华中科技大学学报：自然科学版2011，Vol.39Issue(12)：102-106,5.
基于方向特征融合的感兴趣区域提取算法
Extraction of region of interest using orientation feature fusion
摘要
Abstract
Low-level features in image would influence the points of extracted region of interest（ROI）.Thus,an extraction ROI algorithm was proposed based on orientation feature and visual attention map.The algorithm got the real ROI in image by the data of eyelink,and calculated similarity between superposition map and the real ROI,then integrated superposition map and the Stentiford visual attention map.The algorithm got suitable weight and ROI of image after feedback and adjusted weight.According to experimental data,the different effects of extraction ROI by orientation feature were analyzed.Experimental result shows that the algorithm can exactly extract ROI in image when direction feature is obvious and there are a few objects in image.关键词
特征提取/反馈/方向特征/融合/感兴趣区域/注视焦点Key words
feature extraction/feedback/orientation feature/fusion/region of interest（ROI）/attention focus分类
信息技术与安全科学引用本文复制引用
陈再良,邹北骥,李海冰,史操..基于方向特征融合的感兴趣区域提取算法[J].华中科技大学学报：自然科学版,2011,39(12):102-106,5.基金项目
国家自然科学基金资助项目 （）
国家自然科学基金重大研究计划资助项目 （）
湖南省自然科学基金资助项目 （）
中央高校基本科研业务费专项资金资助项目 （）
