Research on Parts Measurement Method Based on Machine Vision

Research on Parts Measurement Method Based on Machine Vision

Jun Zhang (School of Electrical Engineering and Automation, Tianjin University, Tianjin, China), Shuhua Wang (School of Electrical Engineering and Automation, Tianjin University, Tianjin, China) and Zhengling Yang (School of Electrical Engineering and Automation, Tianjin University, Tianjin, China)
DOI: 10.4018/ijapuc.2013070104
OnDemand PDF Download:
No Current Special Offers


This paper has described the measurement method of the phone slot on the assembly line. The method is based on computer vision. After reducing the effect of noise through fast median method, the authors locate the target area. Thus the authors extract the target area. The authors have proposed an edge detection algorithm based on the improved canny operator. The authors also have put forward the linear fitting method based on RHT-LSM and then the authors delete the straight line whose slope is greater than the given threshold. Therefore the authors can find the corresponding angular point coordinates. Then the authors have compared the semi-circular data of the slots ends and find the points whose tangent line’s slop are biggest. These points are absolutely tin the leftmost and in the rightmost. Thus the authors have got the length and the width of the image in the coordinate system. Then the authors can get the camera’s internal parameters and external parameters after the camera calibration. The practice shows that the system is feasible and it has high use value.
Article Preview

1. Introduction

Based on the image processing, the technique, which takes camera as its image sensor, is a non-contact measurement method. The method has been widely used in industrial field, for that the machine vision technology has the advantages, such as non-contact measurement, wide spectrum range and the ability of working long time(Wen-Ge Wang, Zhang-Yi Hu, & Zong-Yu Shun, 2002).In this paper, we have researched the method to detect the slot of mobile phone shell which is on the line and measure the size of it. The method is based on the theory of digital image. The implementation process of the system is shown in Figure 1.

Figure 1.

The system flowchart


The gray image of the mobile telephone shell can be seen on the conveyor belt. The black outline region of the gray image is Mobile phone shell’s slot whose length and width are 22mm and 1.5mm respectively. In addition, the measured precision required is 0.08mm.


2. Binarization Processing

When we gray the original image in different experimental condition, the gray levels of the image are different. In order to improve the universality; we determine the binarization threshold which is used to binary the slot of mobile phone and the surrounding background through the gray image’s histogram. From the gray image, we know that the grey values of the image are mainly focused on three parts, such as: slot part,the background of the image, the parts which are around the slot and illuminated by the light source. The differences between the three parts are obvious. The proportion of slot around which is represented by the smallest peak grayscale range (250 around) is small and located in the mobile phone’s shell. So we can regard it as two kinds of problems. We binary the image in the method of the biggest variance between the classes and find the best separation point between the phone’s slot and the background. We regard it as the threshold of adaptive binarization.

The local average filter is a commonly used algorithm. When we process the random impulse noisy signal in this method, the pulse noise falloff somewhat but it still influences the result of the filter significantly (Krupiski, 2010; Dong Fu-guo, Yuan Da, & Wang Jin-peng, 2007). The median filtering method is to sort the data within the window and we take the middle value of the result. So the impulse noise doesn’t work and it has no effect on the result.(Dong Fu-guo, Yuan Da, & Wang Jin-peng, 2007) The median filtering can decay the random noise without blurring the boundaries, so it can protect the original signal well and has good smoothing effect in the case of small grey value change. The Median filtering method can overcome the image detail blur which is coursed by the linear filter, the smallest square filtering and average filtering. It can effectively filter out the pulse interference and the noise coursed by scanning image. So it can protect the image edge while removing noise.

The filter window moves on the original image from left to right in the following method: we remove the pixels of the most left one column in the window and put the pixels of the one column which is next to the original window into the new window. Because the grey value of the pixels in the original window is well sorted, we only need to sort the new pixel sequence. We use a 3*3 window and we need to compare 3 * (3-1) / 2 times in each column pixels in the window. If we insert the new added list into the orderly sequence, we need to compare 3*3 times and calculate 12 times. Compared with the traditional median filter, the times need to compare have been significantly reduced. In addition, the efficiency of the median filter has been improved.

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 11: 4 Issues (2019)
Volume 10: 4 Issues (2018)
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing