基于秩约束的车辆遮挡点恢复方法

为了恢复车辆跟踪过程中丢失的特征点,提出了一种基于秩约束的车辆遮挡点恢复方法,该方法利用所有车辆图像点组成一个秩3的矩阵,并利用该特性构造一个投影矩阵,利用该投影矩阵求到车辆遮挡点,再将求到的车辆遮挡点代替图像中的车辆遮挡点,经过多次迭代,最后求到车辆遮挡点的真实位置。该方法的优点是所有的图像及图像点都平等地看待,模拟试验和真实试验表明,该方法具有鲁棒性好、收敛性好及误差小等优点。

Traffic Occlusion Recovery Method Based on Rank Constraint

To recover the occlusions during the feature tracking,a traffic occlusion recovery method based on rank constraint was presented.The projective matrix that is from all the traffic image points including all the traffic occlusions that compose a matrix with rank 3 was used to find all the traffic occlusions.At the same time,the obtained traffic occlusions were substituted for the old traffic occlusions.After iterations,the real positions of the traffic occlusions can be found.The innovation of the algorithm is that all the images and all the image points are treated uniformly.The experiments with both simulated data and real data show that the algorithm is efficient,robust and has good property of convergence with small errors.