Research On Quality Evaluation Method Of Color Multiexposure Image Fusion Based On Perceptual Characteristics

In recent years,with the rapid development of multimedia and network technology,digital images,as the main information carrier,are widely used in medical imaging,public safety,computer vision,communications,and many other fields,and their quality requirements are becoming more and more strict.However,restricted by hardware conditions,the dynamic brightness range of the natural scene obtained by the camera is much smaller than that in the real environment,so the captured image has over-exposed or underexposed distorted areas.Multi-exposure image fusion refers to the direct fusion of a group of images with different exposure levels into high-quality digital images with clear details and uniform brightness,thereby effectively solving the problem of insufficient existing hardware conditions.So how to judge the pros and cons of the multi-exposure fusion technology and provide the basis for algorithm optimization is the significance of the multiexposure fusion quality evaluation method.Combined with visual perception characteristics,this paper evaluates color multi-exposure fusion images.The main research contents are as follows:(1)Considering the importance of color distortion to human visual perception,traditional image fusion quality evaluation methods are not suitable for the characteristics of multi-exposure fusion images and multi-reference images.Therefore,the algorithm combines the multi-channel characteristics of the human visual system to design a color multi-exposure image fusion quality evaluation model based on chromaticity and gradient features.First,the color space conversion is used to extract the chrominance information,and then the gradient features of the image are calculated through the Gaussian fuzzy luminance channel,and the image is filtered by the contrast sensitivity function,which is used as the weighted information of the fusion feature similarity,and the final quality score is calculated.Experiments on relevant databases show that the algorithm can effectively evaluate the subjective perception quality of color multi-exposure fusion images,and the computational complexity is improved compared with the mainstream algorithms.(2)To better evaluate the multi-exposure fusion images,consider their fusion characteristics and the fact that the human eye is more sensitive to edge,structure,and other information.A new multi-scale color multi-exposure fusion image quality objective evaluation model based on multi-feature similarity fusion is proposed.First,the edge features,structural features,and saturation features of the input image are calculated,and the fusion idea is used to combine the features of the source image sequence.The image is fused into a new reference feature map and the similarity is calculated with the fused image,and then the spatially weighted fusion similarity map is calculated on a single scale to obtain the quality score.Finally,the scores under multiple scales are fused using the weight to obtain the final quality score.The experimental results show that the model has better subjective visual perception correlation and evaluation performance compared with other mainstream models.