The Camera Calibration And Multi-scale Corner Detection Technology Based On 3D Reconstruction Technique

3D reconstruction technique is an important research direction of digital image processing and machine vision field, with the rapid development of information technology and related needs, the definition of 3D reconstruction is continuous improvement and innovation. Based on 2D image information extraction and feature points matching, and combined with hierarchical map and geometric imaging technique can depict the real scene information, which provides a strong means of scientific research.On the basis of 3D reconstruction technique of domestic and foreign advanced technology research, focus on camera calibration techniques and corner detection techniques, some creative algorithims and methods have been proposed, and the highlighted ideas deseribed as follows:First, a self-calibration method based on the traditional calibration was proposed. The traditional camera calibration method is complex, and requires manual interaction, an improved camera calibration method was proposed. The improved calibration algorithm combines the high accuracy of traditional calibration methods and the speed of self calibration technique, it combines edge detection techniques and camera calibration techniques, to achieve automation of camera calibration.Second, an improved corner detection algorithm based on multi-scale space was proposed. Using the Harris corner detection method process coarse corner detection under multi-scale space, and then detected corner point information mapping to the original image, at last to sub-pixel detection. Improved corner detection algorithm can have both speed and accuracy.