Researches On Perceptual Blur Image Quality Assessment

Images may suffer from different level of distortions during the process of acquisition, compression and transmission. Blurriness is one of the common distortions. The increasing number of demanding consumer video applications, as exemplified by cell phone and other digital cameras, has boosted interest in no-reference objective image and video quality assessment.This paper concludes existing no-reference blur metrics and classifies them into several categories. For each category, we evaluate its performance, complexity and extensibility as well as its advantages and disadvantages.This paper presents a no-reference blur metric, which evaluate image quality at the receiving ends without reference, and it is especially suit for images that blurred during acquisition. This metric focuses on edges of the blurred image and calculates the ratio of blurred edges. It analyzes the image on pixel scale and reaches high accuracy. Besides, this paper presents an edge extraction method and an image segmentation method that support the blur metric.The experimental results show that this metric matches well with perceptual scores. In the meanwhile, it has little complexity and it is suitable for applications.