基于机器视觉的浇口杯定位研究

Study on Localization of Casting Pouring Cup Based on Machine Vision

针对铸造浇注过程中浇口杯位置无序、数量不定以及高度不同造成的浇口杯定位难等问题,本文提出了一种基于OpenCV视觉技术的铸造用浇口杯定位方法,着重阐述了机器视觉计算方法的设计步骤。首先,根据目标的形状、尺寸等特征获得感兴趣区域(ROI);经过图像预处理后针对传统方法得到的特征点中伪特征点较多的问题,引入自适应高斯滤波平滑二值图像;之后利用Canny算子来检测浇口杯边缘,并引入了3×3的Sobel算子以获取更多的方向信息以及降低Canny算法的敏感性;通过对图像拟合、提取、比较分析等操作,获取浇口杯在砂箱的相对位置。最后以消失模铸造砂箱浇口杯为例,实验结果表明,提出的方法精度高、实时性好,能够快速得到浇口杯的边界信息和位置坐标,为全自动浇注奠定基础。

Aiming at the difficult location of the pouring cup caused by disordered location, variable quantity and different height in the casting process, a method of locating the pouring cup for casting based on OpenCV vision technology was proposed in this paper, and the design steps of machine vision calculation method were emphatically described. Firstly, the region of interest(ROI) was obtained according to the shape, size and other characteristics of the target. After image preprocessing, the adaptive Gaussian filter was introduced to smooth the binary image in view of the problem that there were many pseudo feature points in the feature points obtained by traditional methods;then the Canny operator was used to detect the edge of the gate cup,and the 3 × 3 Sobel operator was introduced to obtain more directional information and reduce the canny computation cost. Through image fitting, extraction, comparative analysis and other operations, the relative position of the gate cup in the sandbox was obtained. Finally, an example was taken of the lost foam casting sand box pouring cup. The experimental results show that the proposed method has high accuracy and good real-time performance, and can quickly obtain the boundary information and position coordinates of the sprue cup, which lays the foundation for fully automatic pouring.