Synthetic aperture radar remote sensing image blocking reconstruction method of boundary gray level distribution correlation

Abstract

The invention discloses a block reconstruction method of a synthetic aperture radar remote sensing image related to boundary gray scale distribution, comprising the following steps: dividing a single-frame image g formed by fusing SAR remote sensing image sequences into several image blocks, and dividing the block into blocks in order to avoid boundary effects Reserve a certain redundant boundary for each image block; construct the most suitable feature-driven prior for each image block according to its own characteristics, and perform super-resolution reconstruction on each image block one by one; finally cut off each reconstruction Construct the redundant boundaries of image blocks, and then recombine them into a complete SAR remote sensing image. The present invention improves the pertinence of the priori model area and the efficiency of computer storage operation through SAR image segmentation and reorganization, and can effectively maintain the correlation of image gray scale distribution, and the form of prior information is determined by the statistical characteristics of the image, ensuring The SAR image reconstruction process is efficient, the reconstructed visual continuity is good, and the data is accurate.

Description

technical field [0001] The invention relates to a remote sensing image processing technology, in particular to a block reconstruction method of a synthetic aperture radar remote sensing image related to boundary gray scale distribution. Background technique [0002] Synthetic Aperture Radar (SAR) remote sensing image super-resolution reconstruction is a technology that breaks through hardware physical limitations and artificial information blockade, and obtains radar remote sensing images with high spatial resolution through signal processing methods. Among them, the maximum a posteriori method (maximum a posteriori, MAP) SAR image super-resolution, which introduces the prior information of the image and follows strict Bayesian probability reasoning, has become a commonly used reconstruction method. [0003] However, there are following problems in the prior art: [0004] ①The prior information model usually introduced is an overall description of the statistical characteri...

AI Technical Summary

Problems solved by technology

[0004] ①The prior information model usually introduced is an overall description of the statistical characteristics of SAR image pixels, and does not consider the statistical characteristics of different objects or regions in SAR images. At the same time, the observation scale of SAR images is large, and the amount of data is higher than that of ordinary images. , the formed super-large matrix will put pressure on the storage and operation of the computer;

[0005] ②If the image block processing is used to improve the regional pertinence of the prior model and reduce the pressure of computer storage and operation, it is difficult to eliminate the resulting boundary effect of low correlation and poor continuity of the gray distribution between the image blocks. This is mainly because Neighborhood pixels (only three or even two neighborhood directions) participating in boundary pixel estimation are missing;

[0006] ③The form of prior information of traditional SAR images can be summarized into two basic models, Gaussian and Laplacian, which one to choose as prior information depends entirely on empirical assumptions, even the example-based method should be used The external training set is determined, and the reliability is not high

Images