Three-Dimensional Photo Studio Research Based On Kinect

With the development of computer vision theory and technology,3D-Object reconstruction has become a research hotspot in recent years, more and more applications is found in various aspects, such as design and simulation, game and film production, industrial measurement, heritage conservation, medical research. With the continuous improvement of precision and technology in sampling device, it’s prospects will certainly become broader and broader. Microsoft launched the Microsoft Kinect device in recent years, now 3D-Object reconstruction research based on Kinect has achieved some results and applications. Some work about 3D-Object reconstruction based on Kinect was did in this paper, Kinect device can obtain color data and depth data at the same time, and provides methods of matching color and depth data. With the characteristics of high speed, high efficiency and low cost, it has great application value.First of all, the background and significance,research status,practical application of 3D-Object reconstruction were briefly introduced in this paper. Then the configuration, data structure and access to data of Kinect were introduced.Then the conversion of depth image to the point data and the display of depth and color data were analyzed.The complete three-dimensional data of the object needs to be gathered from multiple angle and integrated.The integration of multi-angle data can be registered to the same coordinate system by coordinate rotation and translation matrix. Kinect Fusion enables real-time three-dimensional reconstruction based on GPU parallel computing, it operate simplely. Kinect device itself does not have the positioning function, but its proposed point fusion algorithm based on voxel technology, during acquisition of each frame data, Kinect Fusion integrates new data with the original data by ICP iterative registration algorithm, obtains the spatial transformation matrix of the camera with respect to the initial position and integrates data. Kinect Fusion although worked quickly and easily, but the stability is not enough, it is easy to track failed, the tracking process is hard to control. We consider manually match point cloud data, and then using the point cloud filtering and smoothing method to improve the smoothness of the point cloud integration.Raw data acquired from Kinect contains background data and noise data, segmentation, filtering, smoothing of the point cloud dada are need, and then matched color information points can be displayed stereoscopically by OpenGL, this effect is more realistic, and carving file of the point cloud can be outputted. But the communication with more than one computer must be considered during the acquisition of multi-angle data, it could be achieved by communication socket. PCL is a cross-platform open source library of point cloud processing tools, which encapsulates many common point cloud processing algorithms and powerful.Finally, combining OpenGL and PCL open source library, the software system about three-dimensional point cloud acquisition and processing and carving file generation were designed with fine results based on Kinect and the Visual Studio 2010 platform, showing a three-dimensional camera project.