Kinect Based3D Color Reconstruction

With the development of the theory of computer vision and computer hardware technologyand the innovation of3D printing technology in recent years which has brought about the newcamera technology for imaging, which once made3D reconstruction become a hot topic in currentresearch, more and more experts and scholars have begun to study the depth camera. The depthcamera is made with infrared image technique. It can obtain the color video stream and can timelyoutput the distance between the objects in physical space and the camera as well, which is alsocalled the depth information. Moreover, in terms of cost, the depth camera is in low price and insmall size that make it portable and easy to be operated. Those characteristics are what thetraditional equipment the3D reconstruction does not have.At present, the study on the depth camera are about3D reconstruction, object recognition,intelligent navigation, human-computer interaction, image processing and augmented reality. Inthis dissertation, the study of three-dimensional color human body surface reconstructiontechnique is discussed to rapidly reconstruct a3Dmodelwith color and texture information byusing a depth camera scanning of the human body. A depth imaging device Kinect from Microsoftis used in this paper, which is fast, convenient and low cost and has been a focus in recent years, inareas of3D reconstruction.With the Kinect handheld scanner integrated with a color camera and infrared sensors, a threedimensional image of the static human body is obtained by scanning in different angle (includingthe human body surface depth information and color texture information). Compared with theexisting human body scanning system, this system has the advantages of light weight, portable,easy operation and other characteristics. The three-dimensional object or surface of the humanbody can be easily and quickly scanned, with no installation or calibration before the job.Moreover, it moves freely, without limitations of scanning direction and narrow space, whichbreak through the difficult of traditional body scanning equipment acquiring complex installation,time-consuming and laborious, and moving. Any special skills are acquired to use this device.In this paper, the innovation and the main contributions are below: 1.A complete set of3D human body surface color reconstruction system based on Kinect isproposed. In this system, the fast and accurate3D body scanning method is used, breaking throughthe difficult of traditional body scanning equipment acquiring complex installation,time-consuming and laborious, and moving, and making it possible for large quantities of accurate3D human body scanning.2. The problem of three-dimensional color information distortion calibration has been solved.Because the superposition from color information of two-dimensional to three-dimensional depthinformation will make white noise appears in a single frame of3D point cloud edges, the colorinformation distortion would be solved by superimposing RGB information of interest in a verysmall range which is extracted from the edges of objects in RGB image, the threshold ofextracting edges should be amplified.3. The accuracy of reconstruction for3D human body model is high. The experiments haveverified its accuracy of the results. Due to the human scanning process, some part of the bodymight not be scanned, such as head, arm. So the spatial information of these areas are unable to beobtained, leading to some cavities. At this moment, some patching technique is acquired foroptimization, so as to improve the accuracy of the model.4. Some operations, zoom, rotation, contrast setting for examples, is provided for the3Dhuman body model, in order to make it easy for the multi-viewport, multi-level and multi-aspectobservation and analysis, and to make up for the lack of imaging equipment on imaging, providingreliable data for further clinical diagnosis and treatment.