Research Of Real-time Indoor Scene Reconstruction System Based On Optical Coding

price, slow speed and poor quality. The bilateral filtering algorithm is utilized to process the three-dimensional image generated by Kinect. The corresponding software is developed to display the image. The major contents of this thesis are as follows:Firstly, the three-dimensional reconstruction techniques are compared from several aspects. The research achievements at home and abroad of three-dimensional reconstruction techniques are summarized. The advantages of the Kinect sensor in three-dimensional modeling are discussed. The procedure of generating the depth image with the Kinect is presented. The initial data content is briefly analyzed for the following data processing.Secondly, the camera calibration is carried out for the Kinect sensor to reduce the hardware influence on the image quality. In view of the wrong depth information generated by the Kinect sensor discussed above, the results generated by different algorithms are compared. A suitable algorithm is selected to ensure that the processed image is available for image fusion.Thirdly, the processed images are fused with matching algorithm which is particularly developed for the depth image. The detail problems in the fusing procedure are analyzed. The algorithm is improved to solve the problem that feature point search causes a large amount of calculation, thus reduces the time of image matching, and meets the requirements of real-time display.Finally, the three-dimensional model is displayed through the PC software. The overall performance is evaluated through the analysis of the result constructed by the Kinect sensor in different ways. With the rapid development of science and technology, the reconstruction of the three-dimensional model becomes important in daily life, which is widely adopted in many fields. After the invention of Microsoft’s Kinect sensor, the three-dimensional system exhibits a huge price advantage on hardware equipment. In this paper, the three-dimensional reconstruction is implemented with the Kinect sensor, in order to overcome the disadvantages of the traditional three-dimensional modeling such as the high...