Obstacle distance detection system for sight-disabled people

Abstract

In this study, an obstacle distance detection system is designed for sight-disabled people. This system finds distances between the obstacles and sight-disabled people and is alerted according to distance value. Images captured from two USB cameras are processed in MATLAB environment. The obstacle detection is performed in three phases. The system is calibrated in the first phase. With the implementation of calibration steps, intrinsic and extrinsic parameters of the cameras are calculated as a result of the calibration. Images are provided with the rectification of these parameters. The second phase is stereo matching step, the simultaneous images which are captured from the left and right cameras processes and disparity map is obtained in this phase. At this stage, points, are tried to be identified, where are the three dimensional points correspond to different image planes. Barriers, through the help of stereo disparity maps, are identified with discovery of the points that occurs consecutively and that have the same difference value. Then, with the help of identified difference value in the identified barrier area and geometrical location of the cameras, distance of barrier to the camera is calculated. In last phase, the sight-disabled individuals alerted with the microcontroller hardware.