Video Vehicle Detection Based on Texture Analysis

This paper presents a video vehicle method that combines local binary pattern and motion histogram. First, use LBP modeling and updating the background in video image. Second, detect the video vehicles by means of the motion histogram. Finally, eliminates shadows from the detected vehicle region, and improves accuracy of vehicle detection. Experiments in some vehicle database show that our method has a better performance.