Study On Performance Upgrade Of Structured Light 3D Scan

As follow-up study of the Structured Light 3D Scanning System in our lab, improving the performance of the prototype system in speed and accuracy is the main work of this paper. These improvements will make the system to better meet the needs of high precision, high efficiency and suitable for large-scale application of artifacts digitalizing. Based on detailed analysis of the key points in scanning process, this paper improved the function of acquiring digital images and picking up the initiative clues, thus significantly improved the performance of prototype system.The input data of this prototype system are digital striped images. The scanning process is composed of three steps, analyzing the input images, extracting the initiative clues of them and triangulation based on the extracted results. Therefore improving the technology of image acquisition is the first step to upgrade the scanning performance.Among scanning process, this prototype system stores the Bayer data acquired by the camera. So we ensured the high quality of Bayer data acquisition through automatically adjusting the camera's exposure time. Based on analysis of different Bayer interpolation algorithms, we selected and improved the most appropriate algorithm applied to the system. Experiments show that the precision of the scanned model, the scanning speed and storage space occupation have been upgraded significantly.The quality of scanning can be improved further through redevelopment of another core technology, which is image initiative clues extraction. The linear interpolation algorithm used currently only applies to ideal striped images, because of noise pixels in actual acquisition, this method can not lead to a satisfactory outcome. We first analyzed the current sub-pixel edge detection algorithms, such as polynomial interpolation, gray moment, and space moment and so on. Then by combining the features of each algorithm and the striped images, we had the most appropriate algorithm improved and used in the prototype system. At the same time image based noise points filtering and ROI region selection are integrated with the system, all together greatly developed the current technology of image analysis.As the result, all these improvements can significantly improve the quality of the scanned model and speed up the scanning process.