双目立体视觉中角点检测的方法研究

Research on Corner Detection Technique for Binocular Stereo Vision

双目立体视觉是计算机视觉研究中最为活跃的一个分支,是智能机器人科学发展的重要标志。它是由不同位置的2台或者1台摄像机(CCD)经过移动或旋转拍摄同一场景,通过图像获取、摄像机标定、特征提取和匹配,进而通过计算空间点在2幅图像中的视差,获得场景的深度信息。本文在SU-SAN角点检测的基础上,提出了基于图像平均灰度的阈值选取方法,使阈值的选取不再盲目,保证了用SUSAN算法进行角点检测的实用性和适用性。仿真试验表明,用该方法进行特征点的提取,减少了匹配时间,提高了匹配精度,得到了较好的效果。

Binocular stereo vision is one of the most active branches of researching fields of computer vision, it is also the important symbol of the aptitude robots development. It can calculate the disparity in the two images and obtain the scene depth information based on the different locations, vision system composed of two or one CCD cameral or personal comput- er＇s moving or rotating the same scene shot, through image acquisition, camera calibration, feature extraction and matc- hing, space points. The paper presented selection way based on the average gray scale image on the condition of SUSAN al- gorithm to corner detection, in order to make the threshold^s selection not aimless any more, and also guarantee the practi-cality and applicability when using SUSAN algorithm to corner detection. A large number of experiments＇ results indicated that it can reduce the time of extraction of feature points and matching, improve the matching accuracy, and obtain good re-sults.