基于MatLAB的轨道交通轴类零件图像裂纹识别设计
滚珠裂纹是热处理中滚珠零件常常会产生的裂纹，是一种不可修复的工艺缺陷。如果裂纹明显到一定程度，则会使已经加工好的滚珠零件彻底报废，不仅造成金属材料和加工费用的损失，而且这是个一直困扰热处理工作者的理论和工艺的重大疑难问题。因此，对零件表面的图像信息进行考察是判断其是否出现裂纹的一个重要途径，本文以此为出发点，对表面裂纹的图像处理识别方法将进行了深入的研究，针对表面微裂纹信息极易湮没于背景图像而直接用图像分割的方法难于将裂纹信息从背景中分离的特点，提出一种基于边缘检测的裂纹提取方法。这就解决了裂纹检测的问题，使得能够方便准确的测得滚珠上的裂纹信息。20804
本文首先介绍了裂纹图像的采集处理系统，然后对裂纹图像的处理、算法进行了深入的分析，并采用MATLAB 混编的方法实现了算法，编制了识别软件，最后通过实例证明了该裂纹识别方法的有效性，并指出其不足。
毕业论文关键词： 滚珠；表面裂纹识别；裂纹提取；图像处理；边缘检测
Rail Transit based on the Matlab Image Recognition of Crack Shaft Parts design
Abstract：Ball bearing s an important component parts of the machine, from support to withstand、laden the torque effect. On the rail transportation，Machine intact is related to human life and health, resulted from harsh environment which has been one of the main sources of The machine malfunction，once have crack to rise, then accelerate to expand and cause breakdown occurrence. Therefore, aperiodical check is necessary to discover tiny crack in time, to further judge whether it can continue in service or not.
Ball Ball cracks are cracks in the heat-treated parts are often produced, Is a kind of irreversible process defects, If crack clearly to a certain degree, can make good ball parts that have already been processed completely scrapped, not only resulting in loss of metal materials and processing costs, and it is a theory and technology of the heat treatment of workers had significant problems. It is an important path to investigate into the image information of parts surface for judging whether crack or not. the image processing and recognition method of surface crack is thoroughly researched in this thesis. The problem is that surface tiny crack information easily lost in background image. Moreover, it is difficult to separate crack information from the background directly using image segmentation method. So a crack extraction algorithm proposed which is based on image subtraction, Moreover，it achieve the purpose for crack extracting through image subtraction and then image segmentation, morphology processing etc are adopted to carry through crack quantification processing.
This thesis firstly introduce the image acquisition and processing system, then thoroughly analyze the registration and quantification algorithm of crack image and adopt mixed programming with MATLAB to realize algorithm and image recognition. The proposed strategy is ultimately verified by experimental results.
KeyWords： ball; surface crack identification; crack extraction; Image processing；edge detection
目录
1 绪论 1
1.1 课题研究的背景和意义 1
1.2 滚珠的介绍 1
1.3 图像识别技术在裂纹检测中的应用现状 2
1.4 裂纹图像识别技术研究现状及其发展 2