身管内膛表面几何特征参数的准确量化一直是身管疵病检测和寿命预测的一大难点, 而身管膛线的高度差是其中最重要的几何参数之一。本文依托结构光三维检测手段, 采用激光三角法对身管膛线高度差进行定量检测。首先将特定结构光栅投影到模拟身管内膛的标定圆筒内壁上, 并采集经圆筒内表面散射后的变形结构光图像; 然后利用图像边缘分割算法中的多个算子分别对标定圆筒内壁的结构光图像进行分割, 同时引入灰度共生矩阵概念客观评价出最优的分割算子, 对图像中结构光条的边缘分割进行优化; 之后通过图像转换比例反推算法计算得到内壁图像高度差和实际凹槽高度差之间的转换关系, 最终应用于身管膛线高度差的测量过程中。试验结果表明: 绝对偏差控制在0.04 mm内, 满足系统的精度要求, 同时该方法检测手段方便、快捷。本文方法为精确量化身管内膛疵病的几何参数奠定了坚实基础。

In view of correct quantization of geometrical characteristic parameters on barrel's inner surface has always been one of difficult points for barrel defect detection and life forecast, and height difference of barrel rifling is one of the most important geometric parameters, this paper performed quantitative determination to height difference of barrel rifling by relying on the structured light three-dimensional detection means and adopting laser triangulation method. Firstly, certain structured raster was projected on inner wall of calibration cylinder of simulated barrel bore, and deformed structured light image scattering by internal surface of cylinder was collected. Secondly, structured light image on inner wall of calibration cylinder was segmented by utilizing numerous operators in image edge segmentation algorithm (IESA), the gray-scale co-occurrence matrix concept was introduced simultaneously to evaluate optimal segmentation operator objectively and edge segmentation of structured light in image was optimized. Finally, image conversion ratio backward inference algorithm (ICRBIA) was used to derive transformation relation between height difference of inner wall image and that of actual groove. Height difference measurement experiment on actual barrel rifling indicates that the absolute deviation is controlled within 0.04 mm, which meets accuracy requirement of measurement system, and at the same time, the detection means of the method are convenient and fast. It can be conclusion that the proposed algorithm may lay solid foundation for precise quantization of geometrical parameter of barrel bore defect.