Multisource image fusion method based on direction wavelet domain hidden Markov tree model

Abstract

The invention discloses a multisource image fusion method based on a direction wavelet domain hidden Markov tree (HMT)model. The multisource image fusion method mainly solves the problem that the present method has simple fusion rules and an obvious blocking effect. The multisource image fusion method has the following implementation steps of: (1), obtaining an initial fusion image from the original image; (2), resolving the original image and the initial fusion image by using direction wavelets; (3) building the direction wavelet domain HMT model of the initial fusion image, and training so as to obtain an estimated value of a parameter set by; (4), obtaining the posterior probability of a high-frequency coefficient of the direction wavelet of the original image; (5), fusing the high-frequency coefficient of the original image by a remarkable measurement fusion rule according to the posterior probability; (6), fusing a low-frequency coefficient of the original image by adopting an adaptive weight fusion rule; and (7), performing inverse direction wavelet transformation on the fused direction wavelet coefficient so as to obtain a fusion image. The multisource image fusion method based on the direction wavelet domain hidden Markov tree model can extract abundant singularity characteristics of the original image and fully excavate the correlation in data, and can be used for object identification on multisource images and subsequent processing on computers.

Description

Technical field [0001] The invention belongs to the technical field of image processing, and relates to a fusion method of multi-source images, which can be used for target recognition of multi-source images and subsequent processing of computers. Background technique [0002] Image fusion is a technology for comprehensive processing of images from different sources. It was early applied to the analysis and processing of multispectral satellite remote sensing images; in the early 1980s, Daily, Laner and Todd conducted radar images and Landsat-RBV Image, Landsat-MSS image fusion experiment. In the late 1980s, image fusion technology began to arouse people's attention, and the fusion of remote sensing spectral images was gradually used to conduct geological, mineral, climate, and environmental exploration and research. After the 1990s, with the emergence of remote sensing satellites JERS-1, ERS-1, Radarsat, etc., image fusion technology has become a research hotspot in remote sens...

AI Technical Summary

Problems solved by technology

[0009] The purpose of the present invention is to propose a method based on the directional wave domain hidden Markov model in view of the existing methods that have simple fusion rules, fail to fully capture the singularity features such as edges in the image, and have problems such as fuzzy details and block effects. Multi-source image fusion method to improve the quality of fused images

Images