Improvement Algorithm of Background Updating Based on Kalman Filter

In the process of target detection under complex background, the color of target (foreground) is often similar to the color of the background which causes noise during background updating, and the integrity of the target extracted from the image has been affected. In view of this situation, an improved algorithm based on Kalman filter is proposed. The algorithm analyzes the difference of background subtraction results in different channels of multiple color spaces, then the mix-color space is built and the detection results of each channel in the mix-color space are integrated. The background updating algorithm based on Kalman filter has been improved from the update region and the updated background was closer to real background. The experiment results show that under the complex background, the algorithm can remove the background noise effectively with high accuracy and good applicability.