SAR image change detection method based on high-order neighborhood tmf model

Abstract

The invention discloses a SAR image change detection method based on a high-order neighborhood triple Markov random field model, which mainly solves the problems of high number of false detections and low overall precision in the existing method. 2. Initialize the label field X; 3. Initialize the likelihood parameters; 4. Define and initialize the auxiliary field U with a 3×3 neighborhood on the initialized label field X; 5. Use 5×5 The higher-order neighborhood of constructs a priori potential energy function composed of homogeneous area, heterogeneous area and U field; 6. Update the label field X and auxiliary field U; 7. Update the likelihood according to the updated label field X parameter; 8. Iteratively update the label field X and the auxiliary field U to obtain the final change detection result. Compared with the prior art, the invention reduces the number of false detections, improves the overall detection precision, and enhances the robustness to noise, and can be used for the recognition of SAR images.

Description

technical field [0001] The invention belongs to the technical field of image processing, relates to image change detection, and can be used for monitoring and evaluating ground object state changes on SAR images. Background technique [0002] Synthetic aperture radar SAR image change detection refers to detecting the change information of the ground objects in the area by analyzing two SAR images of the same area at different times. Due to the all-weather and all-time characteristics of synthetic aperture radar SAR, SAR image change detection technology has more and more applications in fields such as agricultural and forestry exploration, environmental monitoring, disaster assessment, and resource utilization, so it has high detection accuracy SAR image change detection method become the focus of research. [0003] One of the more classic methods of SAR image change detection is based on statistical models, such as the Markov random field MRF model and the triple Markov ra...

AI Technical Summary

Problems solved by technology

However, the study found that the auxiliary field U in this paper cannot accurately reflect the non-stationary characteristics of SAR images, and cannot distinguish homogeneous regions from non-homogeneous regions.

In addition, due to the existence of speckle noise in SAR images, when faced with SAR images with high noise intensity, the four-neighborhood system cannot suppress the influence of noise very well, and there are still high numbers of false detections in the obtained results.

Images