Research On Asodvs Based On Color Structured Light

The quality of the video signal source has always been the difficult problem in the field of computer vision. To overcome this problem, an Active Stereo Omni-directional Vision Sensor (ASODVS) has been developed, that can obtain an omni-directional depth map through the use of a Panoramic Color Structured Light Generator (PCSLG) and an Omni-directional Vision Sensor(ODVS). However, the ASODVS is a kind of new technology which is unripe in the navigation of the mobile robot. There are still some key technologies to be studied, especially on robustness and timeliness. Therefore, this project has become a very meaningful and challenging task.In order to solve the problem that how to use ASODVS to achieve robot navigation, this paper has studied several key technologies about this system which based on summarizing previous work. The main work can be summarized as follow aspects.1. Study on the technology of PCSLG. According to the comprehensively advantage of ODVS with a single view point, this paper proposed a PCSLG technology which with a single emission point and large transmitting range to match with it. It can make every pixel which on the two dimension video panoramic image to take the depth information in measured scene. And improving the quality of signal source in image space solves the fast matching problem in stereo vision. It provides a kind of new method to achieve the panoramic stereo vision perceive. Study on the principle and calibration of the PCSLG, create corresponding relation between launch angle and launch color.2. Study on the technology of ASODVS. Integration PCSLG and ODVS to construct a kind of ASODVS. It makes corresponding pixel point which on the panoramic image plane has the scenery object’s depth data, orientation and color information. Also, this system can achieve omni-directional perception, expression and3D reconstruction with only one panoramic image immediately. Set up a scientific and unified of3D vision coordinate system. It can be founded on rapid calculations for simplified model, space object point match, measurement and reconstitution etc., and conducive to modeling math geometry. 3. Study on the scientific visualization technology of panoramic depth information. Study on the expression way of panoramic depth information, demonstrate the correctly of the depth data directly. According to the technology of pseudo-color, using the relationship between the depth information which is calculated from the ASODVS system and the value of chromaticity on color code, estimate chromaticity based on the depth information itself, and obtain a pseudo-color image. The color of every pixel on image is not present the color in real scene, but stand for the depth information of every point on real scene. This kind of image is called pseudo-color image.3D points cloud reconstruction can be utilized for distinguishing contour information of measured scene clearly. Experiment results have been shown that the proposed method can accurately express the depth information vividly and directly, and also be appropriate for human observation.The contribution of this work is the achievement of an ASODVS technology which can get the image with depth and color information corresponding to the actual object. The main innovations are as follows:1. Design a kind of PCSLG with a single emission point and the body-structure type. All of the3D spatial objects points projected by the panoramic color structured light have the space geometry property which has their own color.2. Design a kind of ASODVS which is composed of a PCSLG with a single emission point and an ODVS with a single view point. According to analyzing the position and color properties of every pixel on ODVS imaging plane, and obtaining the structure parameters of ASODVS, it makes any pixel points on panoramic image has the scenery object’s depth, orientation and color information, and implements omni-directional perception, expression and dense reconstruction with only single panoramic image immediately.