Medical Image Fusion Algorithms Based On Sparse Representation

With the rapid development and popularization of medical imaging equipment,Medical imaging technology has become an essential means of checking and diagnosing diseases in clinical medicine,however the information which is provided by the single-model medical images has some limitations.Therefore,the medical image fusion is proposed by the scholars.Medical image fusion combines their respective advantages between different models of medical images to make up the limitations of single-model medical images,which provides the information of anatomical structure,physiological status and pathological characteristics of human body in the single medical images more intuitively.According to sparse representation can extract a few features to represent all the information of the image,therefore,the paper combines the sparse representation with image fusion technology,the main contents are as follows:1)According to characteristics of complexity and diversity of medical image,the paper proposed an adaptive medical image fusion algorithm based on online dictionary learning.The algorithm first applies the theory of online dictionary learning to train over complete dictionary of source images and improves the adaptive ability of image feature extraction;Then it adopts orthogonal matching pursuit algorithm for sparse decomposition of source images to gain sparse codes,and reduces fusion data dimension.Besides,it adjusts the fusion rules adaptively according to the degree of energy difference and gradient difference of sparse code between source images.If the degree of energy difference is more than of gradient difference,the sparse codes are fused on the basis of the rules of maximum energy.On the contrary,the sparse codes are fused according to the rule of maximum gradient.Finally,the fused sparse codes and over complete dictionary are reconstructed to obtain fused images.The experiment result shows that compared with multi-scale geometric analysis,k-singular value decomposition and other image fusion algorithms,objective evaluation indexes of images fused by this algorithm-information entropy and edge evaluation factor improve.Subjectively,the texture of the fused image is clearer,and the contrast ratio is higher.In addition,this algorithm can well preserve edge information of source images effectively.2)According to the problem that regularized orthogonal matching pursuit need to forecast sparsity level in the in compressed sensing,and would be unstable if the sparsity level is improperly estimated,the improved regularized orthogonal matching pursuit is proposed.Because observation signal can inherit the original signal,an adaptive weak selection criterion is introduced in the process of selecting candidate atoms.A weak selection strategy is set by the information of observation signal,then it adjust the sparsity adaptively.The algorithm is applied in the medical image fusion of the compressed sensing,then a fusion rule based on the similarity of observation signal structure is proposed.Taking the weight of observation signal information as fusion rule,when the structural similarity between the observed signals is more higher,it shows that the original signals have the same similarity.In the same way,choosing the larger information of observation signal as fused observation signal when the structural similarity between the observed signals is more lower.The experimental results show that the reconstructed image quality of the improved ROMP algorithm is better than OMP,ROMP,SAMP,the peak signal to noise ratio is improved by 6%.When applied to medical image fusion,the fused image has good human visual effect and retains most of the feature information in the source image,further it can get high quality fusion results in a short time.