Aiming at the influence of tracking-learning-detection（TLD） on illumination change

which is susceptible to target occlusion and fast motion cause target

an improved algorithm of TLD target tracking based on local binary pattern（LBP） is proposed. Firstly

the LBP algorithm is combined with the nearest neighbor classifier

so that the improved nearest neighbor classifier can obtain the bounding frame which is closer to the tracking target. In addition

when the target has good texture

the improved nearest neighbor classifier has better classification effect; Secondly

if the tracking target selected by TLD algorithm is obscured or shaken in the process of tracking

the Kalman filter is used to predict the region of the target

which can reduce the detection range of the tracking device and enhance the efficiency of the algorithm. The experimental results show that the improved tracking algorithm has better robustness

higher tracking precision and faster tracking speed than conventional TLD.