仿人形机器人火炬传递中的高精度目标特征提取

Study on High Accurate Feature Extraction in the Process of Transferring Touch with Humanoid Robots

以仿人形机器人的火炬传递为背景,针对立体视觉系统的图像处理展开研究,提出了一种基于色标的高精度特征提取方法。利用火炬上的矩形色标,在RGB空间基于色彩进行图像分割,并利用K-L变换对边缘点进行分组,通过Hough变换和最小二乘法对色标边缘进行直线拟合,得到矩形色标顶点的高精度图像坐标,为提高火炬位姿测量的精度提供了基础。实验结果验证了本文方法的有效性。

The background of this work is to transfer the touch to each other with humanoid robots automatically. The image processing for binocular stereovision system is selected as the main research aspect, and a novel method for object segmentation and feature extraction is presented. Based on the mark attached to the touch, the object is identified in the RGB color space. Then after the edge points grouped by K-L transform, Hough transform and Lease Square Error （LSE） method are employed to fit the edge of the mark. The accurate image coordinates of the feature points of the mark are acquired through the cross points of these fitting lines. The experimental results show its effectiveness.