Abstract
This paper describes a method of obstacle detection in the fish-eye camera image for GPS-Non-Line-Of-Sight(NLOS) satellites detection. Recently the development of the autonomous mobile system(AMS) is more expected for example an auto-pilot car. A GPS measurement is widely used for estimating self-position for such AMS. It can measure the position without premeasured emvironmental data like a map-matching. However, in the urban canyon, GPS measurement accuracy deteriorates seriously using NLOS satellite. Because of obstacles between NLOS satellites and GPS antenna, electrical waves omitted from GPS satellites are reflected or diffracted by obstacles. We propose the method to detect the obstacle using a fish-eye camera image to detect NLOS satellite. Firstly fish-eye image is parted into various small segments by k-means clustering method. And the segments are divided into sky and obstacle areas according to its movement using SIFT matching. The result of a field experiment shows the proposed method is effective to detect GPS-NLOS satellites.
