An Enhancement Algorithm for Video Surveillance Image Based on Multi-scale Retinex

Aiming at the similarity of adjacent frame image´s background and light in video surveillance, it legitimately uses of the information between frames and builds an enhancement algorithm for video surveillance image based on multi-scale Retinex. This algorithm uses different scales and parameters Gauss low pass filter to filter each frame image, takes Minimum operation for filter results, extracts each frame background´s information, By taking the maximum operation, fuses multi frames of image´s background, obtains the video image´s accurate background for that moment, Using Retinex algorithm enhances the background. The experimental results show that after the image enhancement, the image detail information is richer and obtains more comfortable visual effect.