基于视觉注意机制与局部描述子的物体检测

摘要

为实现复杂图像场景下的物体检测,提出整合视觉注意机制与局部描述子技术的检测模型.通过计算探测场景的显著图及提取其SIFT局部描述子特征,采用层次化的匹配策略对任务物体与探测场景进行关键点匹配以实现物体检测.该策略能将匹配范围界定于场景中富含物体区分性信息的显著区域,并且匹配的门限也可由这些区域的显著性自适应地调节.定性及定量的对比实验验证了该模型的性能.%To detect objects in cluttered scenes, a novel detection model is proposed, which combines visual attention guidance and local descriptors representation. Firstly, the saliency map is computed and the SIFT local descriptors are extracted for input scene. After that, by matching keypoints of a hierarchical and saliency-based strategy, only the "support" local descriptors are selected to represent the distinctive features of pop-out objects. Simultaneously, the matching thresholds are adjusted with saliency weights. Qualitative and quantitative experiments on highly cluttered scenes are employed to validate the effectiveness and robustness of the proposed model.