Method for automatically segmenting multi-band image by adaptive filter

Abstract

The invention discloses a method for automatically segmenting a multi-band image by an adaptive filter. The method realizes automatic fine segmentation of the multi-band image by using a good adaptive mechanism, neighborhood fuzzy category determination and safe neighborhood threshold segmentation. The entire process mainly comprises the following steps of: (1) designing the filter; (2) segmenting the image (segmenting patches); (3) combining the minimum drafting area of patches (combining small patches); and (4) gradually clustering (classified variable and clustering algorithm). The method has the advantages of dynamic adjustment, adaptive selection of vector spatial direction, adaptive selection of weight and rapid convergence of an iterative process based on adaptive increase of a fuzzy region and capability of realizing anisotropic adaptive filtering and regional fuzzy increase.

Description

technical field [0001] The invention belongs to the technical field of computer remote sensing image information extraction and classification, and relates to a method for realizing automatic segmentation of multi-band images by an adaptive filter. Background technique [0002] Image segmentation divides the pixels in the image into different categories based on spectral features, and such a method is needed in the process of patch division. However, the feature features expressed by the gray value of each band of a pixel in a multi-band image are affected by factors such as soil background, vegetation coverage, plant composition, growing season, and utilization status, showing great complexity and ambiguity. There is a continuous transition feature in the edge zone between them. In addition to the interference information of the sensor, different ground object categories may have similar spectral characteristics, but there may be great differences between the same type of ...

AI Technical Summary

Problems solved by technology

[0004] These methods have difficulties and deficiencies in determining the category of pixels. First, some methods require complete samples of different object categories, which is almost impossible according to the principle of object spectrum. Second, the filter kernel used is static and automatic. The adaptation mechanism is poor; the third is that one-time judgment is often used in solving the attribution of pixel categories, which cannot fully adapt to the characteristics of neighborhood fuzzy sets; the fourth is that the number of supported ground object categories is limited, and the segmented image is not fine enough

Images