No-reference multi-focus image fusion evaluation measure based on alpha mutual information

Abstract

The invention discloses a no-reference multi-focus image fusion evaluation measure based on alpha mutual information, comprising construction of conditional Renyi entropy, construction of alpha mutualinformation, no-reference image fusion evaluation measure based on alpha mutual information and other steps. First, character of the Renyi entropy is studied to construct the conditional Renyi entropy; second, alpha mutual information is constructed according to Arimoto's conditional Renyi entropy definition, character of the alpha mutual information is studied, and the alpha mutual information is used to measure dependence degrees of two random variables; third, alpha mutual information of an input image and fused image is estimated to provide a no-reference image fusion evaluation measure that evaluates the effects of different multi-focus image fusion algorithms. The no-reference multi-focus image fusion evaluation measure based on alpha mutual information has the advantages that statistical information of the input image and fused image is comprehensively considered, informatics measure is used to measure the fused image to obtain information quantity, no reference to images is required during fusion effect evaluation, and the no-reference multi-focus image fusion evaluation measure based on alpha mutual information is widely applicable and has good evaluation effect.

Description

technical field [0001] The present invention relates to the quality evaluation of multi-focus image fusion algorithms. Combining the statistical characteristics of images, a non-reference image fusion evaluation metric based on alpha mutual information is proposed, and the metric is used to evaluate commonly used multi-focus image fusion algorithms, which belongs to digital image processing fields. Background technique [0002] Image fusion is an important technology in the field of digital image processing. It plays an important role in multi-focus images, multi-spectral remote sensing images, multi-modal medical imaging, weather forecasting, military target recognition, image-guided surgical treatment and radiotherapy planning. effect. Image fusion is the process of integrating two or more source images into one image, and the fused image contains more information than any single source image. Image fusion not only reduces the amount of data for storage and transfer, but...

AI Technical Summary

Problems solved by technology

[0006] However, for most of the current image fusion evaluations, it is usually lacking or difficult to define an accurate reference value. In most cases, subjective evaluation criteria are used to measure the fusion results.

Images