Blind user indoor environment mapping system for self-localization

This paper described blind user indoor environment mapping system for self-localization via webcam camera. The method used is based on robot navigation systems which capture and build panorama map of the floor. The map is then incrementally update its position as it moves throughout an environment. Image registration techniques are applied to measure translation distance and rotation angle between consecutive frames. We used Phase only correlation (POC) method to matched image differs only by a translation. For rotated image, RI-POC and Radon Transform is used to recovered rotation. The accuracy of these two methods for rotated image is compared for best result.