The Vision Measurement Research Of Color Structured Light Coding

Structured light vision measurement is a branch of computer vision, which has beenapplied widely in reverse engineering, quality inspection of production, object recognition andso on, becoming a hot topic in recent years. Structured light measurement principle is theprojection of one or multiple coding mode to the measurement scene, a projection image ofthe scene is captured by using a video camera, and then through the matching of thecorresponding points on the projection mode and the decoding mode, the use of triangulationmethod to draw the depth information, so as to obtain three-dimensional information fields,the corresponding point match of the projection mode and capture images is an importantissue in structured light vision measurement. In order to obtain accurate matching relationship,both have a reasonable structured light coding scheme, as well as accurate decoding andpost-processing programs. The3D measurement system consisted of one projector and onecamera is established.Based on the principle of color code, the color information of De Bruijn sequence of fourhue stripes in two order is added into digital stripe pattern in which high and low intensitystripes vary in periods, forming space code measuring sense pattern of color grating, onlyproject pattern can realize the unique code in pixel dimension, satisfying the demand of3Dmeasurement.In this paper, decoding of captured color structured light stripe image is studied. Fullyconsidering the projection pattern, projection and capture environment, color distortionfactors, the decoding scheme is determined. The scheme consists of the extraction of high andlow intensity stripes split, center locating, K-means color clustering, stripe matching based onwindows unique. Calculate the3D point cloud coordinates combined with structured lightsystem calibration parameters.Simplify the RGB space to the Lab space conversion process, the use of L-channel filterdifferential results, split high and low intensity stripes. The innovation is high and lowintensity stripes split combined with K-means color clustering analysis: the captured stripescolor is not the same as the standard projecting color, since the influence of factors such as theambient light, the crosstalk between the colors and the absorption and reflection of objectssurfaces. However, each color in the RGB color space has a certain degree of clustering.Based on the color clustering, the classic K-means adaptive color clustering algorithm ispresented. According to the principles of the K-means clustering, each of iteration generates new origin and direction vector, which is the new center, the color does not change after a fewiterations, the completion color distinction of the center stripe.Traversing the region of interest strip center obtains decoded sequences. According to thewindow uniqueness of De Bruijn sequence, the position of each window in the projectionpattern is determined. The location of each match point is found. A coordinate matrix ofmatched points is generated. By using triangulation method, the depth information of3Dsurface is obtained. Reconstruct the three-dimensional surface of the object modelThe static3D vision measurement system has high accuracy and resolution, can realizecomplex three-dimensional surface measurement and reconstruction. The research expandsthe application range of the color structured light3D measurement technology.