基于Harris-SIFT算法的双目立体视觉定位

Object Location Technique for Binocular Stereo Vision Based on Harris-SIFT Algorithm

针对基于尺度空间对图象保持不变性的SIFT算法在双目立体视觉应用时实时性差、误匹配等问题,提出一种运用Harris-SIFT算法进行双目立体视觉定位方法。通过介绍双目立体视觉的模型原理,利用Harris-SIFT算法从左右摄像机分别获取的图像中检测目标,并获取匹配目标的特征点,对两幅图像中目标物体的坐标标定,通过计算可得到目标物体的深度距离,还原其三维信息。实验证明,运用Harris-SIFT算法使该系统的实时性能和距离精度得到提高。

The object location technique for binocular stereo vision has some problems of the poor real-time performance and the mismatching based on the SIFT algorithm of the scale space to image keep invariance. A method of the object location technique for binocular stereo vision developed based on the Harris-SIFT algorithm is presented. Through the introduction of the binocular stereo vision model principle, adopting new Harris-SIFT algorithm to get the camera image form the two detection target,and for matching the feature points of the target,according to the target object coordinate calibration two images of the two images. Through the calculation can get the depth of the object distance ,reducing the 3d information. The experiments improve the real-time performance and the accuracy distance based on the Harris-SIFT algorithm