Research And Application Of Multi-focus Image Fusion Algorithm Based On Multi-resolution Analysis

Image fusion is a key technique in image processing, and multi-resolution analysismethod is commonly used method in the field of image fusion and plays an important role inimage fusion. This paper mainly studies the multi-focus image fusion algorithm based onmulti-resolution analysis, mainly including the following three contents:First of all, this paper emphatically introduces the basic theory of the multi-focus imagefusion, including the imaging principle and its own characteristics of the multi-focus image,the three levels of the multi-focus image fusion, and the main algorithms of the multi-focusimage at pixel level, and the main algorithms have been carried on the comparative analysis,besides, the performance evaluation systems of the fused image effect are briefly proposed toproviding the judgment standards for the evaluation of the follow-up experiment results;Secondly, this paper proposes a multi-focus image fusion algorithm based on dual-treecomplex wavelet transform. For the imaging principle and its own characteristics of themulti-focus image and the correlation of the high frequency coefficients and low frequencycoefficients after transformation, the fusion rule based on “local region-energy matching” and“improved gradient operator” is applied in the high-low frequency coefficients. The fusedalgorithm makes full use of the good expression ability of the details of dual-tree complexwavelet transform. The objective evaluation indexes show that the clarity of fused image canbe significantly improved;Finally, according to the marginal distribution characteristics of wavelet coefficient, thispaper introduces the gaussian mixture model, combines it with wavelet transform andproposes a new fusion algorithm. The standard deviation estimated in the gaussian mixturemodel is used to choose coefficient in fusion rules of high frequency sub-band, while region entropy is adopted as a measurement of image clarity in the fusion rules of low frequencysub-band. The simulation results show that the contrast and the edge of fused image obtainedby the proposed algorithm have been enhances, and it greatly improves the quality of thefused image.