Obstacles Detection in Dust Environment with a Single Image

Abstract

Because of light scattering and absorbing in dust environment, it is difficult to detect obstacles with camera. In order to solve this problem, a method of obstacles detection in dust environment from a single image was presented. The method realized distance detection and contour detection for obstacle in dust environment. First, depth map of dust image and geometric reasoning approach based on imaging model of a camera were combined to detect the distance between camera and obstacle of any shape. Then, depth map was applied to detect the contour of the obstacle in dust environment. Namely, edges belonging to contours were selected by using depth map. The validity and feasibility of the method was fully demonstrated by the experiments. The method provides a simple and economical way to detect obstacles in dust environment.