Research On Object Quality Evaluation Method Of Image Quality Based On HVS And Semantic Information

Image information technology is widely used in various fields of image processing, the image quality assessment has become a widely and basic problem. Reasonable evaluation of image quality has important application value in algorithm analysis and comparison, system performance optimization, Image quality monitoring of communication and so on. Image quality evaluation method is objective evaluation method by designing mathematical model to achieve the image intelligent modeling, analysis, and automatic scoring. At present the mainstream of image quality evaluation model have the structure similarity and improve of natural images statistical model and the human visual system (HVS) model. They analyze their essential attribute from the perspective of image quality evaluation system of the object and the subject. In this paper, combining the subject and object in image quality evaluation system, using structural similarity, boundary strength and the theoretical basis of HVS visual features, we are proposed based on HVS and the semantic information of the image quality objective evaluation method. The method in this paper has Partial superior performance evaluation, strong commonality, high accuracy and so on. The main research work of this paper has the following three aspects:1. The image decomposition method based on boundary strength. In this paper, using the definition of the boundary strength, combining with image decomposition approach based on gradient operator mathematical principles, we design intensity the based on boundary strength of image decomposition method. Under the same weight breakdown threshold, the decomposition results of this method can better keep the target boundary and eliminate the boundary of the area of the background.2. Image visual repair method is based on HVS characteristics. In this paper, using the visual multi-channel and brightness is nonlinear of the HVS, combining with image visual error figure, we design a weight coefficient method which is based on visual error visibility and brightness nonlinear repair. Using the visual contrast to cover effect and the entropy effect of the HVS, combining with image visual error figure, we designed a weight coefficient method which is based on HVS visual masking effect repair; The finally, we have the weight coefficient method which will be the two weights coefficient based on HVS characteristics of the image obtained from fusion to repair. The method has the effective visual repair to distortion image, and can largely replace the image in cerebral cortex’s perception of the HVS.3. Image quality assessment method is based on HVS and semantic information. In this paper, using the properties of image points in the border, combining with the definition of the structure similarity, we put forward the theory of boundary similarity, and design based on the boundary similarity of image quality evaluation mathematical model according to the theory. Then, using image point show different information in different regions which is mainly structural information in plain area and show the boundary information in texture or the fringe area. We design a kind of improved image quality evaluation mathematical model based on content. These two models are very well in the local image quality assessment evaluation result, and the evaluation method based on image content has a strong commonality and high accuracy. And we is based on HVS characteristics to repair the distortion of the image as to evaluate the image. We can accordingly get the image quality evaluation method based on HVS. The method show a more superior performance in image quality evaluation on the structural distortion and half structural distortion. Finally, this article uses the image quality assessment to evaluate the performance of image denoising algorithm. The purpose of it is used to analyze the advantages and disadvantages of various denoising algorithm and the denoising characteristic, and study the application value of the image quality assessment.