Abstract
An aged society is approaching in Japan. Electric-powered mobile scooter becomes widely used for the supports of livelihood and locomotion of aged.
Meanwhile, accident of mobile scooter is of a problem. In order to realize lane keep assist and obstacle avoidance, information of the shape of the roadway is needed as a road map. This paper shows the technique of road map generation from camera images. Road images are transformed to birds-eye view images. Road map are generated with concatenated birds-eye view images. We propose the method for birds-eye view concatenation. Road images are obtained a camera on mobile scooter. Birds-eye view images are generated from road images with perspective transformation. To concatenate images, some characteristic points are picked up by SURF. Texture information is not even over the birds-eye view image according to expansion of road image. So, characteristic points are picked up from high-resolution area of birds-eye view images. Experimental results show the high resolution road map is generated. It, however, contains non-continuous part according to concatenation errors. We will approach the improvement of increasing number of effective characteristic points.
