Introduction - If you have any usage issues, please Google them yourself
Tracking process to adapt to changes in the target lighting conditions, and the target feature for online updates, proposes a local binary pattern (LBP) features and image intensity information integration, combined with incremental linear discriminant analysis for target tracking algorithms. Track begins, in order to obtain a more accurate description of the objectives, the use of Gaussian mixture models and expectation maximization algorithm for target segmentation tracking process, through the Monte Carlo method of the target area and the background area sampled and updated feature space parameters. Get the optimal target and background classification surface finally Using Particle Filter optimal classification predict the state of the face of goal. By varying illumination simulation video and natural scenes video tracking experiment to verify the effectiveness of the proposed algorithm.