基于视频处理的游梁支点位置度在线检测

Real-time measurement of pivot position based on video processing

抽油机游梁支点的位置度是衡量抽油机设备安全、经济运行的重要指标。为克服传统位置度人工测试中动态误差大、效率低的缺点,本文提出一种基于视频图像的位置度测试方法,从标志、边线识别、标尺、目标跟踪、精度和速度等方面,详细讨论了测试方法。基于矩形标志物对边平行的特点,通过一维灰度矩算法和曲线拟合算法,准确地测定了标志物的图像尺寸,提高了标尺精度;利用亚像素匹配算法,提高了十字标志的匹配定位精度;利用搜索策略和卡尔曼滤波预测十字标志的运动轨迹,提高了匹配定位的速度。实验表明,这些措施保证了检测过程的实时在线完成。

The pivot position on beam oil pump is a primary specification to assess the facility security and power dissipation. In order to overcome the shortcomings of significant dynamic error and low efficiency of traditional manual measurements, a technique based on video image is presented, and the test algorithms are discussed in detail about the mark, cross line, scale, target tracking, accuracy and processing speed. According to the parallelism of opposite lines, the width of mark is acquired accurately using 1 st gray-level moment and curve fitting. Sub-pixel matching is utilized to improve the tracking precision. Search strategy and iterative Kahnan algorithm are used to predict the trajectories of the mark. Experiments show that the accuracy and speed are satisfied in the pivot position online measurements.