项目名称: 物体形状部分视觉显著性度量及其应用
项目编号: No.61501208
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 无线电电子学、电信技术
项目作者: 赖重远
作者单位: 江汉大学
项目金额: 19万元
中文摘要: 本项目研究物体形状部分的视觉显著性的度量方法。定义物体形状部分及其视觉显著性,给出其应满足的三个性质:完备性、独立性和直观性。通过简化部分形式,将部分视觉显著性度量表征为感知相关三角形视觉显著性的最大值。通过挑选和量化几何因素对感知相关三角形的影响,得到满足上述三个性质的视觉显著性度量的表达式。通过设计视觉心理学实验,验证提出度量的视觉一致性。提出度量可用于基于骨架的形状解析、基于角点的手势识别和基于视觉的形状编码,从而为推动计算机视觉、模式识别和多媒体技术等相关领域的发展提供一种新的工具。
中文关键词: 形状表示;视觉显著性;形状解析;手势识别;形状编码
英文摘要: This project investigates a method to measure the salience of object shape parts. The object shape part and its visual salience are defined, and its three required properties, namely completeness, independence, and intuition, are given. By simplifying the part form, the part salience will be characterized as the maximum salience among the perceptual relevance triangles. By selecting and quantifying the influences of the geometric factors on the perceptual relevance triangle, the visual salience measure that satisfies the above three properties is expressed. By conducting the visual psychology experiments, the visual consistency of the proposed measure is verified. The proposed measure can be used for skeleton-based shape parsing, corner-based hand gesture recognition, and visual-based shape coding, thus providing a new tool to promote the development of fields related to computer vision, pattern recognition, and multimedia technology.
英文关键词: shape representation;visual salience;shape parsing;hand gesture recognition;shape coding