MorphologicalClassification¶
对输入图像通道执行形态上的凸面、凹面和平面分类
描述¶
该算法基于以下出版物：Martino Pesaresi和Jon Alti Benediktsson，IEEE成员：一种用于高分辨率卫星图像形态分割的新方法。《IEEE地球科学和遥感学报》，第39卷，第2期，2001年2月，第309-320页。
This application perform the following decision rule to classify a pixel between the three classes Convex, Concave and Flat. Let f denote the input image and \psi_{N}(f) the geodesic leveling of f with a structuring element of size N. One can derive the following decision rule to classify f into Convex (label \stackrel{\smile}{k}), Concave (label \stackrel{\frown}{k}) and Flat (label \bar{k}): f(n) = \begin{cases} \stackrel{\smile}{k} & : f-\psi_{N}(f)>\sigma \\ \stackrel{\frown}{k} & : \psi_{N}(f)-f>\sigma \\ \bar{k} & : \mid f - \psi_{N}(f) \mid \leq \sigma \end{cases}
输出为带标签的图像(0：平面，1：凸面，2：凹面)
参数¶
Input Image -in image Mandatory
The input image to be classified.
Output Image -out image [dtype] Mandatory
The output classified image with 3 different values (0: Flat, 1: Convex, 2: Concave)
Selected Channel -channel int Default value: 1
The selected channel index for input image
Structuring Element Type -structype [ball|cross] Default value: ball
Choice of the structuring element type
- Ball
- Cross
Radius -radius int Default value: 5
Radius of the structuring element (in pixels), default value is 5.
Sigma value for leveling tolerance -sigma float Default value: 0.5
Sigma value for leveling tolerance, default value is 0.5.
Available RAM (MB) -ram int Default value: 256
Available memory for processing (in MB).
实例¶
从命令行执行以下操作：
otbcli_MorphologicalClassification -in ROI_IKO_PAN_LesHalles.tif -channel 1 -structype ball -radius 5 -sigma 0.5 -out output.tif
来自Python的评论：
import otbApplication
app = otbApplication.Registry.CreateApplication("MorphologicalClassification")
app.SetParameterString("in", "ROI_IKO_PAN_LesHalles.tif")
app.SetParameterInt("channel", 1)
app.SetParameterString("structype","ball")
app.SetParameterInt("radius", 5)
app.SetParameterFloat("sigma", 0.5)
app.SetParameterString("out", "output.tif")
app.ExecuteAndWriteOutput()
局限性¶
形态分类的生成是不可流畅的，在设置结构元素的半径大小时要注意这一事实。