Research On Image Fusion Algorithm Based On Low Rank-sparse Decomposition

The image fusion technology is currently widely used.The aim of fusion is to integrate complementary information of multi-sensor data,makes the new image more suitable for human visual perception and computer processing tasks such as segmentation,feature extraction and object recognition purposes.Pixel-level image fusion is primarily intended to observe the fusion image for the next analysis.For the fusion of video images is to detect and track the moving target,the feature-level fusion is more suitable than the pixel-level fusion.In this paper,the infrared and visual images are fused to obtain a more complete and accurate moving target.we propose a fusion algorithm of gray level confidence map based on edge information.Then qualitative analysis and quantitative analysis of the results of the test were analyzed.On the edge of the target feature extraction,this paper puts forward an improved double threshold target edge detection algorithm based on K-means cluster,it get a low time complexity,and a better representation of the edge of the characteristics of target.The main work of this paper is as follows:This paper analyzes the mathematical model of the low-rank sparse decomposition algorithm and the working principle in the field of background modeling,introduces the improved incPcp algorithm in video incremental processing,and compared with other target detection algorithm;We propose a double threshold edge detection algorithm based on K-means clustering,the algorithm in significant edge enhancement of moving target is using gray value information,and the use of pixel neighborhood information,so the correct characterization of better effect on the target edge;In order to overcome the shortcomings of the fusion detection algorithm based on gray confidence map,this paper introduces the edge information,and proposes a fusion framework based on the joint.Making full use of the temporal and spatial information of the video image,the image edge feature and gray level information are combined,and the infrared and visible images are complementary.The results of the fusion framework proposed in this paper are compared with the qualitative and quantitative analysis of non-fusion test results and other fusion test results.Finally,the validity of the method is verified by comparing the results of the final analysis.