RTG视觉纠偏的局部引导图像滤波算法

Automatic Steering for Rubber Tyred Gantry Based on Local Guided Image Filtering

针对轮胎式集装箱门式起重机(RTG)在基于视觉的行走纠偏过程中环境因素干扰导致的定位失败问题,提出了基于局部引导图像滤波和均值漂移聚类的RTG纠偏视觉辅助算法。该算法首先对图像进行基于局部引导图像滤波的高效预处理,在降噪的同时保留图像边缘特征;然后基于预先标定的引导线宽度和间距等特征,获取自适应分割阈值及简化均值漂移聚类算法的参数,从而实现引导线边缘的筛选提取和偏移量的计算。现场试验表明,该算法能够克服夜间环境下偏色光谱对识别的影响,有效提高纠偏的实时性和精度。

In order to solve the problem of locating the rubber tyred gantryprocess,an algorithm based on local guided image filtering and mean shift clustering algorithm was introduced to extract deviation information. The guided image filter on specific area was used in the algorithm process to preserve the edges while reducing the computation time of noise filtering. Consequently, features including guideline width and spacing were used to determine the adaptivethreshold and the parameters in simplified mean shift clustering algorithm. There by offset calculation and extraction of the guide line edges were achieved. The experiments indicate that the algorithm can provide accurate position information even in poor lighting conditions, which improves the real-time performance and accuracy of steering effectively.