A Real-Time Image Recognition Method for Downward Stairways Using Hough Transform
Kakuzi OGAWARA, Toshiyuki Fukuda, Takuya Naka · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C · 2004
In this paper, we propose a real-time image recognition method for downward stairways. Our proposed system uses a part of images is various situations. This system judges from existence of stair slope with databases made beforehand to predict danger. In Addition, it observes the distance to downward stairways. If this distance becomes smaller than a threshold value, the system starts to measure an inclination angle between a floor and downward stairways with two CCD cameras by stereovision. It uses a Hough transform in order to extract lines from edges of stair steps, too. One of characteristics of this paper is a limitation of polar coordinate area to search maximum values. In result, processing time becomes about 200 msec and maximum recognition time is about 50 msec. We consider that these scores are effective to control an electric wheelchair because in the processing time the distance it moves is small enough for a human.