Research on real-time detection of moving target using gradient optical flow

Abstract

Real-time detection of moving target is a key part of surveillance, robot navigation, and tracking system. Based on the detail analysis of optical flow field, an integrated method for target velocity and location estimation, which incorporates the differences of images, Gaussian filter to the computation of gradient optical flow, is proposed in this paper. Firstly, in order to get the low computation demand needed for the optical flow calculating and especially to reduce the noise, the differences of the images are obtained. Secondly, the Gaussian filter is used to eliminate some noise. These two above steps are very important. Thirdly, gradient optical flow based on the two constraints of apparent brightness remaining constant and the velocity smoothness is adopted. Finally, the experimental results show the effectiveness and robustness of this method.