Improved Direct Linear Calibration Based on a New Corner Detection Algorithm

Abstract

This paper highlighted the technology of gray transformation to target the image captured using the camera, which converted a colored image into grayscale, filtered out the noise using the median, and extracted the image edge by using Canny operator. Because the extracting edges were not clear, it could not obtain the target pixel point accurately. For improving the accuracy and real-time feature extraction, corner feature points detection algorithm via randomized Hough transform based on spatial moment was proposed. To obtain accurate coordinates of the corners, the image was processed using matlab2011 software in the experiment. In order to achieve camera calibration rapidly and effectively, the linear calibration algorithm was improved. Using the least square method, the calculation process was simplified, and the calibration error was reduced. The experimental results show that the proposed algorithm was simple and effective, which improved the calibration accuracy, and verified its validity and feasibility.