Title :
Research on Key Technique of Weed Locating Based on Binocular Vision System
Author :
Wei-xing, Zhu ; Li-bing, Xia ; Feng, Xiong
Author_Institution :
Lab. of Jiangsu Province, Jiangsu Univ. Zhenjiang, Zhenjiang, China
Abstract :
There are two aspects for locating precision of weed recognition system based on binocular vision. One is extraction of the feature points, and the other is distortion of camera lens used for image collecting. Some improvement has been made for these problems in this paper. Improving Harris operator makes the precision of feature point reach sub-pixel and rectifies image-distortion before calibrating the camera. Firstly, the one order radial distortion parameter is computed by the cross ratio invariability, and then collected image is rectified by the distortion parameter, finally, the camera is calibrated on the basis of the rectified image. The simulation results show that there is a light noise effect on the improved Harris operator, and the approach has a higher accuracy which can satisfy the need of locating precision of weed recognition system.
Keywords :
agriculture; calibration; cameras; computer vision; feature extraction; binocular vision system; camera calibration; feature extraction; image rectification; image-distortion; rectified image; weed recognition system; Approximation algorithms; Calibration; Cameras; Feature extraction; Lenses; Machine vision; Nonlinear distortion; Optical distortion; Polynomials; Spraying; binocular vision; camera calibration; image rectification; sub-pixel;
Conference_Titel :
Computer Science and Computational Technology, 2008. ISCSCT '08. International Symposium on
Conference_Location :
Shanghai
Print_ISBN :
978-1-4244-3746-7
DOI :
10.1109/ISCSCT.2008.221