The Measurement Of Industrial Components' Shape And Dimension Based On Image Processing System

1.IntroductionThe technology of measurement is the foundation and prerequisite for industry development.The improvement of mechanical processing technology is closely related to the development of the measuring level.Image measuring technology is the application of digital image processing technology to the field of accurate measurement. The principle of measurement is to detect the edge of the object and obtain the geometric parameters.Compared to the traditional methods, image measurement system has the advantage of high precision, large measurement range, good flexibility; no need for the complex mechanical movement and no special demands on the environment and etc.It is very fit for some field that is difficult for the traditional means. So it attracts a large number of outstanding scholars home and aboard to do some researches in this field. In recent years, image measurement technology has developed rapidly.With the increasing demand for the manufacturing and measuring technology, how to realize an efficient and intelligent measurement has became an important direction for the precise measurement.So the speed and accuracy are the two guidelines for measurement. How to realize a high precise and rapid measurement is the main task of this paper, and it is also the key technology.2. Research ContentThis paper focuses on a number of key technology that affects the measurement in industrial application.The purpose is to realize a high precise and rapid measurement.The image processing technology includes the research such as image filtering and edge detection and etc .In the aspect of image filtering, the paper first analyzes the source and type of noise that may affect the precision of the measurement, then adopts a improved median filtering arithmetic to remove the noise, and enhance the filter efficiency; In the aspect of edge detection, according to the characteristic of the image edge, the paper adopts eight improved direction masks based on classical Sobel mask to obtain the gradient information which can be used to find the single pixel edge of the image. This paper adopts an improved Hough transform to identify the geometrical shape of the image.This method has greatly reduced the space and time consumption. Then the paper brings forward a quadratic polynomial interpolation method to detect subpixel edge at grads direction of edge on grads image.In the hardware part, we choose a cheap, flexible USB transportation implement to complete the date transmission.At last, we setup a software platform to realize the algorithm introduced above.(1) Image filtering and edge detectionImage pretreatment includes noise processing and edge detection.The paper first analyzed the source of the noise.In view of that the impulse noise and salt&pepper noise are the two sources, we proposed a fast median filtering algorithm to remove the noise, reduced the time complexity. Image edge detection is one of the key technologies affecting the measuring accuracy.How to achieve the accurate position of the image is the main problem that will be solved this part.By contrasting various of gradient operators, we proposed eight improved direction masks based on classical Sobel mask to obtain the gradient information, it provides an effective solution to get a consecutive edge.Meanwhile, the value and the direction of the gradient have been got.It can be used to find the single pels pixel of the image.(2) Geometrical shape detection and dimension calculationThis chapter first introduced the calibration of the system to get the relationship between the unit pixel and actual physical size, and then discussed the application of Hough transform in circle and beeline detection, and put forward a improved Hough transform according to the geometric feature of the image which can reduce the amounts of computation.Using subpixel technology can realize the accurate positioning and improving the measureing accuracy.This section first discussed the conception of the subpixel and principle, then introduced several subpixel algorithm and gave a simple comparison.At last, we choose the quadratic polynomial interpolation method to detect subpixel edge at grads direction of edge on grads image.Experiments show that through this method,we can get a smoother edge and the accuracy can be enhanced.(3) The design of transmission module for the systemThis chapter first simply introduced the hardware platform of the image acquisition and processing system and its component modules, then its main work flow.The system adopts two kinds of technology to heighten the processing speed.Frist, the image acquisition and storage is done in a manner of Ping-Pong, which speeds up the parallel processing speed.Second, the application of universal serial interface USB in the image data transmission which is the paper's main task in hardware.We use a USB chip PDIUSBD12 supported by the USB1.1 protocol to realize the transmission.On the basis of the structure of the USB protlcol,the paper introduced the design of the firmware and its realization.In the process of transmission,the USB chip uses 3 endpoints in a mode of asynchronism.Endpoint 0 for the control information ,endpoint 1 for the system parameters and endpoint 2 for the image data transmission.Finally,we test the transmission rate and analyse the factors affecting the transmission and give some methods to improve its speed.(4) The design and realization of the software measurement systemThis chapter is the realization of the image processing algorithm above. Include reading the image data, system calibration, image filtering algorithm, image edge detection, geometrical shape detection and dimension calculation.The result of the experiment proves the feasibility of the algorithm and at last discusses the reason for the testing errors, and gave some methods to reduce the error.3. ConclusionThis paper focuses on the main factors that affect the application of the image processing method in industry field-precision and speed.it does some studies in both algorithms and hardware.The simulation results demonstrate the feasibility of the method and the repeated error is within 1/5 pixel. It shows that the algorithm design is basically successful, so it provides a theoretical basis for the image measuring equipment and has broad prospects and practical value. Keyword: Image Measurement, Geometrical Shape, Edge Detection, Subpixel,...