SAR (Synthetic Aperture Radar) image change detection method based on mixing model

Abstract

The invention discloses an SAR (Synthetic Aperture Radar) image change detection method based on a mixing model, which mainly solves the problems of single mode structure, poor matching degree and poorer speckle noise suppression of the traditional SAR image change detection technology. The SAR image change detection method comprises the following steps of: (1) implementing geometric registration and radiation correction for two images with different time phrases in a same area; (2) inputting images to be matched and corrected; (3) preprocessing the input image; (4) extracting difference images for the pre-processed images ; (5) constructing a mixing model of the difference images; (6) constructing an evaluation function and a penalty function in the mixing model; (7) solving the evaluation function to obtain an optimal threshold value; and (8) outputting a result image according to the optimal threshold value. According to the SAR image change detection method based on the mixing model, the mixing model is modeled according to the extracted different images, the problem that the pixel distribution of the SAR image cannot be perfectly matched in a single model is solved, the false change information is reduced, and a better detection result is obtained.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a hybrid model-based construction method in the technical field of SAR image change detection. The invention can be used for change detection of SAR images such as forest images, flood images, urban change images, etc., so as to realize real-time monitoring of disaster monitoring, land utilization, agricultural survey and other fields. Background technique [0002] As a representative of microwave remote sensing, synthetic aperture radar (SAR) has the advantages of all-weather, all-time, high resolution, and side-view imaging. In recent years, with the expansion of this technology from the military field to the civilian field, it has gradually become the focus of attention in the field of earth observation. SAR image processing is a new discipline developed in recent years, and it is also an important discipline closely related to national economic development and...

AI Technical Summary

Problems solved by technology

In this paper, the hypothetical model in the K&I algorithm is improved, and the traditional Gaussian model is replaced by a generalized Gaussian model, which better conforms to the pixel distribution in the difference image, and the accuracy evaluation effect is also improved compared with the traditional K&I algorithm, but For the suppression of coherence spots, the effect is still not obvious

Images