Distributed Compressed Sensing Algorithms And Its Applications In Image Fusion

Distributed Compressed Sensing is based on the theory of Compressed Sensing.It can use not only the inner correlations but also the inter correlations to improve the reconstruction accuracy of the signals.It has a wide range of applications in image processing and other fields.As an important branch of image processing,image fusion has attracted extensive attention from domestic and foreign researchers due to its broad application prospects.In view of the strong correlations between the source images to be fused in the same scene,this paper introduces distributed compressed sensing to take advantage of the inter correlations of the images to achieve image fusion.The specific work is as follows:1.Multi-focus image fusion aims to identify the focus areas of multiple single-focus images in the same scene,and fuse all the focus areas into a all-focus image.Considering the correlation of multiple single-focus images,the paper proposes a multi-focus image fusion algorithm based on distributed compressed sensing.First,obtain the low-frequency image and high-frequency image by simply comparing the variance of the source image,and further obtain the low-frequency dictionary and high-frequency dictionary;Second,we introduce distributed compressed sensing to exploit the correlations of the source images and reconstruct the fine high-frequency images,and based on all high-frequency images,a decision method is given to obtain the initial decision map;Then,the morphological processing is used to obtain an accurate decision map;Finally,the decision map is utilized to guide the source image to select the focused pixels in order to obtain the final fusion image.In the experiments,two sets of pictures are used for subjective evaluation,the proposed algorithm all obtain satisfactory visual effects.In terms of objective evaluation,three indexes are used and compared with the existing nine algorithms in ten sets of pictures.The experimental results prove that the method has excellent objective performance.Therefore,regardless of subjective vision or objective indexes,the proposed multi-focus image fusion algorithm based on distributed compressed sensing is more competitive than the existing mainstream algorithms.2.Multi-exposure image fusion aims to obtain weight maps of multiple images with different exposure levels and merge them into a high-quality image with good contrast and vivid colors according to the weights.Considering the correlations of images with different exposure levels,the paper proposes a multi-exposure image fusion algorithm based on distributed compressed sensing.First,the low-frequency image and the high-frequency images are obtained by simply comparing the variance of the source image,and further to get the corresponding dictionary;Secondly,the fine high-frequency images are reconstructed based on distributed compressed sensing,and the local contrast measurement is achieved by comparing the reconstructed fine high-frequency images to obtain a local contrast weight maps;Thirdly,brightness measurement is achieved directly based on the brightness feature value to obtain a brightness feature weight maps;Then,multiplying the local contrast weights and the brightness feature weights to obtain the initial weight maps;Finally,the guide filter is used to eliminate the noise and discontinuity in the initial weight maps,and fuse the images based on Laplace pyramid decomposition.In the experiments,the proposed algorithm is subjectively and objectively compared with the existing eight algorithms.The method is not only effective in vision,but also significantly better than the other eight algorithms under the objective index.Experimental results show that the proposed multi-exposure image fusion algorithm based on distributed compression is superior to the existing state-of-the-art algorithms in both subjective vision and objective indexes.