基于机器视觉的轨道车辆零部件形位尺寸检测方法研究

在高铁车辆中,包含着种类繁多的零部件,每个零部件对于车辆的正常稳定运行都具有及其重要的影响,传统的检验用时较长且容易出错,检验效率较低。出错的车内装饰部件会造成不同程度的返工,大大的降低了生产效率。目前,高铁零部件形位尺寸检测研究领域尚很少视觉检测相关的研究,针对高铁零部件的大型不固定的特点,提出了完整的视觉检测高铁零部件形位尺寸的方法,首先从Snake动态轮廓模型分割零部件感兴趣区域,之后通过透视投影变换进行空间校正,采用Sauvola局部二值化进行轮廓提取,进而进行零部件视觉形位尺寸测量四个步骤实现高铁零部件的视觉检测,并对高铁上的三种零部件进行测试和验证,实验结果表明,基于视觉的高铁零部件形位尺寸检测具有较好的检测效率和可拓展性,可用于高铁多种零部件的形位尺寸入库检验中。

Research on Shape Position Detection Method of High-speed Rail Parts Based on Visual Inspection

In high-speed rail vehicles, there are many kinds of parts, each part has an important impact on the normal and stable operation of vehicles. The traditional inspection takes a long time and is prone to errors, and the inspection efficiency is low. The wrong interior decoration parts will cause different degree of rework, which greatly reduces the production efficiency. At present, there are few researches related to visual inspection in the field of shape and location dimension detection of high-speed railway parts. In view of the large-scale and unfixed characteristics of high-speed railway parts, a complete visual inspection method for shape and position dimensions of high-speed railway parts is proposed. Firstly, the interested area of parts is segmented from snake dynamic contour model, then perspective projection transformation is used to solve the problem that parts cannot be fixed, then use Sauvola local binary method to segment features, at last, extract contours and measure the shape and size of parts .Three kinds of parts on high-speed rail are tested and verified. The experimental results show that the visual based shape and position dimension detection of high-speed rail parts has better detection efficiency and scalability, which can be used in the shape and position dimension warehousing inspection of various parts of high-speed rail.