Research For Data Fusion Of The Distance Image And Optical Image

The information of single sensor is always imperfect, discontinuity or inexactitude. The data fusion of sensors can improve the integrality, continuity and veracity of sensor information. So, the related problems research has been hot spot. The data fusion method for laser range finder and camera can revise, fuse the information of laser range finder and camera to make the process of checking and recognizing target more easily and increase the accuracy rate.In this paper, the data fusion procedure of laser range finder and camera have been divided into four parts: data collection and preprocessing, the scene feature drawing , the calibration and registration of laser range finder and camera, the data fusion of laser rang finder and camera. The research content and fruit are provided as follows:A) Nowadays, most of the registration methods at home and abroad have certain limits, such as the firmly in comparative position and pose of laser range finder and camera, the reference features and road remarks. This paper points out a new method. This method extracts the edge information from the images of laser range finder and camera firstly and gets the imperfect lines. Then defining the registration errors by the distance of perfect lines in two kinds of images. Searching the min-value of the registration errors to register the laser range finder and camera.B) The environmental information which obtained through the camera and laser range finder may has some measure errors which occurs inevitably in the process. In this paper, line is selected as the main scene feature. As for the traditional HOUGH transform, the interference of noises and the high-cost compute cannot be ignored. This paper presents a improved method to improve th compute speed of hough transform on the bases of the two-point-vote algorithm.C) An automatic data fusion system for laser range finder and camera has been exploited. This system can get the comparative position and pose between the laser range finder and camera on the condition that exists the relative movement between the camera and the laser range finder, then register their images.