Enhancement Technique for Improving the Reliability of Disparity Map under Low Light Condition

Abstract

Abstract A collision avoidance system is an automobile safety system designed to reduce the severity of an accident. It uses a combination of radar, laser and camera to detect the obstacles and avoid an imminent crash. Being relatively cheap, and capable of giving depth information stereo camera pair alone can be used for obstacle detection. Prime step in obstacle detection by stereo camera system is the generation of disparity map which gives the relative distance for all the obstacles in the scenario. But the accuracy and reliability of the disparity map is decided by quality of stereo images which might vary with external lighting conditions. Images captured under low light or night time conditions will have very poor quality and thereby affecting the reliability of the disparity map. In this paper, pre-processing the stereo images using histogram based enhancement techniques along with filtering is proposed to improve the reliability of the disparity map. The enhancement techniques like histogram equalization with the combination of low pass and median filter is used to process the stereo images before disparity map creation. Simulation results indicate the improvement in the disparity map accuracy and scene visibility with the proposed enhancement technique for the stereo image captured in low light.