Multi-temporal SAR image multi-layer Bayesian blind deconvolution method for anti-impulse interference

Abstract

The invention discloses a multi-layer Bayesian blind deconvolution method for multi-temporal SAR images that is resistant to pulse interference. Multi-temporal SAR images of the same scene are input, registered and fused into a single-frame SAR image g; the Laplace distribution is used to The noise in g is modeled, and the original SAR image f, point spread function h and its prior model parameters αim and αh are modeled a priori in turn, and h, αim and αh are initialized; through the multi-layer Bayesian decomposition framework , to realize the modeling of joint probability distribution; based on the posterior distribution and multi-layer Bayesian decomposition framework, using integral and variational Bayesian reasoning, iteratively estimates the values ​​of f, h, αim and αh. The present invention uses multi-frame SAR images to increase the amount of information available for blind deconvolution; iteratively couples the estimation of a priori model parameter values ​​with the blind deconvolution of SAR images, thereby improving the estimation efficiency and accuracy of parameter values; using pull The noise model of Plath distribution is more suitable for suppressing pulse interference in SAR images and improves the accuracy of blind deconvolution.

Description

technical field [0001] The invention relates to an anti-pulse interference multi-temporal SAR image multi-layer Bayesian blind deconvolution method, which belongs to the technical field of remote sensing image processing. Background technique [0002] The generation of Synthetic Aperture Radar (SAR) images is affected by many factors such as hardware physical limits, operating state stability, complex environmental conditions, imaging algorithm performance, and cost constraints, and it is difficult to get rid of distortion, blur, aliasing, noise, etc. If there is such interference, the image quality will be reduced to varying degrees, resulting in a resolution that cannot meet the requirements of scientific research and application. [0003] Image deconvolution is an economical, feasible and effective original SAR image estimation (image resolution enhancement) technique, in which image blind deconvolution first performs prior modeling on the original SAR image and point spr...

AI Technical Summary

Problems solved by technology

[0006] First, the effect of deconvolution processing on a single frame image is not ideal;

[0007] Second, the setting of prior model parameter values ​​is not convenient and effective enough;

[0008] Third, it is difficult to suppress pulse interference in SAR images by using Gaussian distribution to characterize noise

Images