基于数字图像处理技术的弹体表面疵病检测系统的研究/

2019-04-28 13:49:25

The 检测 image processing detection



无损检测系统是利用物质的某些物理性质因为存在缺陷或组织结构上的差异而使其物理量发生变化的现象,在不使被检物使用性能及形态受到损伤的前提下,通过测量这些变化来了解和评价材料、产品和设备构件的性质、状态或者内部结构等的一种特殊检测技术.目前,工业中用于无损探伤检测的方法主要有:超声检测(UT), 磁粉检测(MT),涡流检测(ET),声发射检测(AE),泄漏检测(LT),渗透检测(PT),射线检测(RT),中子射线检测(NRT),目视检测(VT)等。随着图像处理技术和计算机技术的发展,基于图像处理的磁粉检测检测方法得到了越来越多的应用。实践也证明磁粉检测是目前弹体缺陷无损检测中采用较多并且行之有效的方法之一。但是一直都是由操作人员人眼目视辨别检测结果并进行判别,自动化和生产效率较低,无法对检测过程和结果进行回溯,并且存在大量不安全因素。我们在原来检测手段的基础上,大量引进和采用计算机技术、传感技术、通信技术、数字图像处理分析技术实现信息的获取、传输、处理、控制等功能,使炮弹磁粉无损探伤检测工作实现自动化,具有十分重大的意义。
文中设计的检测系统由相机、相应传感器、图像采集卡、计算机、显示设备、存储设备、打印机、报警设备、控制系统等部分组成。根据检测物体和检测技术指标的要求可以选择相机(包括相机的分辨率、个数,线阵相机,面阵相机等),相机采集的图像信号通过配套的图像采集卡送到计算机进行处理,同时由传感器获得的物体转速等信息送到计算机进行分析,计算机根据这个信息控制相机的采集速率等参数,另外,根据实际需要计算机可以外接显示、存储、打印等设备,计算机对图像采集卡采集的图像信号进行处理、分析后获得有用的信息控制报警设备并把相关控制信号送往外部控制设备(可编程逻辑控制器PLC等)。
系统中里选择的是由加拿大达尔沙公司生产的DS-44-04M30标准数字相机,该相机能够很好地满足我们这个系统所要测量的最小精度(0.5mm);而且高的灵敏度,四通道高数据率输出和LVDS传输方式使它的数据传输率达到160MB/S,能够满足系统实时性的要求。选择的数据采集卡是由加拿大的一家公司CORECO公司生产的The Viper-Digital卡,该数据采集卡可以很好地同DS-44-04M30标准数字相机相容,能够四通道接收各种常用的数据格式的数字图像信号,并且通过PCI总线快速地传送给主机或者是其它的专用图像处理卡,还可以嵌入一个像素处理器,以便完成图像的高速实时处理。磁粉探伤机的中控系统我们选择了由日本欧姆龙公司生产的CPM1A-40CDR-A型号的PLC。该型号的PLC完全可以满足我们所设计的这个系统需要,并且具有结构紧凑、使用方便等优点
文中论述了数字图像处理的基本原理和我们所选用的在弹体表面疵病自动检测系统所使用的处理方法并做了Matlab仿真和分析。主要介绍的基本原理有:平滑、锐化、二值化。在图像的平滑处理方法中我们说明了直方图均衡化、均值平滑滤波以及加权均值平滑滤波等方法;而在图像的锐化部分阐述了梯度运算原理和拉普拉斯算子原理。此外,文中也阐明了目标图像的识别原理以及我们这个无损检测系统所采用的目标图像识别的方法。目标图像识别的原理主要介绍了边缘检测和门限处理等方法。在边缘处理中说明了一阶梯度算子和二阶拉普拉斯算子,并且介绍了使用它们的利与弊。在门限处理中,论述了基本全局门限、基本自适应门限和最佳全局与自适应门限等方法,并且给出了部分理论推导。图像边缘检测技术给出了matlab仿真程序、结果及分析。
检测系统采用的图像处理分析的技术随着检测环境、系统检测精度、技术要求的不同而不同。市场上也有大量的图像处理软件。加拿大Coreco公司的Sapera软件是一种适合大多数检测要求的图像处理软件。Sapera软件是一种新的应用程序接口(API),32位面向对象编程, 多线程库、模块化设计、与设备无关、与C++ 兼容、可用Activex、Sapera API易读并且容易理解。
该检测系统有分辨率高、速度快、硬件简单、软件功能灵活丰富、抗干扰能力强、可回溯性好等优点,在弹体表面无损探伤检测中有很好的应用前途。同时,它具有较强的通用性和功能扩展等特点,在其它产品的无损检测中具有广泛的推广价值,可用于各种无损检测领域,对于检测过程的安全性,提高检测水平,检测精度和生产效率,大幅度地降低人的劳动强度,降低生产成本也具有重大的实用意义。 当然,这样一个复杂的检测系统需要许多研究人员的共同努力,论文中仅把系统相关的硬件、软件原理和设计过程等内容做了详尽的讲述。但是,光学系统设计、电部分硬件单元设计、 SAPERA软件、图像处理分析技术分析和仿真等部分内容的阐述将对其它检测系统整体设计、开发、调试有一定的参考、指导价值。



The nondestructive detection is a kind of special measuring technique. The objects’ physical variables can be changed by the physical character which exists the flaws or the difference in the organization structure. On the condition of without destroying the object’s usability and the formation, with measuring these physical variables can we know and value the feature, state or inner structure of the material, product and equipment part. At present, there are many nondestructive detection methods in the industry, such as: ultrasonic test (UT),magnetism dust test (MT),electron test (ET),acoustic emission test(AE),leakage test (LT),percolation test (PT),ray test (RT),neutron ray test (NRT),visual test (VT etc. With the development of the image processing technique and computer technique, more and more people use the measuring methods which base on the image processing. The paper introduces the application of image processing and image analysis techniques in MT system. Practice also, process that MT is one of all the effective methods that is widely utilized to detect bomb surface flaws. But MT is usually practiced through the eyes of the operators by identifying the test results and pointing out the defects, which is quite low in the automation level and productivity. Meanwhile it’s impossible to recall the test results. Moreover, there is a great deal of unsafe factors existing. As a result , it is of great importance to introduce and utilize the computer techniques ,communications techniques, data image processing and analysis techniques in order to serve the functions ,such as obtaining information ,transferring information, processing information, controlling information and so on ,which can realize the automation of bomb surface flaws detection.
The system is constructed of the cameras, corresponding sense organ, image sampling cards, computer, display device, printer, annunciator and control system etc. According to the detection demands of the examined objects and the technical requirement, we can choose the cameras (including the resolution, amount, linear camera, frame camera etc). The image signals acquisitioned by camera transmit to the computer to be processed through suited image sampling card. Simultaneously, the signals of object rotate speed acquisitioned by the sensors are sent to computer to be processed, then the computer controls the parameters of sampling speed etc according to the result .On the hand, according to the actual demands, computer may be connected to the display device, printer, memory storage etc. After the computer processes and analyzes the image signals acquisitioned by the image sampling card, it can get useful information which can control the annunciator, and transfers correlative controlling signals to the external controlling equipment (programming logic controller (PLC) etc).
The system chooses DS-44-04M30 standard digital camera, which is produced by Canadian DALSA Company. The camera can satisfy the minimum accuracy (0.5mm) demands of the system, and high-sensitivity ,four-channel data outputting can make the data transfer rate reach 160MB/s,which also satisfy the real-time demand of the system . The data sampling card is produced by Canadian CORECO Company. The Viper-Digital card, which can cooperate well with the camera and receive all kinds of digital image signal in common data mode through four channels, which can be transfer to the host computer or other special image processing card. We can also embed a pixel processor to make a high-speed and real-time process of the image. We choose CPM1A-40CDR-1 PLC produced by Japanese OMRON Company as the CPU of the Magnetism-particle machine, which can satisfy the demand of the system .It also has a lot of advantages such as compact structure and convenience in operating.
The paper discuses abut the principles of digital image processing which is a the processing mode we choose in the auto detection system of the flaws on the surface of the bomb .it also mentions MATLAB emulation and analysis. The principle mainly includes: image smoothing, image sharpening etc. In the introduction of the processing modes of image smoothing ,we illustrate the different modes such as : histogram equalization,smoothing linear filtering and order-statistics filtering,while in the section of image sharpening we expound gradient and Iaplacian operator. Moreover ,the author also elaborate target identification principle and the modes we choose in the nondestructive detection system .The principle of target identification mainly introduce edge detection and threshold processing etc. In the part of processing the author explains first derivative gradient and second derivative laplacian operator, in addition to the advantage and the disadvantage of that. In the threshold processing modes, the principle of global threshold, optimal threshold etc is discussed .In the explanation of image edge detection techniques, MATLAB emulation program, results and analysis has been elaborated in details.
The different detection systems adopt different image analysis techniques that vary with detection circumstances, system detection accuracy and specifications. There is a lot of image processing software in the market. The Sapera software produced by the Canadian Coreco Company is software that suits most of the detection requirements. The software is a new application program interface (API) with the following characteristics: 32 bits object-orientated programming, modular program. The software has no relationship with equipments and is compatible with C++, also can be called by Activex, easily read and understood.
The detection system has many advantages as follows: high resolution, high speed, simple hardware, flexible and abundant functions of software, strong anti-interference and good plasticity. It has a bright application future in nondestructive detection on object surfaces. Meanwhile, it is outstanding characteristics, such as combinability and expandability; it also has extensive value in other product’s nondestructive detection systems. It is of great practical importance, as to the security of the detection procedure, improvement of detection level, precision and productivity of detection, great reduction of people’s labor intensity, and reduction of the production cost. Certainly, such a complicated detection system needs a lot of researcher’s joint efforts .The paper only focuses on the hardware principle, software principle and design process, which is correlated with the system. But, the explanation of the design of optical system, the design of hardware units, SAPERA software, image processing and analysis techniques, is of referential and instructive value to other detection system’s design, exploitation, and debugging etc.