Research On3D Active Stereo Omni-Directional Vision Sensing Technology

The quality of the video signal source has always been the bottleneck of computervision in acquiring video image information. The ideal signal source for computervision device is the visual sensor that can get the image with scenery depth and colorcorresponding to the actual object. The technology of active stereo omni-directionalvision sensing provides a new way to solve the problem, but studies in domestic orforeign are not efficiently, many key technologies need to be studied. Therefore, thissubject is significant and challenging.The key to accurately identify the illuminant color in achieving active stereoomni-directional vision sensing technology is to obtain the corresponding relations ofthe colors and the depth. This paper does the research on the technology of theilluminant color identification in active stereo omni-directional vision sensing basedon the previous achievement. Our work includes several aspects:1. The research on the active stereo panoramic vision sensor. This paper presentsthe principle and calibration of the omni-directional vision sensor with a single viewpoint and a panoramic color structured light generator with a single emission point. Inthis paper, active stereo vision has become the visual perception of the body structurefrom the visual perception of surface structure through the vertical overlappingcombination of the axis lines of the both.2. The research on active vision system model. This paper analyzes the threefactors in influencing surface color (illuminant color, properties of surface reflection,the spectral sensitivity of the observer) and their relationships. On this basis, we studythe structure of active vision system and group it into three models: projector model,camera model, surface illuminantion model. And then we build the active visionmodel, describe the mathematical expression of the relationships of the projectionilluminant color with the corresponding image point color. 3. The research on illuminant color identification. This paper studies therelationships of hue, wavelength and brightness. We also analyze color representationof active vision system. From the view of chromaticity, this paper proposes anestimation algorithm of illuminant chromaticity based on CIE chromaticity diagram.This algorithm gets chromaticity coordinates of natural color and mixed color ofobjects in CIE chromaticity diagram by using the images of the same scene with andwithout light source and finding chromaticity coordinates of light source inconnection of their extension cords. Experiments show that the algorithm only suitsthe scene of monochromatic light source and various surface colors. From the view ofoptical, a color correction algorithm of light source based on dichromatic reflectionmodel has been proposed here. Firstly, the algorithm eliminates the couplingdifferencing between digital camera and projector through calibrating the chromaticityof them. Secondly, in the manner of estimating the spectral reflectivity of objectthree-color-channel and separating the diffuse reflection and the spectral reflectioncomponents, the algorithm revises the impact on the color of projection light sourcecaused by surface color and spectral reflection. Experiments show that there are noconstraints of light sources and objects material in using the algorithm and itsignificantly improves the accuracy of identification of the projected light color.