Motion Object Detection Based on Adaptive Mixture Gaussian Model and Four-Frame Subtraction

摘要

In view of background subtraction is influenced by the light and the frame difference is easily affected by double shadow and hole phenomenon in target detection, a motion detection algorithm based on adaptive mixture gaussian model and four-frame subtraction using the dynamic threshold is proposed. In this paper, in order to minimize the negative influences of noise and hole, connected region domain analysis and hole filling algorithm are introduced. The experimental result shows that the algorithm has preferable adaptive performance to the scene with the detection of the moving target accurately.