LS-SVM for a point target motion measurement

In order to determine the 3D motion of a point target from a monocular moving camera, a locating method is proposed by taking advantage of the property that the positions of the target in a short time are relative to each other and the kernel trick of least squares support vector machine (LS-SVM). This method has better noise immunity due to its robust estimation property, and has high accuracy as the kernel function added to the trajectory reconstruction, and has high efficiency because of the incremental and decremental technique. To illustrate the performance of the method employed in the paper, contrastive simulation experiment on location error and execution time is designed. The results show the accuracy and efficiency of the method. This method is applicable for high precision measurement of motion targets by passive observations on various mobile platforms.