Based on Spectrum Cluster High Resolution Remote Sensing Image Multi-criterion Stochastic Tree Division

The high score pigtail rate satellite remote sensing image may provide the rich data for the urban quakeproof disaster reduction information system´s construction. But, because its slight information is specially rich, has brought certain difficulty for the related terrain feature object detection and the extraction. The image division´s goal lies in the original image divides some in the space neighboring, the spectrum similar homogeneity region. Applies in view of the spectrum cluster in the remote sensing image division when the power moment spectrum calculates with difficulty actual problem, has defined the picture element and the kind of between distance, gives a sampling number theorem, has designed an image lamination division algorithm.