Multispectral Image Change Detection Method Based on Kernel Intermodal Factor Analysis Kernel Fusion

Abstract

The invention discloses a multi-spectral image change detection method based on nuclear intermodal factor analysis and nuclear fusion, which mainly solves the problem that the existing difference kernel function cannot fully tap and utilize the correlation between various features of the image, resulting in low detection accuracy . The realization process is as follows: 1) transform the two-time three-channel spectral image into HIS space, and obtain the hue, saturation, and intensity information of the two-time image; 2) extract the color feature and intensity texture feature of the two-time image according to the result of 1); 3) Use the proposed features to obtain the color difference kernel function matrix and the intensity texture difference kernel function matrix; 4) For the weighted fusion of the two matrices obtained in 3), construct a synthetic kernel function matrix, and adaptively select the optimal weighted matrix coefficient; 5) Input the synthetic kernel function matrix into the support vector machine SVM for detection, and obtain the change detection result. The invention has high detection precision, stable result and low calculation amount, and can be used for multispectral image change detection.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to an image change detection method, which can be used for monitoring and evaluating ground object state changes on multi-spectral images. Background technique [0002] Multispectral images are digital images of Earth observation in multiple bands acquired by remote sensing satellites. Multispectral image change detection determines the characteristics and process of ground object changes by analyzing multiple multispectral images observed at different times in the same area. Multispectral images have rich spectral information and can well reflect the characteristics of ground objects. Multispectral image change detection has a prominent role in military and civilian fields. [0003] The most commonly used multispectral image change detection method is the change vector analysis method CVA, which defines the multi-channel original two-time image as a vector, a...

AI Technical Summary

Problems solved by technology

However, when this method deals with multispectral images, it does not consider the correlation between various characteristics, and cannot fully mine and fuse various image features, so its detection accuracy needs to be further improved.

Images