| Stone is a kind of common material in architectural decoration industry, which is widely used in the construction of public facilities and home decoration. In recent years, the investment in infrastructure construction is increased, which further stimulates the demand for the consumption of stone. But the automation level of stone production process is low in China, the production testing of stone is still done by workers, which leads to high cost and low productivity. In order to reduce the cost and improve the productivity, the introduction of visual detection equipment with low cost and high detection efficiency is very necessary. Embedded machine vision has the characteristics of low cost, small volume, easy installation, fast processing speed and stable operation, and it can replace workers to complete the frequent and dry guide detection work in stone cutting process, so it has a very important significance to research.Based on the needs of production, the method of embedded machine vision is used to guide the cutting process of large size stone is studied. The main contents are as follows:(1) According to the requirements of design, a cutting guidance scheme using embedded machine vision is proposed. Based on the characteristics of the detection process, the QY-IMX6 S control board of the embedded processor based on ARM core is used as the hardware processing platform. In view of the demand of large size stone detection, an image acquisition scheme using two cameras cross sampling is proposed, and the problem of large amount of information is brought about by large size, the speed of processing is improved by parallel operation of the two control board.(2) The environment of development and debugging based on embedded Linux system has been built. The V4L2 drive structure of Linux system is studied, and the image acquisition has been realized by using V4L2. Because the image processing algorithm only needs the red component, the method that the red component is extracted directly from YUYV format is used to reduce the amount of storage data and speed the image acquisition.(3) The types of defect of stone are analyzed, and the pretreatment process of image, such as gray scale, filtering and binary is studied, and the appropriate preprocessing algorithm has been selected. The least squares method and Hough transform is studied for the linear fitting of the laser line in this dissertation. Considering the interference of environmental light, Hough transform is chosen as the final fitting method, and the amount of computation in the fitting process has been reduced by limiting the range of angles. Through the determination of the target range and the conversion of coordinate system, the process of Hough transformation is optimized, and the processing speed is improved. Because of the difficulty of image distortion and distortion correction, the method of segmented fitting is used to reduce the effects of distortion. The detection of defects is determined by the change of the gray value of the points on the fitting line, and the average gray value of the laser line has been obtained to improve the accuracy of detection(4) The process of interaction between the embedded vision guidance system and the control platform of stone cutting has been designed, and the software has been designed and developed. Under the laboratory conditions, the test of defect detection has been done, and the results are analyzed. Based on the results of experiment, detection algorithm of system can identify defects of 8 mm and above, and can complete a detection in 0.4 s, which can meet the demands of the system design. |
PDF Full Text Request