项目名称: 基于稀疏表示的视觉感知图像质量评价及其应用研究
项目编号: No.61201328
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 电子学与信息系统
项目作者: 程光权
作者单位: 中国人民解放军国防科学技术大学
项目金额: 23万元
中文摘要: 客观图像质量评价是图像与视频处理领域中的一项基本而又富有挑战的技术。本项目拟从人类视觉系统的视觉感知和视觉心理特性研究出发,探索视觉感知图像信息过程的特点,深入研究图像数据的视觉结构特征,在稀疏表示的基础上挖掘图像数据的自然属性,针对典型失真类型图像数据特点,建立与人类视觉感知保持一致的主客观相结合的新型图像质量评价方法,并在标准图像测试库上仿真验证算法的有效性;同时,研究将新型图像质量评价方法嵌入典型图像处理系统的优化设计中,并设计高效的非线性优化算法进行求解。项目的研究成果将对视觉质量评价方法研究的进一步完善,以及基于视觉感知的图像处理系统设计均具有重要意义。
中文关键词: 图像质量评价;人类视觉系统;视觉感知;视觉心理;稀疏表示
英文摘要: It is a fundamental and challenging work to find an efficient metric of objective image quality assessment, which is an important approach for many image and video processing tasks. Inspired by the characteristics of visual perception and visual psychology, the sensing and information processing mechanism of human viusal system are explored in this project. Following the visual characteristic of image data,the intrinsic and natural property of image data is explored via efficient image sparse resprentation. Some novel objective image quality assessment methods are proposed to fit the typical image distortion models, which are cosistent with the subjective assessment of human beings. The peformance of our methods will be validated by experimental results on standard image databases. Meanwhile, these novel metrics will be embedded into typical image processing system to improve the algorithms' perfomance with efficient nonlinear optimization process. The production of this project will promote the devolopment of visual image quality assessment and visual peception based image process system.
英文关键词: Image quality assessment;Human visual system;Visual perception;Visual psychology;Sparse resprentation