| Color cast detection is a hot issue in the field of computer vision.Now,the color cast detection methods which often used are gray balance,histogram statistics,prior knowledge and so on.But,these methods have a certain degree of limitation.The research of visual saliency put forward a novel thought for the domain of computer vision.In recent years,many traditional application models have met a bottleneck period gradually,and the visual saliency model is booming.Then many investigators wish that the new thought can make big progress in computer visual.Based on the two points above,this paper will apply saliency to color cast detection,the main research work is as follows.According to the research found in the process of learning the model of saliency,two kind of color cast detection methods based on visual saliency are put forward in this paper.First,a video color cast algorithm based on super-pixel and Lab saliency is proposed.This paper extract features of super-pixel,features fusion and normalization to obtain saliency map which can judge the degree of color cast tentatively.Then,using SVM to class and grade.Secondly,an algorithm based on saliency regional characteristics is proposed in this paper.This method aims to obtain salient region to classify for the normal and color cast videos.Then,judge the type and degree of video color cast.Compared with the equivalent circle method and other traditional algorithm,the method in this paper not only improve the accuracy of color cast detection,the degree of partial color is also more close to the subjective judgment.Beside,the method in this paper judge the type of the color cast.In short,this paper put forward a new research direction. |
