Research Of Vision System For Moving Robot Based On Omni-directional Vision System

Robots are automatically designated work machine equipment. In accordance with the surrounding environment information,it may complete some specific tasks. This will people from danger or complex work freed. While mobile robot to perceive the external environment, the main tool is visual sensor. But the visual angle of traditional camera system have restriction, often does not meet the requirements. This paper introduces a kind of omni-directional vision device that uses camera and the mirror image of effective environment around a robot, which can provide information of the robot's 360 space.It found the bases of self-localization and navigation.This paper mainly studies the design of hardware and software development based on the omni-directional vision system of robot. The experimental results verified the effectiveness of the method.Hardware, this paper studies the principle of omni-directional vision system, introduces the influence of single viewpoint, compares the effects of four different shapes of the reflector.And determine the imaging features to make full use of the hypoid mirror mirror. On this basis, we design an omni-directional vision system,which made up of an isomeric horizontal mirror and an isomeric vertical mirror, which effectively limiting omni-directional vision images discrepancy.Software, first introduced the omni-directional vision system processing flow, and research the camera calibration, image collection, etc. And based on the omni-directional vision system images, the paper discuss the image smooth and image sharpen of the image preprocessing procedure.Secondly, the article studied the basic method of color image processing,introduces the composition color spaces,compares the diffenence between several color spaces.The paper also improves the conversion formulas from RGB to YUV color space,studies the basic method of image segmentation.Then the paper uses the method of multi-threshold image segmentation of YUV color to segment the color image. After that, in order to calculate the regional characteristics of feature point coordinates, the artical recognise image regional characteristics, then,removes interference using binary morphological method.Finally, studies the method of self-localization based on the detection of dots and lines. Bring up a modified Hough transform,which proposed to reduce the Hough transform algorithm. The paper compares the method of self-localization based on the detection of dots and lines, determines the triangulation to implement robot self-localization.At the end of the article,the application and development of future direction is discussed.