Image Edge Characteristic Information Detection And Image Assessment

In the 21st century, with the rapid development and improvement of modern information technology, image has become the most important source of information in today human society because of its accuracy, visual, high efficiency and wide adaptability. Image quality correct assessment is an important research topic of the image information in the field of engineering. Image quality subjective assessment method is consistent with the actual situation, but this method of assessment is impacted deeply by different observers, the image type and observing environment and other factors, the process of evaluation methods is tedious and the visual psychological factor is difficult to express by mathematical model, causing the results imprecise. The traditional peak signal to noise ratio in the image quality assessment ignore human visual system characteristics, which leads to assessment and vision inconsistencies. Today, the image quality assessment which is objective and consistent with the observed quality of the human eye has become an important research direction.Image edge is the most basic features of the image and the basis of analysis in understanding image. Changes in information of image edge mean changes in the image basic structure or content. Meanwhile, the masking of the human visual makes edge distortion sensitive to human eyes. Therefore, the image edge is an important role in the human eyes for understanding information, and it is an important factor for image quality assessment.This paper presents the method based multiscale edge integration extracts image edge feature. This method uses dyadic wavelet transform to extract the image multiscale edge. According to edge correlation theory, the noise is removed. Then image edge feature in different scales are integrated based on cross-scale transfer characteristic of signal and noise wavelet transform module. Experimental results show that the method based multiscale edge integration, compared with traditional edge detection method, not only suppress noise, but also ensures the accuracy of the positioning edge.This paper presents a image quality assessment method based on Multiscale Edge Structure Similarity(MESS), which is combined with the image edge and Structure SIMilarity(SSIM). MESS will replace structure of the image edge to structure of the SSIM, so that MESS is a description of the image edge. The results show that MESS considered the importance of edge information for the human eye perceiving structural information, so that this method is better than the SSIM, and more in line with the characteristics of human visual perception.