Simultaneous Localization and Mapping of Mobile Robot Based on Image Enhancement

For the camera optical system aberration, scene image observed by the robot will be disturbed by noise such as illumination, motion, and other noise interference, which will bring visual positioning error, lead to wrong expression of topological structure for the environment, and affect the robustness of Simultaneous Localization and Mapping (SLAM). How to control the accumulation effectively and enhance the effectiveness, robustness and real-time of visual SLAM is a challenging research topic. The fuzzy image set is constructed according to the blurred image. This paper proposed a processing techniques of fuzzy image based on image enhancement. (1) The threshold algorithm is employed to improve the histogram equalization algorithm in image enhancement field. (2) The fuzzy set of illumination is constructed offline and the gray scale histogram of the original image is stretched in nonlinear to increase the dynamic range of the gray value of the pixel. (3) The illumination fuzzy set is integrated into the SLAM, and then the visual odometry, closed-loop detection module of SLAM are tested on the basis of the quantitative experiment in the actual scene. The experimental results show that the proposed algorithm is robust to SLAM and have good real- time performance.