Robust Image Fusion Method Based on Statistical Model of Wavelet Coefficients

Abstract

The invention discloses a robust image fusion method based on the statistical model of wavelet coefficients. It firstly prepares two noisy images to be fused, respectively performs dual-tree complex wavelet transformation on the noisy images, and decomposes the two images into The low-frequency sub-band and high-frequency sub-band; then extract the contraction operator for the high-frequency sub-band of the two images; based on the establishment of a joint probability distribution and an independent marginal distribution between the two coefficients to be melted, a joint and marginal distribution-based Matching degree; Propose a saliency measure based on divergence and information entropy based on marginal distribution, and combine it with the contraction operator; Under the framework of the statistical model, the present invention proposes a parameter estimation method using moment estimation, and denoises the denoising algorithm The noise parameter replaces the moment in the fusion algorithm, so that the fusion rule framework is not affected by noise. At the same time, the denoising operator is combined with the fusion weight to accurately remove the noise while retaining the detailed information of the source image, which further improves the denoising of the fusion algorithm. noise effect.

Description

technical field [0001] The invention relates to an image denoising fusion algorithm, in particular to a robust image fusion method based on a statistical model of wavelet coefficients. Background technique [0002] The purpose of image fusion is to extract, preserve and integrate useful and important information from multiple source images into one image. The source image contains important feature information of different frequency components, which need to be extracted and analyzed at different scales, so as to more specifically determine the content to be retained in the final fusion image. The current multi-scale image fusion method fuses the sub-band design fusion rules at different levels to maximize the fusion of detailed information on different high-frequency sub-bands, and provides accurate image scene information for subsequent detection, tracking or observation work. [0003] Among the multi-scale transforms, wavelet transform is widely used in the field of imag...

AI Technical Summary

Problems solved by technology

[0005] However, in the process of image digitization, transmission, and storage, the image quality is often affected by the interference of imaging equipment, light, temperature and other external factors, and the important details are covered by noise, which not only affects the visual effect, but also directly affects the quality of the fused image.

Images