The Research On Image Blur Assessment Based On Second Distortion

Images are widely used in every field of daily life. Especially with the rapid development of information technology and communication technology, images have gradually become the important medium to get information. Digital images play an important role in machine vision, medicine, telecommunication, astronomy biology and so on. High quality image refers to an image of which the visual system is able to accurately understand the information. On the contrary, images with all kinds of distortion types more or less will affect the understanding of information. Images are very likely distorted during the acquisition, storage, processing and transmission, which can influence the visual effect. So necessary quality evaluation and control is required.Image quality assessment(IQA) has been a basic and classic topic in the field of digital image processing. At present, image quality assessment can be divided into subjective assessment and objective assessment. Subjective assessment depends on the subjective scores of humans, and it is always labor-intensive, time-consuming. Above all, the subjective is not convenient for most applications, so it is necessary for us to put forward an objective assessment method to evaluate images’ quality via simulating the human visual system. So a lot of scholars are devoted to the research subject. Blur is one of the important factors of image quality degradation, and also distortion type most easily detected by eyes. So image blur assessment becomes an important research direction in the field of image quality assessment.This paper focuses on the study of image blur assessment and concrete research contents include following several aspects:First, we analyze the major reasons for image blur, which mainly include imaging system itself factor, natural factor and human factor. These factors cause defocus blur, motion blur, compression blur and Gaussian blur.Second, we introduce and analyze several common image blur assessment metrics which are designed with the construct of image itself. As a result, the local features of images such as texture, edge, etc., will have an effect on result, leading to the failure of blur assessment between images of different content.Third, we study a new good image blur assessment metric based on second distortion. Then, the idea and process of the algorithm are presented. By constructing a second blurred image to do the assessment from an relative angle, the new method solves the drawback of old image blur metrics and gets better performance.Fourth, by in-depth analysis of human visual system, including physiological property and psychological property, we improve and extend the method from the algorithm idea. By testing, we testify that the new metrics have better monotonicity, consistency, instability, accuracy and content independence.The image blur assessment metric based on second distortion not only resolves the drawback of common methods, but also provides a new design concept and great help for image quality assessment.