基于图像深度信息集的Hough圆检测方法

Hough Circles Detection Based on Depth Information Set of Image

针对利用视觉引导的工业机器人进行轴孔装配,提出了一种基于图像深度信息集的工件圆孔检测方法,对三轴磁流体密封装置上的轴承孔进行了边缘特征提取和检测实验。首先利用3D相机采集对齐的三轴磁流体密封装置的彩色和深度数据流样本,根据待检测轴承孔所在平面的深度信息提取出该轴承孔在深度图像中的所有像素坐标。然后对彩色图像矩阵进行相应的数据处理,再经霍夫圆变换即可检测出该轴承孔的相关参数信息。实验结果表明,基于图像深度信息集的霍夫圆检测克服了传统霍夫变换无法单独识别特定的圆以及复杂背景下的错检和漏检等缺点,对复杂情况下的工件圆孔识别和检测具有一定的指导意义。

In view of using visual guiding industrial robot for shaft-hole assembly,this paper proposes a method of circular hole detection based on depth information set of image.And this method is successfully carried out on bearing hole detection of the three-axis magnetic fluid sealer.First,the aligned samples of color and depth data stream are collected via a 3 D camera and all pixel＇s coordinates of the bearing hole in depth image are extracted.Then corresponding data processing are implemented on color image matrices and position information of the bearing hole can be detected by Hough transform eventually.The result turns out that the Hough circles detection based on depth image can be used to improve the weak points of traditional Hough transform which can＇t individually identify the specific circle and often occurs false or missing detection in complex background.This approach can provide some helpful instructions for circular hole detection in complex cases.