Reserch On Fast Color3D Imaging System And Crosstalk Compensation Between Color Channels

Color fringe projection systems for3D shape measurement have been widely studied inacademia and business field because of the advantages of non-contact operation, full-field, andfast data processing. A color CCD camera and a DLP (Digital Light Processing) projector aremostly used. However, crosstalk between color channels of the camera and projector changes thesinusoidal shape of the obtained fringe patterns and then reduces the measurement accuracy.Several methods have been proposed to solve this problem, but they all have some disadvantages.This paper presents a simple method to reduce crosstalk between red, green and blue channels.Crosstalk between color channels can be seen as signal aliasing, so different wavelength in colorchannels can be used to keep them apart. Simulated and experimental results show that theproposed method can significantly reduce the influence of crosstalk on phase calculation andimprove the measurement accuracy of the color fringe projection system.Simultaneously measuring3D shape and color texture of moving objects in real time isbecoming more and more important in many application fields. The existing methods ofmeasuring moving objects have some disadvantages, such as a complicated procedure to buildthe coordinate transformation between cameras, inaccurate3D shape data because of morecapturing time of fringe pattern images, and sensitivity to environmental light due to theprojected visible fringe patterns. This paper presents a novel method to simultaneously capture3D shape and color texture information of moving objects by combining infrared (IR) channeland visible channel together. Shape data are calculated from IR fringe patterns to avoid the effectof environmental light on the measured objects. Shape data and color texture information aresimultaneously captured by IR and visible channels of a2-CCD camera; therefore they havepixel-to-pixel correspondence, without needing any registration. Experimental results on twomoving objects show that the proposed method can simultaneously obtain3D shape and colortexture information with a reliable way.