摘要：轴类零件是工业生产机器中最重要的组成部分，其检测工作被各个生产企业所重视。检测工作主要包括内部探伤和表面探伤。内部探伤已经有较好的方法来检测，如超声波探伤、X光射线探伤等方法。但是表面探伤仍是轴类零件探伤工作中的一大难题。传统的人眼检测法容易受人眼分辨能力和疲劳等主观因素的影响，存在着诸如劳动强度大、工作效率低、工作环境恶劣尤其是遗漏现象严重等问题。因此，研究和开发成熟的轴类零件裂纹识别系统对企业来说是个急于解决的任务。22760
为了降低人工检测的强度和风险，提高轴类零件探伤的效率，本文采用图像处理技术和理论对裂纹的检测方法展开研究 。
由于轴类零件大部分是在工业机械中使用，就免不了会有油污、粘结物、水渍等，再加上采集图像时容易受到光照条件的影响，采集到的图像往往会有除了裂纹外的干扰物。这些干扰物与裂纹有一定的相似性，对后期图像处理带来干扰，影响裂纹检测的准确性。
针对该问题，本文分析总结了裂纹图像的基本特点，提出并实现了一种提取裂纹的方法，实现对原图像的预处理，包括图像平滑处理、灰度化、滤波去噪等，以及后期的图像分析，包括图像边缘检测、裂纹提取等。这一方法的实现都是基于MATLAB平台开发的图像处理软件。这一方法有以下几个特点：
(1) 采用1200万像素照相拍摄，保证图像的高分辨率，确保了后期检测的精度。
(2) 利用了图像的边缘和区域信息，很好的完成了图像分割，从而保证检测的准确度。
(3) 所有算法都是软件实现的，便于系统更新，从而提高了该方法的灵活性。
毕业论文关键词： 数字图像处理；轴类零件；裂纹检测；裂纹提取；滤波去噪
Recognition of Shaft Parts Crack Image Based on MATLAB
Abstract：Shaft parts are the most important part of the industrial machine，The detection work has been paid great attention in each enterprises. Detection work include internal detection and surface detection. internal detection has had  good method to detect，such as ultrasonic testing, X-ray flaw detection and so on. But the surface detection is still a big problem in the shaft parts detection work. The traditional human eyes detection is easily influenced by subjective factors such as human eyes resolving power and fatigue. There is such as labor intensive, inefficient and severe working condition, especially the omission phenomenon and so on. Therefore, the research and development of mature shaft parts crack identification system for enterprise is eager to solve.
In order to reduce the strength and risk of the artificial detection, improve the efficiency of shaft parts inspection. This article uses the technology and theory of image processing to study the method of crack detection.
Since most of shaft parts is used in industrial machinery, so it is inevitable there will be grease, dung, water damage, etc. And it is easily affected by light condition and distractions to collect images. These distractions are similar to crack, so it will interference image processing and affect the accuracy of the crack detection.
Aimed at the problem, article analyze and summarize the basic characteristics of crack image and proposed and implemented a method to extract the crack and preprocess images. Which include image smoothing and gray, filtering noise, etc. Image post processing include the image edge detection and extraction of crack, etc. The implementation of this method is based on the development of MATLAB image processing software platform. This method has the following characteristics:
(1) 12 million megapixel camera are used to ensure a high resolution image, which leads to high accuracy of inspection.
(2) The edge and region information are used together for have image segmentation，which leads to high accuracy of inspection. 基于MATLAB的轴类零件图像裂纹识别设计:http://www.youerw.com/zidonghua/lunwen_15528.html