华中科技大学学报（自然科学版）2013，Vol.41Issue(6)：45-49,5.
融合运动和空间关系特性的显著性区域检测
Salient region detection of interfusing motion and spatial relationships
摘要
Abstract
Traditional visual salient models are likely to fail in complicated scenes.Thus,a salient detection approach to interfusing motion and spatial relationships was proposed.The spatial depth relationships and motion features were added on the basis of color contrast.First,a depth saliency model was built and interfused with color saliency to obtain spatial saliency map.Then the motion feature saliency,acquired by analyzing motion vectors,was applied to the process of computing the final saliency map with additive and multiplicative integration.Experiments were analyzed by several stereo image databases.Both quantitative evaluation and direct observation indicate that the approach can detect saliency regions effectively and the results are better than that of traditional methods.关键词
视觉显著性模型/深度显著性/运动特征显著性/加性融合/乘性融合Key words
visual saliency model/ depth saliency/ motion feature saliency/ additive integration/ multiplicative integration分类
信息技术与安全科学引用本文复制引用
刘晓辉,金志刚,赵安安,卫津津..融合运动和空间关系特性的显著性区域检测[J].华中科技大学学报（自然科学版）,2013,41(6):45-49,5.基金项目
国家高技术研究发展计划资助项目(2009AA01A336). （2009AA01A336）
