燃气阀门密闭特性的检测

论文总字数：31343字

摘 要

本课题通过一个实验装置完成将多个燃气阀门浸入水中，配置相应的光源和图像传感器，对加压后的阀门连续成像，对连续获取的图像进行分析和判断，设计识别算法，完成对燃气阀门漏气的自动检测。最终通过实验对已经获取的样图的算法进行验证。

本文的主要内容实际上就是研究如何对燃气阀门漏气产生的气泡进行自动检测，即如何对气泡检测系统进行设计。主要思路是基于C语言软件开发系统，采用图像测量识别技术，对所需要实现的系统功能进行结构设计，方法研究。气泡检测系统的功能实现主要分为以下几个步骤：1、图像的采集。2、对采集图像的预处理。3、图像分割。4、基于数学形态学的滤波处理。5、图像的二值化。6、连通域的检测和过滤。

图像采集的过程中需要注意摄像机采集气泡图像时，摄像机的拍摄速度应该与气泡产生的移动速度相适应，以便能够更好的捕捉到目标图像。图像的预处理通常包括：图像灰度化和噪声滤波，根据几种图像灰度化方法比较，本文采用了基于全局彩色分量统计特性的灰度化算法。而预处理中的噪声滤波，是对噪声的初级过滤，运用中值滤波的算法进行处理。图像的分割意义在于将所需要研究的图像部分：气泡，从整体图像中提取出来，便于后续分析、判断。通过连续两张图像做差值运算的方法，即可将运动的气泡从无意义的背景中分割出来。形态学的滤波处理不同于中值滤波的简单数值处理，它是运用膨胀、腐蚀算法对图像的灰度值进行逻辑上的滤波处理。图像的二值化分为全局阈值运算和局部阈值运算，通常有Otsu算法、灰度拉伸算法、Bernsen算法以及直方图算法，各方法优缺点比较后，选择最为常用的Otsu算法。连通域的检测和过滤，关键在与对连通域过滤器的设计，思路是根据气泡的大小、形状、密度等特点进行程序设计，以便过滤出不是气泡的连通域，留下目标图像气泡。

最后通过实验对设计出的气泡检测系统进行验证，实现对燃气阀门密闭性的自动检测。

关键词：气泡；图像；算法；检测系统

The leak detection of the gas valve

03012327 HEJUNLIANG

Supervised by ZHOUBIN

Abstract： The leak detection gas valve is one of the important features, however, manual inspection is not only time-consuming, laborious, but also the presence of low efficiency, high cost, error-prone and other shortcomings, Under the background of global aging, increasing labor costs, artificial intelligence, machine instead of the traditional large-scale production, such as labor trend has been open bow arrows. Based on this background we try to use an image detection method, by writing bubble detection system to achieve the tightness of the gas valve automatically detected.

       This topic through a complete test means a plurality of gas valves immersed in water, configure the appropriate light source and the image sensor, the valve pressurized continuous image, images captured continuously analyze and judge, recognition algorithm design to complete the gas valve automatic leak detection. Final algorithms through experiments already acquired sample image verification.

       The main content of this paper is to study how to actually bubble gas valve leak detection automatically generated, that is, how to bubble detection system design. The main idea is based on the C language software development system, using image recognition technology measurement, the SFCs required to achieve the structural design, research methods. Bubble detection system function to achieve is mainly divided into the following steps: 1, the image acquisition. 2, the image acquisition pretreatment. 3, image segmentation. 4, According to the mathematical morphology filtering process. 5, an image binarization. 6, connected domain detection and filtering.

       Image acquisition process needs to pay attention to the camera when collecting bubbles image capture speed of the camera moving speed of bubbles should adapt to be able to better capture the target image. The preprocessing of the image typically include: making the image gray and noise filtering, according to several gray image comparison method, we use the gray algorithm based on the statistical characteristics of the global color components. The pretreatment noise filtering, noise is the primary filter, using the algorithm of median filtering for processing. The significance of image segmentation target image need: bubbles extracted from the background image to facilitate subsequent analysis, judgment. By doing two consecutive image difference calculation method, bubbles can be split from the movement meaningless background. Morphological filtering process differs from simple numerical processing median filter, which is the use of the expansion, corrosion algorithm gray value image filtering process logic. The image is divided into global binarization threshold calculation and local thresholding operation, usually Otsu algorithm, algorithm gray stretch, histogram algorithms Bernsen algorithm and after carefully comparison of the advantages and disadvantages of each method, the most common choice Otsu(Da Jing) algorithm. Connected domain detection and filtering, the key to communicating with domain filter design, ideas are programmed according to the size of the bubble, shape, density and so on, in order to filter out the air bubbles are not connected domain, leaving the target image bubbles.

        Finally, through experiments designed to verify the bubble detection system, automatic detection of the gas valve tightness.

Key words: bubble image algorithm Detection Systems

目录

1、绪论 1

1.1课题的背景及意义 1

1.2课题的研究内容与方法 1

2、图像的预处理 3

2.1图像的采集 3

2.2图像灰度化 4

2.3图像噪声的滤波处理 5

3、图像的分割 6

3.1图像的差运算 6

3.2基于数学形态学的滤波处理 7

3.2.1数学形态学的内容和特点 7

3.2.2腐蚀和膨胀运算 8

3.3灰度形态学 9

3.3.1理论基础 9

3.3.2灰度膨胀、腐蚀运算 10

3.3.3灰度形态学基本运算的性质 12

4、灰度图像的二值化 13

4.1常见的图像二值化算法 13

4.1.1Otsu算法（大津算法） 14

4.1.2拉伸灰度算法 15

4.1.3局部阈值算法：Bernsen 15

4.1.4直方图算法 15

4.2灰度图像二值化算法的选择 15

5、连通域的检测和过滤 17

5.1连通域标记算法 17

5.1.1算法的有关术语 17

5.1.2像素点扫描法的基本原理 17

5.2连通域的过滤 18

5.2.1过滤器 18

5.2.2过滤器的设计思路 19

6、气泡检测系统的编写 21

6.1系统开发环境及实现的功能： 21

6.2主要程序的编写 21

7、实验检测 30

结语 36

致谢 37

参考文献 38

燃气阀门的密闭性检测

剩余内容已隐藏，请支付后下载全文，论文总字数：31343字