项目名称: 基于数据挖掘和感知分析的非对称失真视觉质量评价模型研究
项目编号: No.61502429
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 计算机科学学科
项目作者: 周武杰
作者单位: 浙江科技学院
项目金额: 20万元
中文摘要: 立体视觉系统由于能够提供深度信息来增强视觉的真实感,给用户以身临其境的全新视觉体验而越来越受到人们的欢迎,被认为是下一代媒体的主要发展方向。本项目研究主要解决两个基本科学问题:一是如何深入挖掘非对称失真立体视觉感知质量的影响机理;二是如何将立体视觉感知特性进行数学建模,解决非对称失真视觉质量评价问题。.主要研究内容为:首先,深入挖掘单双目视觉感知影响机理,提取单双目感知特征向量,为评价建模提供输入特征;其次,采用数据挖掘等方法,力争揭示非对称失真视觉感知特性的规律,为研究双目感知特征融合机制提供科学判据;最后,采用感知调制和统计训练两种策略对单双目感知特征向量质量图进行整合,从而构建非对称失真视觉质量评价模型。.通过本项目的研究,将形成一套高效的非对称失真视觉质量评价理论与方法,使模型预测值能够更加准确地反映人类视觉主观感知,为立体视觉处理系统性能优化提供理论支持。
中文关键词: 感知分析;预测模型;视觉质量;非对称失真
英文摘要: Stereo vision system, can provide more realistic and immersive experience to end users by providing depth perception, is considered to be the next generation of multimedia. This work mainly attempts to solve the two following basic scientific problems: one is how to dig deeper into asymmetric distortion influencing the visual quality; The other is how to establish an effective mathematical model of stereo visual perception, and to solve the problem of asymmetrically distorted visual quality assessment (VQA).. The main research contents of this work: Firstly, through the deeply understanding of the monocular/binocular vision perception mechanism, monocular/binocular perceptual feature vectors provided for the VQA model are extracted; Secondly, the rule of asymmetrically distorted visual perception, which provides scientific criterion for the study of binocular fusion, will be revealed by using data mining; Finally, two strategies, perceptual modulation and statistical training, are used to integrate the perceptual quality maps of monocular and binocular perceptual feature vectors into an overall quality score. Thereby, the asymmetrically distorted VQA model is constructed.. Through this research, an effective VQA theory and method will be formed. The proposed VQA prediction model will be achieve much higher consistency alignment subjective judgments, which plays an significant role in the optimization of the stereo vision processing system.
英文关键词: perceptual analysis;prediction model;visual quality;asymmetrical distortion