基于三次样条插值的指针式仪表识别方法研究

Research on pointer instrument recognition method based on cubic spline interpolation

针对变电站巡检机器人指针式仪表自动识别中易受指针阴影影响的问题,提出了一种基于三次样条插值的指针式仪表图像分割方法。提取包含指针的圆形区域作为信息图,对信息图进行中值滤波、灰度拉伸;使用16个固定的阈值对信息图进行分割,并统计指针的长度和面积;使用三次样条插值法对面积序列插值并计算其一阶差分,以指针长度大于信息图半径的最小阈值作为起点,求得一阶差分的第一个波谷点,以该点对应的灰度值作为分割阈值对信息图进行图像分割。图像分割实验表明,文中算法对含有指针阴影的仪表图片分割成功率由10%提高到94%,识别成功率由82%提高到98%,提取的指针中仅包含少量的无效信息,减少了后续算法的运算浪费;鲁棒性实验表明,该算法对各种环境干扰具有较好的鲁棒性。

Aiming at the problem that the pointer recognition instrument of the substation inspection robot is susceptible to pointer shadow,a pointer-based instrument image segmentation method based on cubic spline interpolation is proposed in this paper. Firstly,the circular area containing the pointer is extracted as the information map,the median filtering and grayscale stretching are performed on the information graph;secondly,the information map is segmented by using 16 fixed thresholds,and the length and area of the pointer are counted;finally,the strip interpolation method interpolates the area sequence and calculates the first-order difference. The first threshold of the first-order difference is obtained by taking the minimum threshold of the pointer length larger than the radius of the information as the starting point,and the gray value corresponding to the point is used as the segmentation threshold to conduct image segmentation of the infographic. The image segmentation experiment shows that the segmentation success rate of the instrument image with pointer shadow is increased from 10% to 94%,and the positioning success rate is increased from 82% to 98%. The extracted pointer contains only a small amount of invalid information,which reduces the wasteful operation of the follow-up algorithm. The robustness experiments show that the algorithm is robust to various environmental disturbances.