect 厄特沃什效应ephyra larva 碟状幼体epibenthic sledge 底表撬网epifauna 底表动物epilithion 石面生物epipelagic organism 大洋上层生物epipelagic zone 上层epipelos 泥面生物epiphyte 附生植物epiplankton 上层浮游生物epipsammon 沙面生物Equatorial Countercurrent 赤道逆流Equatorial Current 赤道流"Equatorial Undercurrent, EUC " 赤道潜流equilibrium profile 平衡剖面equilibrium tide 平衡潮equinoctial tide 分点潮equivalent duration 等效风时equivalent fetch 等效风区estuarine chemistry 河口化学estuary 河口湾estuary improvement 河口治理euphotic layer 真光层Eurasian Plate 欧亚板块eurybaric organism 广压性生物eurybathic organism 广深性生物euryhaline species 广盐种euryphagous animal 广食性动物"eurythermal species, eurythermic species " 广温种eustasy 全球性海面升降eutrophic water 富营养水eutrophication 富营养化[作用]euxinic environment 静海环境event deposit 事件沉积exclusive economic zone 专属经济区exogenous organic matter 外源有机物"expendable bathythermograph, XBT " 投弃式温深计exploitative engineering of offshore petroleum/gas reservoir 海上油气开发工程exploratory engineering of offshore petroleum/gas reservoir 海上油气勘探explosive energy source 炸药震源exposed waters 开阔海域failure probability 破坏概率fan delta 扇[形]三角洲fast ice 固定冰fatigue break 疲劳断裂fault coast 断层海岸feather angle 羽角feathering 羽状移动fecal pellet 粪粒fecundity 生殖力feeding migration 索饵洄游fertility 肥力fetch 风区fetch-limited spectrum 有限风区谱fictitious body 假想天体"filter feeder, suspension feeder " 滤食性动物finestructure 细结构fiord 峡湾fish finder 鱼探仪fish resources 鱼类资源fisheries oceanography 渔业海洋学fishery management 渔业管理fishery resources 渔业资源fishing effort 捕捞能力fishing intensity 捕捞强度fishing mortality coefficient 渔捞死亡系数fishing season 渔期fixed oceanographic station 定点观测站fixed structure 固定式结构flare boom 火炬臂"flat coast, low coast " 低平海岸floating breakwater 浮式防波堤floating hose 浮式软管floating structure 浮式结构floating-type wharf 浮式码头floe ice 浮冰"flood, flood tide " 涨潮food chain 食物链food organism 饵料生物food pyramid 食物金字塔food web 食物网foraminiferal ooze 有孔虫软泥fore-arc 弧前fore-arc basin 弧前盆地forerunner 先行涌foreshore 前滨fouling organism 污着生物foundation bed 基床foundation capability 地基承载能力fracture zone 破裂带freshwater plume 淡水舌frictional depth 摩擦深度"fringing reef, shore reef " 岸礁fully developed sea 充分成长风浪gas exploder 气爆震源gateway 峡口general circulation 总环流geographical barrier 地理障碍geological oceanography 地质海洋学"geomagnetic electrokinetograph, GEK " 电磁海流计geostrophic current 地转流geotechnical test 土工试验glacial effect 冰川效应globigerina ooze 抱球虫软泥Gondwana 冈瓦纳古陆gravitational tide 引力潮gravity corer 重力取芯器gravity platform 重力式平台gravity wave 重力波gravity-type structure 重力式结构grazing angle 掠射角groin 丁坝gross primary production 毛初级生产量growth efficiency 生长效率growth overfishing 生长型捕捞过度Gulf Stream 湾流"gulf, bay " 海湾guyed-tower platform 拉索塔平台guyot 平顶海山gyre 流涡habitat 生境"hadal fauna, ultra-abyssal fauna " 超深渊动物"hadal zone, ultra-abyssal zone " 超深渊带half-tide level 半潮面halmyrolysis 海解作用halobiont 盐生生物halocline 盐跃层halophile organism 适盐生物harbor accommodation 港口设施harbor entrance 口门harbor hinterland 港口腹地harbor land area 港口陆域harbor siltation 港口淤积harbour basin 港池harbour site 港址Hardy continuous plankton recorder 哈迪浮游生物记录器harmonic analysis of tide 潮汐调和分析harmonic constant of tide 潮汐调和常数hatchability 孵化率"headland, cape " 岬角heave 垂荡"hekistoplankton, ultraplankton " 超微型浮游生物helium-nitrogen-oxygen saturation diving 氦-氮-氧饱和潜水helium-oxygen diving 氦-氧潜水hemipelagic deposit 半远洋沉积"herbivore, grazer " 食植动物hermatypic coral 造礁珊瑚heterogeneity 异质性heterotroph 异养生物high energy marine environment 海洋高能环境high sea 狂浪"high water, HW " 高潮"highest astronomical tide, HAT " 最高天文潮位holophytic nutrition 全植型营养holoplankton 终生浮游生物homogeneity 同质性homogeneous layer 均匀层horizontal distribution 平面分布hot spot 热点hot spring 海底热泉"Huanghai Coastal Current, Yellow Sea Coastal Current " 黄海沿岸流"Huanghai Cold Water Mass, Yellow Sea Cold Water Mass " 黄海冷水团"Huanghai Sea, Yellow Sea " 黄海"Huanghai Warm Current, Yellow Sea Warm Current " 黄海暖流humification 腐殖化[作用]hummocked ice 堆积冰hydraulic model test 水力模型试验hydraulic piston corer 液压活塞取芯器hydrobiology 水生生物学hydrobiont 水生生物hydrodynamic noise 流体动力噪声hydrothermal circulation 热液循环hydrothermal process 热液过程ice cover 冰盖ice edge 冰缘线ice field 冰原ice period 冰期ice rind 冰壳ice shelf 冰架ice thickness 冰厚iceberg 冰山ichthyology 鱼类学implosive source 聚爆式震源in situ density 现场密度in situ measurement 现场测定in situ salinometer 现场盐度计in situ specific volume 现场比容in situ temperature 现场温度incident wave 入射波"incubation, hatching " 孵化Indian Ocean 印度洋Indian Plate 印度洋板块indicator species 指示种infauna 底内动物influx 输入通量inshore 内滨instanteneous mortality rate 瞬间死亡率interface exchange process 界面交换过程intermediate water 中层水internal tide 内潮internal wave 内波interstitial fauna 间隙动物"interstitial water, pore water " 间隙水intertidal zone 潮间带"Intertropical Convergence Zone, Equatorial " 赤道辐合带intraplate volcanism 板内火山活动inversion layer 逆置层in-vivo fluorescence technique 活体荧光技术ion-exchange membrane 离子交换膜irregular wave 不规则波island 岛island arc 岛弧island shelf 岛架island slope 岛坡isohaline 等盐线isotherm 等温线jacket pile-driven platform 导管架桩基平台jack-up platform 自升式平台jetty 突堤jetty 导堤juvenile 幼年个体Kelvin wave 开尔文波key species 关键种knuckle joint 万向接头Knudsen's burette 克努森滴定管Knudsen's pipette 克努森移液管Knudsen's tables 克努森表Kuroshio 黑潮lag effect 滞后效应lagoon 湖lamellibranchia larva 瓣鳃类幼体land and sea breezes 海陆风land fabrication 陆上预制land-origin ice 陆源冰larva 幼体lateral reflection 侧反射launching 下水Laurasia 劳亚古陆law of the sea 海洋法lead lane 冰间水道level bottom community 平底生物群落level ice 平整冰life support system 生命支持系统light acclimation 光驯化light adaptation 光适性light and dark bottle technique 黑白瓶法light boat 灯船light house 灯塔light saturation 光饱和Lloyd's Register of Shipping 劳埃德船级社long heavy swell 长狂涌long low swell 长轻涌long moderate swell 长中涌long-crested wave 长峰波Longhurst-Hardy plankton recorder 朗-哈浮游生物记录器longshore current 顺岸流"longshore drift, littoral drift " 沿岸泥沙流"low water, LW " 低潮"lowest astronomical tide, LAT " 最低天文潮位luminous organism 发光生物lunar tide 太阴潮lunar tide interval 太阴潮间隙lysis 溶菌lysocline 溶跃层macrobenthos 大型底栖生物macrofauna 大型动物macroplankton 大型浮游生物magnetic lineation 磁条带magnetic quiet zone 磁场平静带main thermocline 主[温]跃层major constituents of sea water 海水主要成分malacology 软体动物学"manganese nodule, ferromanganese nodule " 锰结核mangrove coast 红树林海岸mangrove swamp 红树林沼泽manifold system 管汇系统mantle bulge 地幔隆起mantle convection 地幔对流mantle plume 地幔柱marginal basin 边缘盆地marginal sea 边缘海marginal-type wharf 顺岸码头mariculture 海产养殖mariculture 海产栽培marine accident 海损事故marine acoustics 海洋声学marine aerosol 海洋气溶胶marine bio-acoustics 海洋生物声学marine biochemical resource 海洋生化资源marine biochemistry 海洋生物化学marine biogeochemistry 海洋生物地球化学marine biological noise 海洋生物噪声marine biology 海洋生物学marine chemical resource 海洋化学资源marine chemistry 海洋化学"marine climate, ocean climate " 海洋气候marine climatology 海洋气候学marine contamination 海洋玷污marine corrosion 海洋腐蚀marine detritus 海洋碎屑marine ecology 海洋生态学marine ecosystem 海洋生态系marine element geochemistry 海洋元素地球化学marine engineering geology 海洋工程地质marine environment 海洋环境marine environmental assessment 海洋环境评价marine environmental capacity 海洋环境容量marine environmental chemistry 海洋环境化学"marine environmental forecasting, marine " 海洋环境预报marine environmental monitoring 海洋环境监测marine environmental protection 海洋环境保护marine environmental quality 海洋环境质量marine environmental sciences 海洋环境科学marine erosion 海蚀作用marine geochemistry 海洋地球化学marine geology 海洋地质学marine geomagnetic anomaly 海洋地磁异常marine geomagnetic survey 海洋地磁调查marine geomorphology 海洋地貌学marine geophysical survey 海洋地球物理调查marine geophysics 海洋地球物理学marine gravimeter 海洋重力仪marine gravity anomaly 海洋重力异常marine gravity survey 海洋重力调查marine heat flow survey 海洋地热流调查marine humus 海洋腐殖质"marine hydrography, marine hydrology " 海洋水文学marine installation 海上安装沉放marine isotope chemistry 海洋同位素化学marine meteorology 海洋气象学marine microorganism 海洋微生物marine natural hydrocarbon 海洋天然烃marine natural product 海洋天然产物marine natural product chemistry 海洋天然产物化学marine organic chemistry 海洋有机化学marine organic geochemistry 海洋有机地球化学marine pharmacognosy 海洋生药学marine photochemistry 海洋光化学marine physical chemistry 海洋物理化学marine physics 海洋物理学marine policy 海洋政策marine pollutants 海洋污染物marine pollution 海洋污染marine pressure hydrophone 海洋压力水听器marine reflection seismic survey 海洋反射地震调查marine refraction seismic survey 海洋折射地震调查marine resource chemistry 海洋资源化学marine resources 海洋资源marine salvage 海难救助"marine sciences, ocean sciences " 海洋科学marine sedimentology 海洋沉积学marine seismic profiler 海洋地震剖面仪marine seismic streamer 海洋地震漂浮电缆marine seismic survey 海洋地震调查marine seismograph 海洋地震仪marine stratigraphy 海洋地层学marine technology 海洋技术marine towage 海上拖运marine wide-angle reflection seismic survey 海洋广角反射地震调查maritime air mass 海洋气团marking 标记marsh organism 沼泽生物mass balance 质量平衡mass budget 质量收支mass transfer 质量转移"mean sea level, MSL " 平均海平面"mechanical bathythermograph, MBT " 机械式温深计medical security for diving 潜水医务保障megafauna 巨型动物megalopa larva 大眼幼体megaplankton 巨型浮游生物meiobenthos 小型底栖生物meiofauna 小型动物"meroplankton, transitory plankton " 阶段性浮游生物mesocosm 中型实验生态系mesopelagic fish 中层鱼类mesopelagic organism 大洋中层生物mesopelagic zone 中层mesoplankton 中型浮游生物mesopsammon 沙间生物mesoscale eddy 中尺度涡meteorological tide 气象潮microbenthos 微型底栖生物microbivore 食微生物者microcolony 小菌落microcontinent 微大陆microcosm 小型实验生态系microdistribution 微分布microecosystem 微生态系microfauna 微型动物microfouling 微生物污着microhabitat 微生境micronutrients 微量营养物microplankton 小型浮游生物microstructure 微结构Mid-Atlantic Ridge 大西洋中脊mid-ocean ridge 洋中脊mid-ocean ridge basalt 洋中脊玄武岩midwater trawl 中层拖网migratory fish 洄游鱼类minimum duration 最小风时minimum fetch 最小风区minor elements of sea water 海水微量元素mirage 蜃景mixed layer sound channel 混合层声道"mixed layer, mixing layer " 混合层mixed tide 混合潮mixotroph 混合营养生物mobile platform 移动式平台moderate sea 中浪module 模块"monocycle, monocycly " 单周期monophagy 单食性monsoon current 季风海流moored data buoy 锚定资料浮标mooring facilities 系泊设施mooring force 系泊力mortality 死亡率mound-type breakwater 斜坡式防波堤mud 泥muddy coast 泥质海岸multibeam echosounder 多波束测深仪multi-point mooring 多点系泊multistage flash distillation 多级闪急蒸馏multistage separator 多级分离器mysis larva 糠虾期幼体N/P ratio 氮磷比[值]"Nanhai Coastal Current, South China Sea Coastal Current " 南海沿岸流"Nanhai Sea, South China Sea " 南海"Nanhai Warm Current, South China Sea Warm Current " 南海暖流nannoplankton 微型浮游生物nauplius larva 无节幼体navigation channel 航道navigation equipment 导航设备neap tide 小潮nearshore zone 近滨带nectochaeta larva 疣足幼体nektobenthos 游泳底栖生物nekton 游泳生物nepheloid 雾状层neritic organism 近海生物neritic sediment 浅海沉积neritic zone 浅海带neritic zone 近海区net plankton 网采浮游生物net primary production 净初级生产量net primary productivity 净初级生产力neurotoxin 神经毒素niche 生态位Ninety East Ridge 东经90度洋中脊Niskin water sampler 尼斯金采水器nitrogen cycle 氮循环nitrogen-oxygen diving 氮-氧潜水no swell 无涌non-conservative constituents of sea water 海水非保守成分nonharmonic constant of tide 潮汐非调和常数non-saturation diving 非饱和潜水Norpac net 北太浮游生物网North American Plate 北美洲板块"North Atlantic Deep Water, NADW " 北大西洋深层水not fully developed sea 未充分成长风浪nursing ground 育幼场nutrient depletion 营养[盐]耗竭nutrients in sea water 海水营养盐obduction plate 仰冲板块obduction zone 仰冲带oblique haul 斜拖observation platform 观测平台ocean 洋ocean basin 洋盆ocean bottom seismograph 海底地震仪ocean circulation 大洋环流ocean color scanner 海色扫描仪ocean current 海流ocean current energy 海流能ocean energy conversion 海洋能转换ocean energy resources 海洋能源ocean engineering 海洋工程ocean exploitation 海洋开发ocean management 海洋管理ocean observation technology 海洋观测技术"ocean optics, marine optics " 海洋光学ocean power generation 海洋能发电ocean salinity energy 海洋盐差能ocean thermal energy 海洋温差能ocean wave 海浪ocean wave spectrum 海浪谱ocean-atmosphere heat exchange 海气热交换oceanic crust 洋壳oceanic front 海洋锋oceanic optical remote sensing 海洋光学遥感oceanic plate 大洋板块oceanic sound scatterer 海洋声散射体oceanic tholeiite 大洋拉斑玄武岩oceanic troposphere 大洋对流层oceanic turbulence 海洋湍流oceanic zone 大洋区oceanization 大洋化作用"oceanographic survey, oceanographic investigation " 海洋调查"oceanography, oceanology " 海洋学offshore 外滨offshore bar 滨外坝offshore engineering 近海工程offshore loading and unloading system 海上装卸油系统offshore oil-gas flowline 海上输油气管线offshore platform 近海平台offshore storage unit 海上贮油装置oil fence [围]油栅oil-gas-water treating system 油气水处理系统oligohaline species 寡盐种oligostenohaline species 低狭盐种oligotaxic ocean 少种型大洋oligotrophic water 贫营养水omnivore 杂食动物ooze 软泥ophiopluteus larva 长腕幼体opportunistic species 机会种optimum catch 最适渔获量organic coating layer 有机覆盖层overfishing 捕捞过度overlying water 上覆水overpopulation 种群过密overtide 倍潮overwintering 越冬oxide film 氧化膜oxygen maximum layer 氧最大层oxygen minimum layer 氧最小层oxygen partial pressure 氧分压Oyashio 亲潮oyster reef 牡蛎礁"Pacific Equatorial Undercurrent, Cromwell Current " 太平洋赤道潜流Pacific Ocean 太平洋Pacific Plate 太平洋板块Pacific-type coastline 太平洋型岸线Pacific-type continental margin 太平洋型大陆边缘pack ice 浮冰群paleoceanography 古海洋学paleocurrent 古海流paleodepth 古深度paleomagnetic stratigraphy 古地磁地层学paleoproductivity 古生产力paleosalinity 古盐度Pangaea 泛大陆Panthalassa 泛大洋parallel dike 顺坝parasitism 寄生"particulate inorganic carbon, PIC " 颗粒无机碳particulate matter in sea water 海水颗粒物"particulate organic carbon, POC " 颗粒有机碳"particulate organic matter, POM " 颗粒有机物"particulate organic nitrogen, PON " 颗粒有机氮"particulate organic phosphorus, POP " 颗粒有机磷passive continental margin 被动大陆边缘patch reef 点礁patchiness 斑块分布pediveliger larva 具足面盘幼体pelagic deposit 远洋沉积pelagic division 水层区pelagic egg 浮性卵pelagic fish 上层鱼类pelagic organism 水层生物pelagic organism 大洋生物pelagic phase 浮性生活期peleotemperature 古温度peninsula 半岛periphyton 周丛生物permanent thermocline 永久性温跃层phaeophytin 脱镁叶绿素phosphorus cycle 磷循环photo-autotroph 光能自养生物photobacteria 发光细菌photochemical transformation 光化学转化photophilous organism 适光生物photosynthetic activity 光合活性"phototaxis, phototaxy " 趋光性phycology 藻类学phyllosoma larva 叶状幼体physical oceanography 物理海洋学phytoplankton 浮游植物pile group 群桩pile-driving barge 打桩船pilidium larva 帽状幼体pipe-laying ship 敷管船piston corer 活塞取芯器pitch 纵摇planktobacteria 浮游细菌plankton 浮游生物plankton equivalent 浮游生物当量plankton indicator 浮游生物指示器plankton net 浮游生物网plankton pump 浮游生物泵plankton recorder 浮游生物记录器"planktonology, planktology " 浮游生物学planula larva 浮浪幼体plate 板块plate boundary 板块边界plate collision 板块碰撞plate convergence 板块会聚plate tectonics 板块构造学pleuston 漂浮生物plunging breaker 卷碎波poikilotherm 变温动物Poincare wave 庞加莱波polar ice 极地冰pollutant 污染物polymetal crust 多金属结壳polymorphism 多态现象polyphagy 复食性polystenohaline species 高狭盐种polytaxic ocean 多种型大洋population 种群population dynamics 种群动态population ecology 种群生态学porcellana larva 磁蟹幼体porosity 孔隙度"port engineering, harbor engineering " 港口工程post-larva 稚期practical salinity 实用盐度practical salinity scale 1978 1978 实用盐标precipitous sea 怒涛predation 捕食[现象]predator 捕食者preformed nutrients 原存营养盐pressure-relief tank 减压舱pressurized compartment 加压舱prey 猎物primary production 初级生产量primary productivity 初级生产力producer 生产者。

GCAT-U-Net嵌入全局坐标注意力机制的遥感地块分割网络

计算机测量与控制．２０２２．３０（２）　犆狅犿狆狌狋犲狉犕犲犪狊狌狉犲犿犲狀狋牔犆狅狀狋狉狅犾　·２２２　·收稿日期：２０２１０９３０；　修回日期：２０２１１０２６。

作者简介：苏　耀（１９９５），男，山西运城人，硕士研究生，主要从事模糊数学与人工智能方向的研究。

通讯作者：周　伟（１９８９），男，湖南益阳人，博士研究生，主要从事模式识别方向的研究。

引用格式：苏　耀，于　濂，周　伟．ＧＣＡＴ－Ｕ－Ｎｅｔ嵌入全局坐标注意力机制的遥感地块分割网络［Ｊ］．计算机测量与控制，２０２２，３０（２）：２２２２２８，２３６．文章编号：１６７１４５９８（２０２２）０２０２２２０７ＤＯＩ：１０．１６５２６／ｊ．ｃｎｋｉ．１１－４７６２／ｔｐ．２０２２．０２．０３２中图分类号：ＴＰ１８３文献标识码：Ａ犌犆犃犜－犝－犖犲狋嵌入全局坐标注意力机制的遥感地块分割网络苏　耀１，于　濂１，周　伟２（１．北京师范大学数学科学学院，北京　１００８７５；２．北京师范大学互联网教育智能技术及应用国家工程实验室，北京　１００８７５）摘要：遥感影像的地块背景特征复杂，当前地块分割方法不能较好地处理模糊的边缘信息，导致分割精度不理想；文章利用注意力机制处理地块特征，提出了一种基于全局坐标注意力机制的遥感地块分割网络：ＧＣＡＴ－Ｕ－Ｎｅｔ；该方法在Ｕ－Ｎｅｔ网络基础上嵌入了全局坐标注意力机制，加强了深度神经网络对于遥感影像数据中重要特征的关注度；在公开的ＧＩＤ数据集上的实验结果表明，文章提出的模型将准确率从０．９０４１提升到了０．９２２７，比传统Ｕ－Ｎｅｔ网络提高了２百分点；结合特征自身重要性和特征位置信息的全局坐标注意力机制有助于更精确的目标定位，其输出相较于嵌入单一注意力机制，地块边界更为清晰，提升效果更为显著。

关键词：地块语义分割；注意力机制；模式识别；Ｕ－Ｎｅｔ网络；卷积神经网络犌犆犃犜－犝－犖犲狋：犚犲犿狅狋犲犛犲狀狊犻狀犵犘犾狅狋犛犲犵犿犲狀狋犪狋犻狅狀犖犲狋狑狅狉犽狑犻狋犺犌犾狅犫犪犾犆狅狅狉犱犻狀犪狋犲犃狋狋犲狀狋犻狅狀犕犲犮犺犪狀犻狊犿ＳＵＹａｏ１，ＹＵＬｉａｎ１，ＺＨＯＵＷｅｉ２（１．ＳｃｈｏｏｌｏｆＭａｔｈｅｍａｔｉｃａｌＳｃｉｅｎｃｅｓ，ＢｅｉｊｉｎｇＮｏｒｍａｌＵｎｉｖｅｒｓｉｔｙ，Ｂｅｉｊｉｎｇ　１００８７５，Ｃｈｉｎａ；２．ＮａｔｉｏｎａｌＥｎｇｉｎｅｅｒｉｎｇＬａｂｏｒａｔｏｒｙｆｏｒＣｙｂｅｒｌｅａｒｎｉｎｇａｎｄＩｎｔｅｌｌｉｇｅｎｔＴｅｃｈｎｏｌｏｇｙ，ＢｅｉｊｉｎｇＮｏｒｍａｌＵｎｉｖｅｒｓｉｔｙ，Ｂｅｉｊｉｎｇ　１００８７５，Ｃｈｉｎａ）犃犫狊狋狉犪犮狋：Ｔｈｅｐｌｏｔｓｂａｃｋｇｒｏｕｎｄｆｅａｔｕｒｅｓｉｎｒｅｍｏｔｅｓｅｎｓｉｎｇｉｍａｇｅｓａｒｅｃｏｍｐｌｅｘ．Ｔｈｅｃｕｒｒｅｎｔｐｌｏｔｓｅｇｍｅｎｔａｔｉｏｎｍｅｔｈｏｄｓｃａｎｎｏｔｐｒｅｆｅｒａｂｌｙｈａｎｄｌｅｔｈｅｆｕｚｚｙｅｄｇｅｉｎｆｏｒｍａｔｉｏｎ，ｓｏｔｈｅｓｅｇｍｅｎｔａｔｉｏｎａｃｃｕｒａｃｙｉｓｎｏｔｉｄｅａｌ．Ｔｈｅａｔｔｅｎｔｉｏｎｍｅｃｈａｎｉｓｍｉｓｕｓｅｄｔｏｐｒｏｃｅｓｓｔｈｅｌａｎｄｐａｒｃｅｌｆｅａｔｕｒｅｓ，ａｎｄａｒｅｍｏｔｅｓｅｎｓｉｎｇｌａｎｄｐａｒｃｅｌｓｅｇｍｅｎｔａｔｉｏｎｎｅｔｗｏｒｋｂａｓｅｄｏｎｔｈｅｇｌｏｂａｌｃｏｏｒｄｉｎａｔｅａｔｔｅｎｔｉｏｎｍｅｃｈａｎｉｓｍｉｓｐｒｏｐｏｓｅｄ．ＴｈｉｓｍｅｔｈｏｄｅｍｂｅｄｓｔｈｅｇｌｏｂａｌｃｏｏｒｄｉｎａｔｅａｔｔｅｎｔｉｏｎｍｅｃｈａｎｉｓｍｏｎｔｈｅｂａｓｉｓｏｆＵ－Ｎｅｔｎｅｔｗｏｒｋ，ｗｈｉｃｈｓｔｒｅｎｇｔｈｅｎｓｔｈｅｄｅｅｐｎｅｕｒａｌｎｅｔｗｏｒｋｓａｔｔｅｎｔｉｏｎｔｏｉｍｐｏｒｔａｎｔｆｅａｔｕｒｅｓｉｎｒｅｍｏｔｅｓｅｎｓｉｎｇｉｍａｇｅｄａｔａ．ＴｈｅｅｘｐｅｒｉｍｅｎｔａｌｒｅｓｕｌｔｓｏｎｔｈｅｐｕｂｌｉｃＧＩＤｄａｔａｓｅｔｓｈｏｗｔｈａｔｔｈｅｍｏｄｅｌｐｒｏｐｏｓｅｄｉｎｔｈｅａｒｔｉｃｌｅｉｎｃｒｅａｓｅｓｔｈｅａｃｃｕｒａｃｙｆｒｏｍ０．９０４１ｔｏ０．９２２７，ｗｈｉｃｈｉｓ２％ｈｉｇｈｅｒｔｈａｎｔｈｅｔｒａｄｉｔｉｏｎａｌＵ－Ｎｅｔｎｅｔｗｏｒｋ．Ｔｈｅｇｌｏｂａｌｃｏｏｒｄｉｎａｔｅａｔｔｅｎｔｉｏｎｍｅｃｈａｎｉｓｍｔｈａｔｃｏｍｂｉｎｅｓｔｈｅｆｅａｔｕｒｅｉｍｐｏｒｔａｎｃｅｉｔｓｅｌｆａｎｄｔｈｅｆｅａｔｕｒｅｌｏｃａｔｉｏｎｉｎｆｏｒｍａｔｉｏｎｉｓｈｅｌｐｆｕｌｆｏｒｍｏｒｅａｃｃｕｒａｔｅｔａｒｇｅｔｐｏｓｉｔｉｏｎｉｎｇ．Ｃｏｍｐａｒｅｄｗｉｔｈｔｈｅｅｍｂｅｄｄｅｄｓｉｎｇｌｅａｔｔｅｎｔｉｏｎｍｅｃｈａｎｉｓｍ，ｔｈｅｐａｒｃｅｌｂｏｕｎｄａｒｙｉｓｍｏｒｅｃｌｅａｒｅｒａｎｄｔｈｅｉｍｐｒｏｖｅｍｅｎｔｅｆｆｅｃｔｉｓｍｏｒｅｓｉｇｎｉｆｉｃａｎｔ．犓犲狔狑狅狉犱狊：ｐａｒｃｅｌｓｅｍａｎｔｉｃｓｅｇｍｅｎｔａｔｉｏｎ；ａｔｔｅｎｔｉｏｎｍｅｃｈａｎｉｓｍ；ｐａｔｔｅｒｎｒｅｃｏｇｎｉｔｉｏｎ；Ｕ－Ｎｅｔｎｅｔｗｏｒｋ；ｃｏｎｖｏｌｕｔｉｏｎｎｅｕｒａｌｎｅｔｗｏｒｋ投稿网址：ｗｗｗ．ｊｓｊｃｌｙｋｚ．ｃｏｍCopyright©博看网 . All Rights Reserved.第２期苏　耀，等：ＧＣＡＴ－Ｕ－Ｎｅｔ嵌入全局坐标注意力机制的遥感地块分割网络·２２３　·０　引言耕地的数量和质量是保持农业可持续发展的关键，利用高分辨率的卫星遥感影像［１］可以识别并获取耕地区域，准确的耕地分布能够为国家决策部门提供重要支撑。

brainR包说明书

Package‘brainR’October12,2022Type PackageTitle Helper Functions to'misc3d'and'rgl'Packages for BrainImagingVersion1.6.0Date2019-12-03Maintainer John Muschelli<*********************>Description This includes functions for creating3D and4D images using'WebGL','rgl',and'JavaScript'commands.This package relies on the X toolkit('XTK',<https:///xtk/X#readme>).License GPL-2LazyData trueDepends rgl,misc3d,oro.niftiImports grDevicesRoxygenNote7.0.1Encoding UTF-8Suggests servrNeedsCompilation noAuthor John Muschelli[aut,cre]Repository CRANDate/Publication2019-12-0506:50:02UTCR topics documented:brainR-package (2)makeScene (2)scene4d (3)write4D (4)write4D.ﬁle (6)writeTrianglesSTL (7)writeWebGL_split (8)12makeScene Index10 brainR-package Functions to render3D brain images in htmlDescriptionBrain Template from Copyright(C)1993-2009Louis Collins,McConnell Brain Imaging Centre, Montreal Neurological Institute,McGill University6th generation non-linear symmetric brain Author(s)John Muschelli<*********************>ReferencesG.Grabner,A.L.Janke,M.M.Budge,D.Smith,J.Pruessner,and D.L.Collins,"Symmetricatlasing and model based segmentation:an application to the hippocampus in older adults",Med Image Comput Comput Assist Interv Int Conf Med Image Comput Comput Assist Interv,vol.9, pp.58-66,2006.See Alsocontour3d,rglmakeScene Make Leveled SceneDescriptionMake scene returns a list of levels-but makes them mutually distinct.So if cutoff0.1,0.2,then0.1<=x<0.2is an roi,not>0.1and>0.2.Different than contour3d as these are mutually exclusivelevels.UsagemakeScene(data,cutoffs,alpha,cols)Argumentsdata-3D array of values(can be nifti-class)cutoffs-series of levels to be createdalpha-alpha levels for each contourcols-colors for each contourValuescene with multiple objects-can be passed to write4Dscene4d3 scene4d Wrapper to write a4D sceneDescriptionThis function takes inﬁlenames,levels,and creates an output htmlﬁle,with4D elements.The html is based on XTK(https:///xtk/X#readme)Usagescene4d(files,fnames=NULL,outfile="index_4D_stl.html",levels=NULL,alpha=NULL,color="white",useTemp=FALSE,MNITemp=c("1mm","2mm"),objtype="stl",...)Argumentsfiles(character)vector ofﬁlenames(ﬁrst being a brainﬁle if useTemp=FALSE) fnames(character)ﬁlenames for the3D surfaces in the scene-needs to be the same length asﬁlesoutfile(character)htmlﬁlenamelevels(numeric/list)levels to make contours/surfaces for eachﬁle.Either a numeric vector may be passed,one level for eachﬁle.Or a list of numeric vectors ofmultiple levels for eachﬁle.Will be coerced to a list.alpha(numeric/list)alpha opacities for each contours/surface for eachﬁle.Will be coerced to list similarly as levelscolor(character/list)colors for each contours/surface for eachﬁle.Will be coerced to list similarly as levelsuseTemp(logical)whether to use template from brainR as the brainﬁgureMNITemp(character)if(useTemp=TRUE)either"1mm"or"2mm"denoting the resolu-tion of the template usedobjtype(character)object type to write theﬁles to.Either"stl","obj",or"ply"to write....other options to be passed to write4DExamples###Faster-8mm resampled but very coarseimgs<-paste("Visit_",1:5,"_8mm.nii.gz",sep="")ifiles<-sapply(imgs,system.file,package= brainR )files=file.path(tempdir(),basename(ifiles))file.copy(ifiles,files)outfile<-file.path(tempdir(),"index_4D_stl.html")scene4d(files,levels=rep(0.99,length(files)),outfile=outfile,color=rep("blue",length(files)),useTemp=TRUE,MNITemp="8mm",alpha=rep(1,length(files)),rescale=TRUE)##Not run:imgs<-paste("Visit_",1:5,".nii.gz",sep="")ifiles<-sapply(imgs,system.file,package= brainR )files=file.path(tempdir(),basename(ifiles))file.copy(ifiles,files)scene4d(files,levels=rep(0.99,length(files)),outfile=outfile,color=rep("blue",length(files)),useTemp=TRUEge,MNITemp="8mm",alpha=rep(1,length(files)),rescale=TRUE)##End(Not run)write4D Wrapper to write a4D sceneDescriptionThis function takes in a scene and writes it out to a series ofﬁles either with the stl format or obj format(see writeOBJ and writeSTL)Usagewrite4D(scene,outfile,fnames=NULL,captions=NULL,writefiles=TRUE,reprint=TRUE,...)Argumentsscene list of3D triangles(see contour3d).If a multicolored object is to be rendered (multiple contours with one control)-it must be in a listoutfile htmlﬁlename that is to be exportedfnamesﬁlenames for the3D surfaces in the scene-needs to be the same length as scenecaptions labels for checkboxes on html webpagewritefiles(experimental)simply run the code to create the html and not write the.obj or .stlﬁlesreprint(logical,experimental)do you want to reprint the rgl before saving(common use by rgl functions)...other options to be passed to write4D.ﬁleExamples#Brain Template from Copyright(C)1993-2009Louis Collins,#McConnell Brain Imaging Centre,#Montreal Neurological Institute,McGill University#6th generation non-linear symmetric brain##Downsampled to8mm using FSL fslmaths-subsamp2template<-readNIfTI(system.file("MNI152_T1_8mm_brain.nii.gz",package="brainR"),reorient=FALSE)dtemp<-dim(template)###4500-value that empirically value that presented a brain with gyri###lower values result in a smoother surfacebrain<-contour3d(template,x=1:dtemp[1],y=1:dtemp[2],z=1:dtemp[3],level=4500,alpha=0.8,draw=FALSE)###Example data courtesy of Daniel Reich###Each visit is a binary mask of lesions in the brainimgs<-paste("Visit_",1:5,"_8mm.nii.gz",sep="")files<-sapply(imgs,system.file,package= brainR )scene<-list(brain)##loop through images and threshnimgs<-length(imgs)cols<-rainbow(nimgs)for(iimg in1:nimgs){mask<-readNIfTI(files[iimg],reorient=FALSE)if(length(dim(mask))>3)mask<-mask[,,,1]###use0.99for level of mask-binaryactivation<-contour3d(mask,level=c(0.99),alpha=1,add=TRUE,color=cols[iimg],draw=FALSE)##add these triangles to the listscene<-c(scene,list(activation))}##make output image names from image namesfnames<-c("brain.stl",gsub(".nii.gz",".stl",imgs,fixed=TRUE))fnames=file.path(tempdir(),fnames)outfile<-file.path(tempdir(),"index.html")write4D(scene=scene,fnames=fnames,outfile=outfile,standalone=TRUE,rescale=TRUE)if(interactive()){if(requireNamespace("servr",quietly=TRUE)){servr::httd(tempdir())}}6write4D.ﬁle unlink(outfile)write4D.file Write a4D sceneDescriptionThis function takes in a scene and writes it out to a series ofﬁles either with the stl format or obj formatUsagewrite4D.file(scene=NULL,outfile="index_4D.html",fnames,visible=TRUE,opacity=1,colors=NULL,captions="",standalone=FALSE,rescale=FALSE,index.file=system.file("index_template.html",package="brainR"),toggle="checkbox",xtkgui=FALSE)Argumentsscene-list of3D triangles(see contour3d).If a multicolored object is to be rendered (multiple contours with one control)-it must be in a listoutfile-htmlﬁlename that is to be exportedfnames-ﬁlenames for the3D surfaces in the scene-needs to be the same length as scenevisible-logical vector indicating which structures are visible in htmlﬁleopacity-list of alpha values-same length as scene;if sub-structures are present,then the each list element has length the number of structurescolors-character vector of colors(col2rgb is applied)captions-labels for checkboxes on html webpagestandalone-logical-should this be able to be rendered ofﬂine?rescale-rescale the scene?-in betaindex.file-template htmlﬁle usedtoggle-(experimental)"checkbox"(default)or"radio"for radio or checkboxes to switch thingxtkgui-(experimental)Logical to use xtkgui for objectswriteTrianglesSTL7 See AlsowriteOBJ,writeSTL,contour3dExamplestemplate<-readNIfTI(system.file("MNI152_T1_8mm_brain.nii.gz",package="brainR"),reorient=FALSE)dtemp<-dim(template)###4500-value that empirically value that presented a brain with gyri###lower values result in a smoother surfacebrain<-contour3d(template,x=1:dtemp[1],y=1:dtemp[2],z=1:dtemp[3],level=4500,alpha=0.8,draw=FALSE)###Example data courtesy of Daniel Reich###Each visit is a binary mask of lesions in the brainimgs<-paste("Visit_",1:5,"_8mm.nii.gz",sep="")files<-sapply(imgs,system.file,package= brainR )scene<-list(brain)##loop through images and threshnimgs<-length(imgs)cols<-rainbow(nimgs)for(iimg in1:nimgs){mask<-readNIfTI(files[iimg],reorient=FALSE)if(length(dim(mask))>3)mask<-mask[,,,1]###use0.99for level of mask-binaryactivation<-contour3d(mask,level=c(0.99),alpha=1,add=TRUE,color=cols[iimg],draw=FALSE)##add these triangles to the listscene<-c(scene,list(activation))}##make output image names from image namesfnames<-c("brain.stl",gsub(".nii.gz",".stl",imgs,fixed=TRUE))fnames=file.path(tempdir(),fnames)outfile<-file.path(tempdir(),"index.html")write4D.file(scene=scene,fnames=fnames,visible=FALSE,outfile=outfile,standalone=TRUE,rescale=TRUE)unlink(outfile)unlink(fnames)writeTrianglesSTL Write STL triangles(without recalling the ids)DescriptionThis is code extracted from writeSTL in rgl.This allows users to write the triangles in STL without reprinting the rgl(which takes time)UsagewriteTrianglesSTL(scene,con,ascii=FALSE)Argumentsscene list of triangles(that have class Triangles3D)conﬁlename or connection of stlﬁle to writeascii indicator if theﬁle should be written in ascii or binaryValueﬁlename(invisible)of stl objectwriteWebGL_split Write WebGL with split trianglesDescriptionAdapted writeWebGL function that splits the triangles into65535verticesUsagewriteWebGL_split(ids=rgl.ids()$id,writeIt=TRUE,verb=FALSE,...)Argumentsids-rgl ids(see rgl.ids)writeIt-(logical)write theﬁle outverb-verbose output...-further arguments passed to writeWebGLValueif writeIt is TRUE,then returns the value from writeWebGL.Otherwise,returns the split triangles from the rgl objectsExamples##Not run:#Brain Template from Copyright(C)1993-2009Louis Collins,#McConnell Brain Imaging Centre,#Montreal Neurological Institute,McGill University#6th generation non-linear symmetric braintemplate<-readNIfTI(system.file("MNI152_T1_2mm_brain.nii.gz",package="brainR") ,reorient=FALSE)dtemp<-dim(template)###4500-value that empirically value that presented a brain with gyri###lower values result in a smoother surfacebrain<-contour3d(template,x=1:dtemp[1],y=1:dtemp[2],z=1:dtemp[3],level=4500,alpha=0.1,draw=FALSE)drawScene.rgl(brain)###this would be the activation or surface you want to render-#hyper-intense white mattercontour3d(template,level=c(8200,8250),alpha=c(0.5,0.8),add=TRUE,color=c("yellow","red"))###add texttext3d(x=dtemp[1]/2,y=dtemp[2]/2,z=dtemp[3]*0.98,text="Top")text3d(x=-0.98,y=dtemp[2]/2,z=dtemp[3]/2,text="Right")fname<-"knitted_webGL.html"writeWebGL_split(dir=getwd(),filename=fname,template=system.file("my_template.html",package="brainR"),width=500,writeIt=TRUE)browseURL(fname)##End(Not run)Index∗brainRbrainR-package,2∗packagebrainR-package,2brainR(brainR-package),2brainR-package,2contour3d,2,4,6,7makeScene,2nifti-class,2rgl,2rgl.ids,8scene4d,3write4D,2,3,4write4D.file,5,6writeOBJ,4,7writeSTL,4,7,8writeTrianglesSTL,7writeWebGL,8writeWebGL_split,810。

基于改进ＤｅｅｐＬａｂＶ３＋的引导式道路提取方法及在震源点位优化中的应用

２０２４年３月第３９卷第２期西安石油大学学报（自然科学版）ＪｏｕｒｎａｌｏｆＸｉ’ａｎＳｈｉｙｏｕＵｎｉｖｅｒｓｉｔｙ（ＮａｔｕｒａｌＳｃｉｅｎｃｅＥｄｉｔｉｏｎ）Ｍａｒ．２０２４Ｖｏｌ．３９Ｎｏ．２收稿日期：２０２３０６０３基金项目：国家自然科学基金面上项目“基于频变信息的流体识别及流体可动性预测”（４１７７４１４２）；四川省重点研发项目“工业互联网安全与智能管理平台关键技术研究与应用”（２０２３ＹＦＧ０１１２）；四川省自然科学基金资助项目“基于超分辨感知方法的密集神经图像分割”（２０２２ＮＳＦＳＣ０９６４）第一作者：曹凯奇（１９９８），男，硕士，研究方向：遥感图像标注。

Ｅｍａｉｌ：８１９０８８３３８＠ｑｑ．ｃｏｍ通讯作者：文武（１９７９），男，博士，研究方向：人工智能在地球科学的应用、高性能计算。

Ｅｍａｉｌ：ｗｅｎｗｕ＠ｃｕｉｔ．ｅｄｕ．ｃｎＤＯＩ：１０．３９６９／ｊ．ｉｓｓｎ．１６７３０６４Ｘ．２０２４．０２．０１６中图分类号：ＴＥ１９文章编号：１６７３０６４Ｘ（２０２４）０２０１２８１５文献标识码：Ａ基于改进ＤｅｅｐＬａｂＶ３＋的引导式道路提取方法及在震源点位优化中的应用曹凯奇１，张凌浩２，徐虹１，吴蔚３，文武１，周航１（１．成都信息工程大学计算机学院，四川成都６１０２２５；２．国网四川省电力公司电力科学研究院，四川成都６１００９４；３．中国石油集团东方地球物理勘探有限责任公司采集技术中心，河北涿州０７２７５０）摘要：为解决自动识别方法在道路提取时存在漏提、错提现象，提出一种引导式道路提取方法提高修正效率。

在ＤｅｅｐＬａｂＶ３＋原有输入通道（３通道）的基础上添加额外输入通道（第４通道），将道路的４个极点转化为二维高斯热图后作为额外通道输入网络，网络以极点作为引导信号，使网络适用于引导式道路提取任务；设计并行多分支模块，提取上下文信息，增强网络特征提取能力；融合类均衡二值交叉熵和骰子系数组成新的复合损失函数进行训练缓解正负样本不均衡问题。

DC-CBAM-UNet++网络的肺结节图像分割方法

第 22卷第 7期2023年 7月Vol.22 No.7Jul.2023软件导刊Software GuideDC-CBAM-UNet++网络的肺结节图像分割方法徐微，汤俊伟，张驰（1.湖北省服装信息化工程技术研究中心； 2.武汉纺织大学计算机与人工智能学院，湖北武汉 430200）摘要：针对肺结节图像存在体积较小、形状不规则、边缘模糊，导致模型特征提取困难及分割精度不高等问题，提出一种基于UNet++结合空洞卷积与注意力机制的肺结节分割方法（DC-CBAM-UNet++）。

该方法在传统UNet++网络基础上引入空洞卷积（DC-UNet++），并增加注意力机制加强特征图获得更多加权占比，使特征图获得更大的感受野。

在LIDC肺结节公开数据集上的训练与验证结果表明，所提模型精确率、相似系数和交并比分别达到94.98%、90.86%、84.54%，证明了该方法的有效性，为分割肺结节图像提供了一种新方法。

关键词：UNet++；空洞卷积；注意力机制；图像分割DOI：10.11907/rjdk.221582开放科学（资源服务）标识码（OSID）：中图分类号：TP391 文献标识码：A文章编号：1672-7800（2023）007-0125-06Lung Nodule Image Segmentation Method Based on DC-CBAM-UNet++ NetworkXU Wei， TANG Junwei， ZHANG Chi（1.Engineering Research Center of Hubei Province for Clothing Information；2.School of Computer Science and Artificial Intelligence ， Wuhan Textile University ，Wuhan 430200， China）Abstract：To address the problems of small volume， irregular shape and blurred edges in lung nodule images， which lead to difficulty in fea‐ture extraction and low segmentation accuracy， propose a lung nodule segmentation method （DC-CBAM-UNet++） based on UNet++ com‐bined with cavity convolution and attention mechanism. In order to obtain a larger sense field for the feature map， this method improves the tra‐ditional UNet++ network ， introducing the null convolution （DC-UNet++） on the original basis， and also introducing the attention mechanism to enhance the feature map to obtain more weighted occupancy. Experiments were conducted using the LIDC lung nodule public dataset for training and validation， and the accuracy， similarity coefficient and cross-merge ratio reached 94.98%，90.86% and 84.54%， respectively，demonstrating the effectiveness of the method and providing a new method for segmenting pulmonary nodule images.Key Words：UNet++； dilated convolution； attention mechanism； image segmentation0 引言肺癌是一种严重疾病，死亡率较高，但病因尚不明晰，可能由于长期吸烟和所处的环境引起。

基于SWAT-MODFLOW地表-地下水耦合模型的结构与应用研究

基于SWAT-MODFLOW 地表−地下水耦合模型的结构与应用研究刘文冲 1，赵良杰 2, 3，崔亚莉 1，曹建文 2, 3，王　莹 4，李美玲1（1. 中国地质大学(北京), 北京 100083；2. 中国地质科学院岩溶地质研究所/自然资源部、广西岩溶动力学重点实验室/联合国教科文组织国际岩溶研究中心, 广西桂林 541004；3. 广西平果喀斯特生态系统国家野外科学观测研究站, 广西平果 531406；4. 广东省地质环境监测总站, 广东广州 510510）摘　要：为了利用Seonggyu Park 和Ryan T.Bailey 的SWAT-MODFLOW 耦合程序实现地表、地下不同范围模型耦合，同时探究耦合程序输出的以SWAT 计算的地下水补给量和以MODFLOW 网格计算的补给量之间的差异，以及耦合程序在有关地表地下水研究上的优势。

本文以该耦合程序示例模型美国佐治亚州南部小河流域（LRW ）为例，选取模型中SWAT 划分的104号子流域为边界，用GMS10.4建立地下水流模型，最后将地下水流模型和原SWAT 模型进行耦合。

研究结果表明：(1)耦合程序能实现以地表分水岭自然边界为范围的SWAT 模型与以子流域为边界的小范围MODFLOW 模型的耦合，但由于地下水流模型网格边界和子流域边界不能完全匹配，导致MODFLOW 以网格计算的地下水降雨补给量和SWAT 统计的地下水降雨补给量存在差异，误差随网格变小而变小；(2)耦合后各均衡项发生了变化，河道对地下水的总补给量变为耦合前的15.25%，地下水向河道的总排泄量比耦合前多19.29%，总降雨补给比耦合前多17.07%，总蒸发量是耦合前的3.08倍。

经过研究发现耦合模型能更准确的模拟地表地下水文过程，反映降水与地下水、地表水与地下水转化关系。

关键词：SWAT-MODFLOW ；耦合模型；地表−地下水中图分类号：P333；P641 文献标识码：A 文章编号: 1001 − 4810 （ 2023 ） 06 − 1131 − 09开放科学（资源服务）标识码（ OSID ）：0 引　言数值模型是定量分析水文过程的重要工具，其中具代表性的模拟软件有SWAT 和MODFLOW 。

基于SWAT_模型的渠江流域径流侵蚀功率时空规律分析

第22卷　第2期2024年4月中国水土保持科学Science of Soil and Water ConservationVol.22　No.2Apr.2024　D OI :10.16843/j.sswc.2023084基于SWAT 模型的渠江流域径流侵蚀功率时空规律分析黄　幸1,2,莫淑红2†,李平治2,乔殿新3,李斌斌3(1.河海大学水文水资源学院,210098,南京;2.西安理工大学省部共建西北旱区生态水利国家重点实验室,710048,西安;3.水利部水土保持监测中心,100053,北京)摘要:探明流域径流侵蚀功率的演变规律对重点侵蚀区识别和土壤侵蚀防治至关重要㊂以嘉陵江右岸支流渠江流域为研究区,基于SWAT 模型模拟计算流域径流侵蚀功率,分析其时空分布特征与空间尺度效应,并通过聚类分析㊁相关性分析揭示其对流域气象㊁地形㊁土壤等因素的关系㊂结果表明:渠江流域年尺度的径流侵蚀功率大于季尺度,其中第3季度为土壤侵蚀重点防治时段;全年和第3季度的多年平均径流侵蚀功率均呈现出北部大南部小㊁西部大东部小㊁上游大下游小的空间分布特征;渠江干流与其支流大通江的多年平均径流侵蚀功率和流域控制面积之间均呈幂指数关系,且其变化规律存在空间阈值,在年尺度干流和大通江的阈值面积分别为8549.4和8504.4km 2,在第3季度干流和大通江的阈值面积分别为4834.9和6223.5km 2;气象因子㊁地形因子和流域形态因子为渠江流域径流侵蚀功率的主要影响因素㊂研究结果可为制订渠江流域土壤侵蚀治理方案提供决策依据㊂关键词:SWAT 模型;径流侵蚀功率;时空分布;空间尺度效应;渠江流域中图分类号:S157.1;P333文献标志码:A文章编号:2096⁃2673(2024)02⁃0025⁃09引用格式:黄幸,莫淑红,李平治,等.基于SWAT 模型的渠江流域径流侵蚀功率时空规律分析[J].中国水土保持科学,2024,22(2):25-33.HUANG Xing,MO Shuhong,LI Pingzhi,et al.Spatial and temporal analysis of runoff ero⁃sion power in Qujiang River Basin based on SWAT model[J].Science of Soil and Water Conservation,2024,22(2):25-33.收稿日期:20230508　修回日期:20240110项目名称:国家自然科学基金基于侵蚀能量过程的集合式流域水土流失预报模型”(U2040208), 陕北黄土高原区人水耦合系统互馈及协同进化机理研究”(52179024)第一作者简介:黄幸(2000 ),女,硕士研究生㊂主要研究方向:水文水资源㊂E⁃mail:809319042@†通信作者简介:莫淑红(1972 ),女,博士,教授㊂主要研究方向:旱区水文及水资源㊂E⁃mail:moshuhong@Spatial and temporal analysis of runoff erosion powerin Qujiang River Basin based on SWAT modelHUANG Xing 1,2,MO Shuhong 2,LI Pingzhi 2,QIAO Dianxin 3,LI Binbin 3(1.College of Hydrology and Water Resources,Hohai University,210098,Nanjing,China;2.State Key Laboratory of Eco⁃hydraulics in Northwest Arid Region,Xi′an University of Technology,710048,Xi′an,China;3.The Center of Soil and Water Conservation Monitoring,Ministry of Water Resources,100053,Beijing,China)Abstract :[Background ]Soil erosion destroys soil and water resources,exacerbates natural disasters such as droughts and floods,and threatens human survival and development.Qujiang River Basin is severely affected by soil erosion,and runoff erosion power can reflect dynamic characteristics of water erosion better than rainfall erosion force.Therefore,it’s important to use runoff erosion power theory to study erosion in Qujiang River Basin to reveal mechanism of water⁃sand response.[Methods ]This paper中国水土保持科学2024年took Qujiang River Basin as a research object,calculating seasonal as well as annual runoff erosion power based on runoff,which was simulated by SWAT model in terms of utilizing meteorological forcing data such as precipitation,temperature,wind etc.In addition,features of spatial⁃temporal pattern and effects of spatial scale were analyzed.Cluster and correlation method were adopted for investigation into relationships between runoff erosion power with meteorological,topographic and soil conditions. [Results]1)The constructed SWAT model had high accuracy in runoff simulation and good applicability in Qujiang River Basin.R2and NS coefficients were0.69and above,while PBIAS coefficient was below 16.72%in both parameter rate setting period and validation period.2)In aspect of time,annual runoff erosion power outweighed that in season.However,erosion during seasonⅢwas more serious than that in other season,which requires more attention on soil erosion prevention and control.Besides,runoff erosion power for whole year and for season III demonstrated a decreasing trend from north to south,west to east and up to down inspace.3)The thresholds of drainage control area for whole year in Qujiang river and Datong river were8549.4and8504.4km2respectively,while those for season III were4834.9and 6223.5km2respectively,indicating runoff erosion power decreased smoothly with increasing area when the area was larger than spatial thresholds,then gradually tended to a stable value.4)Meteorological, topographic and watershed morphological characters were main factors influenced runoff erosion power in Qujiang River Basin.Erosion in upstream area of basin presented greater performance than downstream, due to the steep topography,uneven precipitation distribution and morphological ease of runoff generation and flow concentration processes in upper reaches.[Conclusions]This paper illustrated feasibility of SWAT model and its simulated outcome in Qujiang River Basin.The spatial⁃temporal runoff erosion power characteristics together with impacts are closely related to meteorological constituents,terrain and basin shape.Therefore,the results contribute to effective identification of key sand producing areas in watershed,and also provide supports for soil erosion prevention,ecology restoration and environmental carrying capacity enhancement.Keywords:SWAT model;runoff erosion power;temporal and spatial distribution;spatial scale effect; Qujiang River Basin 土壤侵蚀是全球公认的最严重的环境问题之一,其破坏水土资源,加剧旱涝等自然灾害,威胁人类的生存和发展㊂嘉陵江流域地质地貌条件复杂㊁降水丰富㊁土壤可蚀性强,且不合理生产活动多,使得流域内水土流失面积和土壤侵蚀总量曾常居长江流域前列,水土流失治理刻不容缓[1]㊂当前有许多量化土壤侵蚀对流域影响的模型,如通用土壤流失方程㊁修正土壤流失方程和中国坡面土壤流失方程等[2]㊂降雨侵蚀力反映降雨及其产生的径流剥离和携带土壤颗粒的能力,其作为水蚀动力指标被广泛应用于土壤侵蚀模型及土壤侵蚀分析中[34]㊂但是降雨侵蚀力的计算通常需要高精度且不易获取的长序列场次降雨数据,资料处理较繁琐,且其仅通过雨滴的击溅效应表征土壤侵蚀作用,并未体现水蚀动力过程中的径流侵蚀和径流输沙作用[56],具有一定的局限性㊂与降雨侵蚀力相比,径流侵蚀功率能更好的反映水蚀动力特性,对于侵蚀动力机制的反应更敏感,数据要求更低[7]㊂目前已有众多学者基于径流侵蚀功率理论研究黄土高原地区的土壤侵蚀空间分布特征及尺度效应,取得较好成果[89]㊂长江流域的降水特征及下垫面条件与黄河流域不同,其侵蚀产沙和主要驱动机制也会有所差异[10],而此类研究在长江流域相对较少㊂因此,利用径流侵蚀功率理论研究长江流域地区的侵蚀情况对揭示其水沙响应机理具有重要意义㊂笔者以嘉陵江子流域渠江流域为例,基于SWAT(Soil and Water Assessment Tool)模型模拟流域径流,计算并分析径流侵蚀功率的时空分布特征㊁空间尺度效应及主要影响因子,以期为流域有效识别重点产沙区㊁合理开发利用水土资源等做出贡献㊂1　研究区概况渠江流域面积为3.92万km2(E106°28′~ 109°00′,N30°00′~32°48′),属亚热带湿润季风气62　第2期黄幸等:基于SWAT模型的渠江流域径流侵蚀功率时空规律分析候㊂其雨季集中于7 9月,多年平均降雨量为1078mm,输沙量主要来自汛期,多年平均输沙模数为347t/km2㊂流域地势东北高西南低,源头地势坡度较大,至下游浅丘区比降逐渐减小,土壤类型以棕壤㊁黄棕壤和紫色土为主㊂渠江流域水系及水文站点分布见图1㊂图1　渠江流域水系及水文站点分布图Fig.1　Distribution of river system and hydrological stations in Qujiang River Basin2　数据与方法2.1　数据来源笔者采用的数据主要包括摘自‘长江流域水文年鉴“的2008 2018年渠江流域上㊁中㊁下游巴中㊁七里沱和罗渡溪水文站实测日流量资料,由中国大气同化驱动数据集整理的逐日降水㊁风速㊁温度等气象数据,以及中科院地理空间数据云㊁中科院数据中心遥感影像㊁世界土壤数据库提供的30m分辨率地形数据㊁1∶10万土地覆盖数据和1∶100万土壤类型等空间数据㊂2.2　径流侵蚀功率计算2.2.1　SWAT模型构建与应用　笔者使用SWAT 模型[11]模拟渠江流域2009 2018年月径流,进而分析计算流域径流侵蚀功率㊂先依据集水面积㊁坡度等级㊁土地利用与土壤类型将渠江流域划分为77个子流域和1377个水文响应单元㊂然后,选取2008年作为预热期进行参数预调,2009 2013年作为率定期,2014 2018年作为验证期㊂在SWAT-CUP中选择径流相关参数进行参数敏感性分析和参数率定,并运用决定系数R2(R-Square)㊁纳什效率系数NS(Nash-Sutcliffe efficiency coefficient)㊁偏差比例PBIAS(percent bias)作为评价指标检验径流模拟精度㊂一般认为R2>0.6㊁NS>0.5㊁PBIAS≤±25%时,结果较好㊂满足精度要求时,输出渠江各子流域出口的模拟月径流过程㊂2.2.2　径流侵蚀功率计算　径流侵蚀功率属于经验模型指标中的侵蚀动力因子,其计算原理及应用详见文献[7]㊂依此推出的季径流侵蚀功率和年径流侵蚀功率计算公式分别见式(1)和式(2)㊂E s=Q s H s㊂(1)式中:E s为季径流侵蚀功率,m4/(km2㊃s);Q s为季最大月流量模数,m3/(km2㊃s);H s为季径流深,m㊂E y=Q y H y㊂(2)式中:E y为年径流侵蚀功率,m4/(km2㊃s);Q y为年最大月流量模数,m3/(km2㊃s);H y为年径流深,m㊂2.3　径流侵蚀功率影响因素分析选用系统聚类方法[12]和斯皮尔曼相关系数[13]对径流侵蚀功率影响因子进行合并归类与相关性分析㊂斯皮尔曼相关系数的绝对值越大,指标之间相关性越高,其表达式见式(3)㊂ρ=1-6∑n i=1d2in(n2-1)㊂(3)式中:ρ为斯皮尔曼相关系数;n为系列长度;d为72中国水土保持科学2024年按升序或降序排列后,同次序指标间的差值,量纲均为1㊂3　结果与分析3.1　模型率定与验证结果率定期㊁验证期巴中㊁七里沱和罗渡溪水文站模拟与实测月径流量值对比见图2~4㊂分析可知,巴中水文站率定期R 2=0.96㊁NS =0.92㊁PBIAS =16.2%,验证期R 2=0.70㊁NS =0.69㊁PBIAS =16.43%,率定结果较好,验证结果评级为可信;七里沱水文站率定期R 2=0.95㊁NS =0.93㊁PBIAS =-0.3%,验证期R 2=0.82㊁NS =0.80㊁PBIAS =-14.81%,率定结果极好,验证结果评级为较好;罗渡溪水文站率定期R 2=0.93㊁NS =0.92㊁PBIAS =-3.5%,验证期R 2=0.87㊁NS =0.84㊁PBIAS =-16.72%,率定结果极好,验证结果评级为较好㊂综上所述,在率定期和验证期内,3个水文站模拟所得月径流过程与实测月径流过程吻合度较高,表明所构建的SWAT 模型能较真实地反映渠江流域水文情势㊂图2　巴中水文站月径流量模拟值与实测值对比图Fig.2　Comparison of simulated and measured monthly runoff at Bazhong station图3　七里沱水文站月径流量模拟值与实测值对比图Fig.3　Comparison of simulated and measured monthly runoff at Qilituo station3.2　径流侵蚀功率的时空分布基于SWAT 模型得到2009 2018年渠江77个子流域季尺度和年尺度的多年平均径流侵蚀功率,其空间分布见图5和6㊂从时间角度看,全流域多年平均年径流侵蚀功率为0.049m 4/(km 2㊃s),总体上大于各季度多年平均径流侵蚀功率;在季尺度,第3季度多年平均径流侵蚀功率最大(0.028m 4/(km 2㊃s)),其次为第2季度㊁第4季度和第1季度,其变化周期与水文周期一致㊂显然,第3季度的径流侵蚀功率对全年贡献最大,应当重视该季度的水土流失治理㊂在空间尺度上,流域全年和第3季度的多年平均径流侵蚀功率均呈现出北部大㊁南部小;西部大㊁东部小;上游大㊁下游小”的空间分布特征㊂第4季度的空间分布虽在南北和上下游部分与全年相似,但在东西部表现出东部大㊁西部小”的特征㊂第1季度和第2季度的空间分布82　第2期黄幸等:基于SWAT模型的渠江流域径流侵蚀功率时空规律分析图4　罗渡溪水文站月径流量模拟值与实测值对比图Fig.4　Comparison of simulated and measured monthly runoff at Luoduxi station与全年相反,表现为南部大㊁北部小;东部大㊁西部小;下游大㊁上游小”,但因其径流侵蚀功率较小,对年尺度空间分布影响也较小㊂3.3　径流侵蚀功率的空间尺度效应渠江流域的径流侵蚀功率空间差异明显,可能存在空间尺度效应,即径流侵蚀功率随流域控制面积变化可能呈现出一定规律㊂首先利用SWAT 模型计算2009 2018年渠江77个子流域出口断面及以上控制面积的全年和第3季度的多年平均径流侵蚀功率,结果见图7㊂由图7可知,对全年和第3季度,渠江各子流域出口断面及以上控制区域的多年平均径流侵蚀功率总体呈现出上游大,下游小;干流大,支流小”的分布特点㊂为量化径流侵蚀功率与流域控制面积的相关关系,选取侵蚀较严重的渠江干流和支流大通江为研究对象,将流域控制面积分别与全年和第3季度的多年平均径流侵蚀功率进行拟合分析㊂干流的拟合结果分别见式(4)㊁式(5)和图8,大通江的拟合结果分别见式(6)㊁式(7)和图9㊂E my =0.0761A -0.158,R 2=0.67;(4)E ms =0.0438A-0.137,R 2=0.52;(5)E my 1=0.0771A -0.153,R 2=0.46;(6)E ms 1=0.0510A-0.165,R 2=0.49㊂(7)式中:E my ㊁E my 1分别为干流和大通江多年平均年径流侵蚀功率,m 4/(km 2㊃s);E ms ㊁E ms 1分别为干流和大通江多年平均季径流侵蚀功率,m 4/(km 2㊃s);A 为流域控制面积,103km 2㊂由图8和9可知,渠江干流和支流大通江存在较显著的空间尺度效应,其多年平均径流侵蚀功率与流域控制面积之间均呈幂指数关系,且干流相关关系优于支流大通江,全年相关系数与第3季度接近㊂随着流域控制面积的增大,多年平均径流侵蚀功率逐渐减小并趋于平缓,减小速率由大变小,说明存在空间阈值㊂为确定阈值,对式(4)~(7)求导,导数方程见式(8)~(11)㊂E′my =-0.0120A -1.158;(8)E′ms =-0.0060A-1.137;(9)E′my 1=-0.0118A -1.153;(10)E′ms 1=-0.0084A-1.165㊂(11)经计算可知,渠江干流在全年和第3季度的空间阈值分别为8549.4和4834.9km 2,大通江在全年和第3季度的空间阈值分别为8504.4和6223.5km 2㊂即当流域控制面积小于空间阈值时,|E |≥0.001,多年平均径流侵蚀功率随流域控制面积增加而迅速减小,反之则变化平缓,并逐渐减小至某一稳定值㊂其中干流的全年和第3季度径流侵蚀功率稳定值分别为0.04和0.02m 4/(km 2㊃s),大通江的全年和第3季度径流侵蚀功率稳定值分别为0.056和0.035m 4/(km 2㊃s)㊂3.4　径流侵蚀功率影响因素为探究渠江流域径流侵蚀功率时空分布及空间尺度效应的主要驱动因素,选用SPSS (statistical product and service solutions)软件对多年平均的年径流侵蚀功率进行聚类分析㊂本文共选取地形(75°以上坡度面积比例)㊁流域形态(圆度)㊁气象(各子流域上的多年平均降水量)㊁侵蚀动力(多年平均径流侵蚀功率)㊁土地利用(林草地面积比例)和土壤(易受侵蚀土壤面积比例)6类聚类因子,经分析将渠江77个子流域划分为3种聚类类型,其空间分布见图10㊂可知,第1聚类与第2聚类主要分布于上游地区,其面积所例分别为26.0%和19.3%,多年平均径流侵蚀功率分别为0.05892中国水土保持科学2024年　4个季度的季径流侵蚀功率量级不同,值越大侵蚀越严重,其总体范围为0.00001≤E s≤0.08000㊂下同㊂Magnitude of seasonal runoff erosion power varies among four periods,with larger values indicating more severe erosion.The over⁃all range is0.00001≤E s≤0.08000.The same below.图5　渠江流域多年平均季径流侵蚀功率空间分布图Fig.5　Spatial distribution of multi⁃year average runoff erosion power at season scale in Qujiang River Basin 和0.052m4/(km2㊃s),均大于多年平均径流侵蚀功率中值0.049m4/(km2㊃s),该地区径流侵蚀能量较大,受侵蚀程度较剧烈㊂第3聚类是渠江流域的主导聚类,主要分布在中下游地区,其面积比例为54.7%,流域内多年平均径流侵蚀功率为0.046m4/(km2㊃s),小于多年平均径流侵蚀功率中值,该地区植被类型多样,下垫面条件较好,受侵蚀程度较小㊂03　第2期黄幸等:基于SWAT模型的渠江流域径流侵蚀功率时空规律分析图6　渠江流域多年平均年径流侵蚀功率空间分布图Fig.6　Spatial distribution of multi⁃year average runoff erosionpower at year scale in Qujiang River Basin 统计侵蚀动力因子与其余聚类因子的斯皮尔曼相关系数见表1,结果表明气象因子㊁地形因子和流域形态因子为渠江流域径流侵蚀功率排名前3的主图7　渠江各子流域出口断面及以上控制面积的多年平均径流侵蚀功率空间分布图Fig.7　Spatial distribution of multi⁃year average runoff erosion power for control area at and abovethe outlet cross⁃section of each Qujiang River subbasin要影响因子㊂且在显著水平为0.05时,第1聚类中的气象因子与径流侵蚀功率呈显著相关,说明降雨的空间分布不均性对该地区径流侵蚀能量影响较大;第2聚类中气象因子㊁地形因子和流域形态因子与径流侵蚀功率呈显著相关,说明该地区75°以上坡度面积比例很大,流域内地势陡峭,降雨空间分布差异较显著,流域形态偏圆形,产汇流过程较快,能显著影响径流侵蚀功率的空间分布和规律;第3聚类中流域形态因子与气象因子与径流侵蚀功率呈一般相关,其对该聚类地区径流侵蚀功率的影响远不如第2聚类地区,说明该地区整体地势较平缓,各子流域多年平均降雨量差距不大,流域形态较狭长,产汇流过程相对较慢㊂4　结论1)所构建的SWAT 模型径流模拟精度较高,在渠江流域适用性好㊂巴中㊁七里沱㊁罗渡溪水文站率定期R 2㊁NS 系数均在0.92及以上,PBIAS 系数均<16.20%;验证期R 2㊁NS 系数均在0.69及以上,PBIAS 系数均在16.72%以下㊂这说明该模型能较真实地反映渠江流域水文情势㊂2)渠江流域多年平均年径流侵蚀功率总体上大于季径流侵蚀功率,其中第3季度>第2季度>第4季度>第1季度㊂流域全年和第3季度的多年平均径流侵蚀功率空间分布均呈现出北部大㊁南部小;西部大㊁东部小;上游大㊁下游小”的特征;第4季度的空间分布在东西部呈现出东部大㊁西部小”的特征,其余部分与全年相似;第1㊁第2季度的空间分布与全年分布规律相反㊂ 3)渠江干流及其支流大通江的径流侵蚀功率具有较为显著的空间尺度效应,全年和第3季度的多年平均径流侵蚀功率与流域控制面积之间呈13中国水土保持科学2024年图8　渠江干流多年平均年和第3季度径流侵蚀功率与流域控制面积关系图Fig.8　Fitting results of multi⁃year average runoff erosionpower and watershed control area at year and seasonscale(season Ⅲ)in the main stream of Qujiang River图9　渠江支流大通江多年平均年和第3季度径流侵蚀功率与流域控制面积关系图Fig.9　Fitting results of multi⁃year average runoff erosionpower and watershed control area at year and seasonscale(season Ⅲ)in Datong River图10　渠江流域聚类空间分布图Fig.10　Clustering spatial distribution of Qujiang River Basin幂指数均关系㊂当流域控制面积分别>8549.4和4834.9km 2时,干流全年和第3季度的多年平均径流侵蚀功率随着面积增加变化幅度极小,并逐渐稳定于0.04和0.02m 4/(km 2㊃s);当流域控制面积分别大于8504.4和6223.5km 2时,大通江全年和第3季度的多年平均径流侵蚀功率随着面积增加变化幅度极小,并逐渐稳定于0.056和0.035m 4/(km 2㊃s)㊂表1　斯皮尔曼相关系数计算结果表Tab.1　Spearman correlation coefficient calculated results table聚类名Cluster name地形因子Topographical factor流域形态因子Watershed morphology气象因子Meteorology 土地利用因子Land use 土壤因子Soil第1聚类Cluster Ⅰ-0.451-0.4840.621*-0.280-0.022第2聚类Cluster Ⅱ0.636*0.622*0.678*0.441-0.336第3聚类Cluster Ⅲ0.0680.349*0.336*-0.2610.072　注:*表示在0.05水平上相关性显著㊂Notes:*indicates significant correlation at the 0.05level. 4)气象因子㊁地形因子和流域形态因子对渠江流域径流侵蚀功率的分布有主要影响作用㊂流域上游区因地势陡峭㊁降水分布不均㊁形态易于产汇流而表现出较大的径流侵蚀功率;下游地区因地势平坦,产汇流过程缓慢受侵蚀情况较轻㊂5　参考文献[1]　万彩兵,程冬兵,李昊.水土保持法修订实施十年来长江流域水土流失治理成效[J].中国水土保持,2021(6):1.WAN Caibing,CHENG Dongbing,LI Hao.Effect of soil and water loss control in the Yangtze River Basin since the revision and implementation of soil and water conser⁃vation law for ten years [J].Soil and Water Conservation in China,2021(6):1.[2]　史志华,刘前进,张含玉,等.近十年土壤侵蚀与水土保持研究进展与展望[J ].土壤学报,2020,57(5):1117.SHI Zhihua,LIU Qianjin,ZHANG Hanyu,et al.Study on soil erosion and conservation in the past 10years:Pro⁃23　第2期黄幸等:基于SWAT模型的渠江流域径流侵蚀功率时空规律分析gress and prospects[J].Acta Pedologica Sinica,2020,57(5):1117.[3]　许功伟,徐立荣,张云苹,等.1970 2019年大汶河流域降雨侵蚀力时空变化分析[J].水土保持研究,2022,29(6):45.XU Gongwei,XU Lirong,ZHANG Yunping,et al.Spati⁃otemporal variation of rainfall erosivity in the Dawen RiverBasin during1970-2019[J].Research of Soil and Wa⁃ter Conservation,2022,29(6):45.[4]　WISCHMEIER W H,SMITH D D.Predicting rainfallerosion losses:A guide to conservation planning[M].Washington:Department of Agriculture,1978:26.[5]　徐晶,时延锋,徐征和,等.1961 2020年沂蒙山区降雨侵蚀力时空变化特征[J].水土保持研究,2022,29(6):36.XU Jing,SHI Yanfeng,XU Zhenghe,et al.Spatiotem⁃poral variation characteristics of rainfall erosivity in theYimeng Mountain area from1961to2020[J].Researchof Soil and Water Conservation,2022,29(6):36. [6]　TU Anguo,ZENG Jianling,LIU Zhao,et al.Effect ofminimum inter⁃event time for rainfall event separationon rainfall properties and rainfall erosivity in a humidarea of southern China[J].Geoderma,2023,431:116332.[7]　鲁克新.黄土高原流域生态环境修复中的水沙响应模拟研究[D].西安:西安理工大学,2006:18.LU Kexin.Study on runoff⁃sediment response simulationin watershed eco⁃environmental rehabilitation on loessplateau[D].Xi′an:Xi′an University of Technology,2006:18.[8]　张译心,徐国策,李占斌,等.不同时间尺度下流域径流侵蚀功率输沙模型模拟精度[J].水土保持研究,2020,27(3):1.ZHANG Yixin,XU Guoce,LI Zhanbin,et al.Simula⁃tion accuracy of runoff erosion power⁃sediment transfermodel at different time scales[J].Research of Soil andWater Conservation,2020,27(3):1.[9]　杨光,鲁克新,李鹏,等.基于SWAT模型的清水河流域年径流侵蚀功率空间分布[J].水土保持研究,2022,29(6):90.YANG Guang,LU Kexin,LI Peng,et al.Spatial distri⁃bution of annual runoff erosion power in Qingshui RiverBasin based on SWAT model[J].Research of Soil andWater Conservation,2022,29(6):90. [10]CHANG Enhao,LI Peng,LI Zhanbin,et al.The impactof vegetation successional status on slope runoff erosion inthe Loess Plateau of China[J].Water,2019,11(12):2614.[11]HUSSAINZADA W,LEE H S.Effect of an improved ag⁃ricultural irrigation scheme with a hydraulic structure forcrop cultivation in arid northern Afghanistan using the Soiland Water Assessment Tool(SWAT)[J].Scientific Re⁃ports,2022,12(1):5186.[12]倪鹏,包为民,张乾,等.基于主成分分析的系统聚类分析方法在洪水预报中的应用[J].三峡大学学报(自然科学版),2018,40(1):1.NI Peng,BAO Weimin,ZHANG Qian,et al.Applica⁃tion of hierarchical cluster analysis to flood forecastingbased on principal component analysis[J].Journal ofChina Three Gorges University(Natural Sciences),2018,40(1):1.[13]张钰荃,高成,陈旭东,等.西江上游流域年径流系数变化规律及其影响因素分析[J].水电能源科学,2021,39(5):37.ZHANG Yuquan,GAO Cheng,CHEN Xudong,et al.Variation of runoff coefficient and its influencing factors inupper reaches of Xijiang River Basin[J].Water Re⁃sources and Power,2021,39(5):37.33。

THE INTERNATIONAL JOURNAL OF MEDICAL ROBOTICS AND COMPUTER ASSISTED SURGERY Int J Med Robot

Introduction
Computer-assisted surgery (CAS) is a methodology that translates into accurate and reliable image-to-surgical space guidance. Neurosurgery is a very complex procedure and the surgeon has to integrate multi-modal data to produce an optimal surgical plan. Often the lesion of interest is surrounded by vital structures, such as the motor cortex, temporal cortex, vision and audio sensors, etc., and has irregular conﬁgurations. Slight damage to such eloquent brain structures can severely impair the patient (1,2). CASMIL, an imageguided neurosurgery toolkit, is being developed to produce optimum plans resulting in minimally invasive surgeries. This system has many innovative features needed by neurosurgeons that are not available in other academic and commercial systems. CASMIL is an integration of various vital modules, such as rigid and non-rigid co-registration (image–image, image–atlas and

1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。

AUTOMATICATLAS-BASEDBUILDINGOFPOINTDISTRIBUTIONMODELFORSEGMENTATIONOFANATOMICALSTRUCTURESFROMBRAINMRI.

JonathanBailleul,SuRuan,DanielBloyetGREYCCNRSUMR6072,Ensicaen&UniversitedeCaen,F-14050Caencedex,France

ABSTRACTWeproposeamethodbasedonaprioriknowledgepro-videdbyanatomicalatlasestobuild-almost-automati-callyaPointDistributionModel(PDM)ofinternalbrainstructures.A3DtrainingsetofshapesisconstructedbyregisteringchosenatlasoveranMRIdatabase,whichisthenland-markedusingamethodrecentlydevelopedbyDaviesetal.ThisPDMglobaloptimizationprocessisdrivenbyaMin-imumDescriptionLength(MDL)principle-basedobjectivefunction.PreliminaryresultsofbuiltPDMsareencouraging.FutureworkmightbuildActiveAppearanceModels(AAM)fromcomputedPDMstosetupasegmentationmethod.1.INTRODUCTION1.1.Atlas-basedMRIsegmentationAsaﬁrstattempttosegmentanatomicalstructuresfrombrainMRIvolumes,wetriedtoimprovethemethod[4]de-velopedinourlabbyJ.H.Xueandrelyingonaprioriin-formationprovidedbyananatomicalatlasAtlasa.Suchanatlasresultsfromsegmentationofstructuresofinterest1inareferenceMRIvolumeRefa.Thisisatime-consuminganderror-pronetaskwhichdemandstheinterventionofanatom-icalexperts,henceourlimitednumberofsuchatlases(Cyc-eron,HarvardSPL,Talairach).Inthatmethod,Atlasaisﬁrstregistered2intoagivenpa-tientMRItoinferfuzzyﬁeldsprovidingimplicitspatialin-formationabouttheexpectedpositionsofstructures.Con-currently,patientMRIisoversegmentedusingFuzzyMarkovRandomFieldsintoregionslabeledamong20fuzzyclassesresultingfrommixturesofanatomicalbraintissues(CSF,GM,WM).Finally,asegmentationresultisinferredslice-by-slicethroughconjunctionofbothsourcesusingaratherstochasticGAapproach([4])or,inourcase,asequenceofamendedmorphologicﬁlters-includinghole-ﬁllfromasafestructurecore-inanattempttoemphasizedeterminis-ticbehaviorandfurtherextensibility.1fornow:ventricles,putamens,caudatenuclei,thalami,hippocampi2RogerP.Woods’sAIR,http://bishopw.loni.ucla.edu/AIR5/index.html1.2.EmergingneedforashapemodelInbothcases,resultsshowgoodglobalpositioningandcov-erageraterelativelytoexpert-segmentation3.Butwedidnotcompletelyovercomelocalnoticeableartifactsonlow-contrastedstructureedges-ventriclesexcepted-thatbasi-callyjustiﬁedournewapproach.Thislimitationincitedustolookforacomplementaryapproachemphasizingshape-correctnessofsegmentationproposalsthroughenforcementofexplicitshapeconstraintsinferredfromthechosenAtlasa.Thoughotherapproacheshavebeenexamined,thePointDistributionModelfromCootes&Taylor([1],[2])seemstobestsuitourpurpose.Itrequiresaninputtrainingsetofnsinstancesofstudiedobject,wherethesamenlabeled

landmarkshavebeenplacedtokeyanatomicallocichar-acterizingtheobject’sshape(nincludesintermediatecon-tourpoints).PrincipalComponentAnalysis(PCA)isthenappliedtoresultingnsshapevectorsofdimensionnp=

2󰀃3×ntoextractmostsigniﬁcantlinearvariationmodesofobject’sshapealongthetrainingset.Theydeﬁneanellip-soidalnp-DAllowableShapeDomain(ASD)determiningthespaceofvariation-centeredtomeanshape-allowedtoanunseenshape,thusprovidingacriterionfordiscriminat-ingunlikelyshapes.AlthoughthisPDMmodelcanrepresentshapeofob-jectsofanydimension,itsapplications(notonlyconﬁnedtoBrainMRI)arefarmorefrequentin2Dthanin3Dcon-sideringthatbuildinganannotatedtrainingsetin3Draisesnewproblemsweattemptedtoovercome.

2.TOWARDSAUTOMATIC3DPDMBUILDING2.1.Automaticbuildingofa3DtrainingsetInBrainMRIdomain,proper2DPDMscanusuallybeachievedusingalimitednumberoftrainingshapes(e.g.󰀂20)demandingreasonableexperteffort.Butmodelingof3Dshapevariationsgenerallyrequiresahigherns,whichisdramaticconsideringbuildingoneinstancerequiresexpertsegmentationonabout50volumeslices.

3applyingcurrentmethodonRefb=aandcomparingtoAtlas

0-7803-7946-2/03/$17.00 ©2003 IEEE.ISSPA 2003Inourpreviousmethod,wenoticedthatthewarpreg-istrationofAtlasa,duetoitsglobal-consistentnatureen-ablinglocaladaptations,managedtopreservetheshapeofourstructureswhilestillallowingthemgoodﬁttingtotargetMRI.Duetoresidualerrors,theresultingstructurescouldnotbeconsideredasaﬁnalsegmentationresult,butcloseexaminationshowedthattheseerrorsremainedsurprisinglysatisfactoryonmostMRIs.Anencouragingfactisthatafterrigidrealignment,thesestructuresshowednoticeableshapevariationsthatshouldbelearnedbyaPDM.AlthoughexactnessofthesestructuresaccordingtooriginalMRIcanbediscussed,itshouldbecompensatedinPDMsensebytheaprioriunlimitednum-berofMRIs(󰀂4000)availablethroughCyceron,thoughvisualcontrolisadvisedtorejectaminority(󰀂20%)ofpoorlyregisteredstructures.Asaconsequence,wewillassumewecanusetheseshapeinstancesasatrainingsetforourPDMmodelinaBoot-strapingapproach,consideringthatthesegmentationresultsofourupcomingPDM-basedmethodmightreﬁnethequal-ityoftheinputset,untilaccordancetoexpertresultscon-vergestoasatisfactorydegree.2.2.AutomaticlandmarkingusingtheMDLprincipleForsimilarreasons,if2Dlandmarkingcansometimesbemanuallypracticableoverlimitedsets,thisgenerallydoesnotscaletothe3Dcasethroughsimplez−iteration,aswecouldexperiment.Despitespecialcaseswithshapesofreg-ularform,3Dshapevariationshouldbeconsideredglob-ally,andoftengoesbeyondsimpleobservationorintuition.Someautomaticmethodsweredevelopedtodetectnotice-ablepointsfrom2D-andsometimes3D-objects,buttheytendtospecializetostudiedobjectsclasses.Furthermore,nonecanstatethatthesemethodschoosepointsintendingtodesigna’good’PDM.Though,recentworkfromRhodriDavies[2]formu-latedtrainingsetlandmarkingasaglobaloptimizationpro-cess.Eachshapeismappedontoacorrespondingspherewhereagivennumberoflandmarksisﬁrstevenlydisposed.Then,theirpositionsareblindlyalteredbyreparameteriza-tionfunctions.Theirevolutionisregulatedbyanobjectivefunctionevaluatingthe’quality’ofthePDMinferredbyback-projectionofdisplacedlandmarksontoshapes.Thus,thewholeprocessconvergestothe’bestpossible’PDMforcurrenttrainingset.TheaforementionedobjectivefunctionreliesontheMin-imumDescriptionLengthprinciple,whichcanbesummedupasfollows.ConsideringPCAanalysisprojectsshapepoints-actuallydeviationsfrommeanshape-tothespacedeﬁnedbycomputedorthogonaleigenvectors,wecanas-sumethetrainingsetgetsencodedbyacenteredmultivari-ateGaussianmodel.Theidea,takenfromthecommunica-tionﬁeld,consistsinattemptingtopacktogetherbothmodelparametersandmodel-encodedvaluesasasinglemessageinthemostcompactform.Balancingmodelcomplexityandaccuracyregardingtooriginaldata,DLissupposedtobeminimalwhenthecurrenttrainingsetinterpretation-i.e.landmarking-ismostbothgeneralizableandcom-pact.ResultsshowbetterinferredPDMsthanviamanuallandmarking,whichoftenattemptstolocateknownpointsofanatomicalsigniﬁcance,thusintroducingapartofhumansubjectivityorinadaptationtocurrenttrainingset.