Optical remote sensing image change detection method based on double dictionary cross sparse representation

Abstract

The invention discloses an optical remote sensing image change detection method based on a double dictionary cross sparse representation. Problems that detection results of existing methods are not stable and leak detection and false warnings are not balanced. The detection method includes (1) reading-in two images with different time phase (2) establishing eigenvectors over each pixel in sequence (3) establishing a partial dictionary in sequence over the pixel construction (4) representing the eigenvector of the time phase two with partial dictionary sparsity of the time phase one for two time phase images at the same position, representing the eigenvector of the time phase one with partial dictionary sparsity of the time phase two (5) utilizing I 1 norms of all eigenvector to form two I1 norms images (6) forming a difference image (7) carrying out a maximum entropy threshold value over the difference image and obtaining an initial classification binary image (8) carrying out a regional growth over the initial classification binary image, then obtaining the final changing and detecting results. The detection results are capable of keeping the edge information at the changing areas, reducing the fake region of variation, effectively improving the detection accuracy and can be used fro monitoring the resources and estimating the disaster.

Description

technical field [0001] The invention belongs to the technical field of image processing and relates to change detection of optical remote sensing images, in particular to an optical remote sensing image change detection method based on double-dictionary cross-sparse representation, which is suitable for remote sensing image processing and analysis. Background technique [0002] Remote sensing change detection is to detect the change information between multiple remote sensing images acquired at different times in the same geographical location, and is widely used in many fields of national economy and national defense construction, such as land resources and land survey, forest resource monitoring, military reconnaissance, disaster forecasting and evaluation, supervision of major national ecological projects, etc. [0003] In the registered multi-temporal remote sensing image change detection method, the common method is to construct the difference map first, and then use th...

AI Technical Summary

Problems solved by technology

The disadvantage is that Treelets cross-filtering will lead to more missing detection information in the detection results, and cannot keep the edge information of the changing area well.

The disadvantage is that this method takes the l of the sparse coefficient 1 l of norm and approximation error 2 The product of the norms is used as the gray level magnitude of the difference map, while the l of the sparse approximation error 2 The norm is sensitive to the edge of the image, but insensitive to the smooth area. It takes a large value at the edge of the change area and a small value at the non-edge part of the change area, so the constructed difference image is unreasonable. Part of it will produce a lot of missed detection; in addition, false alarms will also be generated at the edge of the image in the non-changing area

Images