基于ORB特征的快速目标检测算法

Rapid moving object detection algorithm based on ORB features

在动态场景运动目标检测下提出了一种新颖的快速目标检测算法,针对SURF算法不能满足实时性的需要,提出基于ORB(oriented FAST and rotated BRIEF)特征的特征点匹配算法,接着采用八参数旋转模型,结合最小二乘法求解全局运动参数进行运动补偿,最后使用帧差法来获得运动目标。在此过程中采用PROSAC(progressive sample consensus)算法来去除外点。实验结果表明,该算法不仅保持了SURF本身的优越性,而且提高了检测速度,可以实时准确的检测出运动目标。

A novel method for detecting object under dynamic scene is proposed. Aiming at SURF algorithm can not meet the needs of real-time application, a feature point matching algorithm based on the ORB （ oriented FAST and rotated BRIEF） algorithm is presented. It uses the rotation model of eight parameters, combines with the least squares method to solve the global motion parameters for motion compensation, and the frame difference method is last used to get a moving target. In this process, PROSAC （progressive sample consensus） algorithm is used to eliminate the outliers generated in the process of feature point matching. The experiments demonstrate that, this method not only remains the advantages of SURF but also improves the detection rate, and can effectively detect moving object in real-time and accurately.