| 基于视觉注意模型和进化规划的感兴趣区检测方法 |
| |
| 引用本文： | 张菁,沈兰荪,高静静.基于视觉注意模型和进化规划的感兴趣区检测方法[J].电子与信息学报,2009,31(7):1646-1652. |
| |
| 作者姓名： | 张菁 沈兰荪 高静静 |
| |
| 作者单位： | 北京工业大学信号与信息处理研究室,北京,100124 |
| |
| 基金项目： | 国家自然科学基金(60772069,60402036);;北京市自然科学基金(4062006)资助课题 |
| |
| 摘 要： | 根据生物注意机制，该文提出了一种基于视觉注意模型和进化规划的感兴趣区检测方法。采用进化规划方法分割图像候选区域；区域兴趣度由视觉注意模型产生的局部显著和进化规划计算的全局显著共同度量。在视觉注意模型中，图像经过小波多尺度变换和计算中央周边差得到局部显著度。注意焦点在显著度增强因子的作用下，选取候选区域得到感兴趣区。实验结果表明，所提方法检测的感兴趣区更接近人眼的视觉注意机制，并取得了较为满意的对象检测和兴趣度量结果。
|
| 关 键 词： | 计算机视觉 视觉注意模型 感兴趣区 进化规划 显著度 注意焦点 |
| 收稿时间： | 2008-6-3 |
| 修稿时间： | 2009-2-27 |
Region of Interest Detection Based on Visual Attention Model and Evolutionary Programming |
| |
| Zhang Jing,Shen Lan-sun,Gao Jing-jing.Region of Interest Detection Based on Visual Attention Model and Evolutionary Programming[J].Journal of Electronics & Information Technology,2009,31(7):1646-1652. |
| |
| Authors: | Zhang Jing Shen Lan-sun Gao Jing-jing |
| |
| Affiliation: | Signal & Information Processing Lab, Beijing University of Technology, Beijing 100124, China |
| |
| Abstract: | According to biological attention mechanism, a region of interest detection based on visual attention model and evolutionary programming is proposed in this paper. Candidate image regions are segmented with evolutionary programming algorithm. The interest of a region is measured with local saliency produced by visual attention model and global saliency based on evolutionary programming. After wavelet multiscale transform, local saliency in visual attention model is computed by center-surround differences. Under the action of saliency enhancement factors, focus of attention selects region of interest from candidate regions. The experimental results show the proposed approach is closer to human attention mechanism and performs well for object detection and interest measurement. |
| |
| Keywords: | Computer vision Visual Attention Model (VAM) Region Of Interest (ROI) Evolutionary Programming (EP) Saliency Focus of Attention (FOA) |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《电子与信息学报》浏览原始摘要信息 |
|
点击此处可从《电子与信息学报》下载全文 |
