Abstract: A cable location algorithm based on texture recognition is proposed to solve the high-voltage de-icing robot obstacle problem when arm grasps line. According to the periodic characteristics of texture along the direction of the cable within the complex background, the high light reflection and shadow effects are reduced using homomorphic filter. The direction of the texture is calculated by direction field estimation and texture is extracted by Gabor transform. A linear least squares fitting method is used to obtain the final position of the cable within the region of interest. Obstacle tests in the context of different light show the effectiveness of the algorithm.

Key words: deicing-robot, visual positioning, orientation field estimation, Gabor transform

摘要： 针对高压线除冰机器人越障时机械臂对中问题，提出了一种基于电缆线纹理特征的定位算法。根据复杂背景下沿电缆线方向纹理具有周期性的特征，通过对原始图像做同态滤波，减少高光反射和阴影的影响。用方向场估计纹理的方向，利用Gabor变换进行纹理信息提取，以纹理图像作为感兴趣区域，在感兴趣区域内使用线性最小二乘拟合方法获取电缆线最终位置。通过在不同光线背景下对电缆线定位的对中实验，证明了算法的有效性。

关键词: 除冰机器人, 视觉定位, 方向场估计, Gabor变换