Research On Iamage Fusion Algorithm Based On Low Rank Decomposition And NSST

Image fusion is a technique for aggregating information from multiple images into the same image.In essence,it is a collection of key information from different images.This not only requires that the source image be the same scene,but also tests the ability of the algorithm to transfer key information.Currently,image fusion algorithms are in full bloom.With the advent of the era of big data,the complexity of data is not the same,which leads to a single algorithm being stretched when faced with high-precision image fusion.In order to improve the operational efficiency of the algorithm,this thesis uses the idea of data dimensionality reduction to reduce the complexity of the data.At the same time,in order to reduce the loss rate of information such as image edge structure by a single algorithm,the idea of organic integration of multiple algorithms is adopted.Therefore,this thesis proposes an image fusion algorithm based on low rank decomposition and NSST.Research has found that low rank decomposition theory has remarkable effects in processing complex background images,and can effectively shorten the running time of the algorithm.Combined with the translation invariance and directional selectivity of NSST for image processing,the algorithm can obtain fusion images with rich background information and prominent key information while ensuring the running efficiency.This thesis starts with low rank decomposition theory and combines NSST algorithm to deeply explore the fusion algorithm for multi-source images in this model.The experimental object is a multimodal image group,and its main content is:(1)The first two chapters of this thesis mainly introduce the current research background and significance of image fusion technology,as well as the current status of algorithm research.The second chapter introduces in detail the basic theory and framework of the algorithm proposed in this thesis.(2)In Chapter 3,this thesis proposes a dual channel Pulse coupled neural network(d PCNN)image fusion algorithm(RN1)based on a dual decomposition model.The RN1 algorithm is based on the framework of RPCA low rank decomposition and NSST multiscale decomposition,combined with preprocessing algorithms,and utilizes the d PCNN algorithm,improved local weighted energy algorithm,and eight neighborhood Laplace energy sum algorithm to fuse images.Experiments have shown that the RN1 algorithm is significantly superior to existing contrast algorithms in general,and effectively solves some problems with infrared and visible light.(3)In Chapter 4,this thesis proposes an improved image fusion algorithm based on the dual decomposition model(RN2),which uses the improved low rank decomposition ROSL,and then introduces the CSR algorithm to fuse low frequency coefficients to improve the regional energy algorithm fusion base layer.Experiments have shown that RN2 algorithm is not only significantly superior to the comparison algorithm in terms of subjective and objective indicators,but also has greatly improved compared to RN1 algorithm,making significant progress in multimodal image fusion.