A Recognition Algorithm For Color Vision Deficiency Based On Recoloring

Recent years,with the explosive growth and maturation of computer science and multimedia technology,color perception has become one of the most important ways for people to visual communicate in daily life.There are about 2 billion human with color vision deficiency in the worldwide,who cannot correctly perceive color images under normal situation.Unfortunately,the existing recognition algorithms for color vision deficiency have some shortcomings in terms of practicability and performance.This paper summarizes these shortcomings,and proposes corresponding improved algorithms.The main contents of this paper include:1,research on recognition algorithm for color vision deficiency based on color-to-gray.First,in order to reduce the time complexity of the algorithm,this paper uses region growth algorithm to segment original image,and each region represents a color in the segmented image;Second,use an improved color difference model to calculate the distance between arbitrary segmented region pairs;Finally,construct a global grayscale objective function based on the calculated distance,and can obtain the resulted grayscale image by solving the minimum of the objective function.2,research on recognition algorithm for color vision deficiency based on self-recoloring.In order to make algorithm adaptively determine the optimal number of clusters for each image,this paper adopts an improved octree quantization algorithm;Second,in order to improve the efficiency of algorithm,all the unrecognizable colors for human with color vision deficiency are converted sequentially according to color confidence;Finally,the stopping condition in the iterative conversion process for the unrecognizable color is the contrast between this color and other colors with higher confidence,which can ensure that the contrast between different colors in the recolored image will not be lower than the contrast between the corresponding colors in the original image.3,the experimental part uses test samples from different scenes to test two recognition algorithms for color vision deficiency proposed in this paper.And compare the recognition algorithms proposed in this paper with some existing algorithms in terms of subjective evaluation,objective evaluation and time complexity,to conclude that the algorithms proposed in this paper have a better efficiency.First,compare and analyze the global colorto-gray algorithm proposed in this paper with three state-of-the-art algorithms.From the experimental data,it can be concluded that the global color-to-gray algorithm can improve E-score by approximately 10.76%.Second,compare and analyze the self-adapting recoloring algorithm proposed in this paper with three state-of-the-art algorithms.From the experimental data,it can be concluded that the contrast between different colors in the recolored image is higher than the contrast between corresponding colors in the original image and recolored images obtained by other recoloring algorithms.