Optimal Moving Windows for Real-Time Road Image Processing

In this paper, a "Moving Window" scheme for detecting lanes and obstacles from the images captured by a CCD camera in an automobile is proposed. Processing the input dynamic images in real time requires high performance hardware as well as efficient software. In order to relieve these requirements for detecting lanes and obstacles from the images in real time, a "Moving Window" scheme is proposed. The size of the moving window is optimally determined based upon road and automobile conditions to be big enough to detect the lanes and obstacles and to be small enough to process the detection procedure in real time. For each image frame, the "Moving Window" is newly defined and it is moved in a certain direction that is predicted by the Kalman filtering technique. By detecting the left and right lane marks of the driving lane, it becomes possible to search the obstacles within the lane. The obstacle can be verified through the correlation between the stochastic characteristics of the suspected obstacle and the actual obstacle in the database. The feasibility of the proposed algorithm is demonstrated through the simulated experiments of freeway driving. Keyword : Optimal moving window, Lane detection, Obstacle detection, Kalman filter.