基于单目视觉的轴类零件3D重构技术

The 3D reconstruction technology for shaft parts based on monocular vision

在结构化环境下,提取轴类零件三维信息是实现机械手自动上下料的重要环节。首先采集放置于特定工作台上的轴类零件图像,其次利用超红特征和Otsu法获取超红特征灰度图并进行动态阈值分割,提取标准圆的像素半径,进而获取每毫米所代表的像素值;利用拟合直线方程可获取摄像机与工作台的距离,同时利用亮度特征,采用相同的方法获取轴类零件的像素尺寸,然后利用每毫米所代表的像素值计算轴类零件的实际尺寸,完成轴类零件的三维重构。该技术降低了软硬件的复杂性,可实现机械手智能抓取轴类零件,为数控机床无人化操作的可行性发展提供了理论基础。

In a structured environment,it is an important link for feeding and blanking to extract the 3D information of shaft parts intelligently. It collects the images for shaft parts placed in the specific work station,and gains the gray images using excess red feature,processes the binary images in Otsu＇s threshold method. And then it extracts the pixel radius of the standard circle,obtains the pixel per mm,defines the distance between camera and the specific work station with fitted linear equation. At the same time,it finds the pixel size of shaft parts,calculates the actual size in the pixel per mm,and completes the 3D reconstruction of shaft parts. This technique reduces the complexity of hardware and software,grasps shaft parts intelligently and provides the necessary information for the unmanned operation of CNC machine tools.