基于对象性测度估计和霍夫森林的目标检测方法

实现高效准确的目标检测算法在视频监控、自动导航等诸多领域至关重要的作用。针对现有目标检测算法速度不高且鲁棒性差的缺点,提出了一种基于对象性测度估计和霍夫森林的快速目标检测方法。首先,基于自下而上的视觉注意机制,采用对象性测度估计,提取图像中的物体候选集;然后,在由物体候选集确定的感兴趣区域内进行霍夫森林目标检测,确定目标中心位置;最后,结合目标中心所在的对象性测度估计候选框的尺度信息,确定目标大小。实现结果表明,该方法在提高霍夫森林目标检测算法检测准确度的同时,大大提升了检测速率。

Object detection method based on objectness estimation and Hough forest

Realizing effective and efficient object detection plays an important role in computer vision and has many practical applications including video surveillance and auto navigation. In order to improve the speed and accuracy of the existing detection methods, a simple yet effective object detection method coupled objectness estimation with Hough forest was proposed. Firstly, objectness estimation was utilized to generalize a set of object proposals based on bottom up visual attention mechanism of human vision system. Secondly, Hough forest object was adopted to localize the center of the object in the region of interest which was confirmed by object proposals put forward in the last step. Thirdly, the scale of the object proposal where the center was located was employed to determine the size of the object. A set of experiments demonstrate the effectiveness and efficiency of the proposed method.