Research On Online Inspection System Of Tool Based On FPGA

In the process of product the tool,the surface of the tool is often affected by various factors such as production materials and processing environment,resulting in many spots or scratch defects.If it cannot be eliminated in time,it may greatly damage the tool's follow-up use.Currently,most of tools inspection techniques remain in manual or semi-system integration.The artificial detection is basically done by the eyes,the detection efficiency is relatively low,and the result is not necessarily accurate;the semi-system detection is to uniformly transmit the collected tool image data to the software for analysis.Although the semi-system test can complete large-scale detection,the pre-detection time is much more time-consuming and the acquisition speed is relatively slow.FPGA has the advantage of parallel processing,which can achieve the highest real-time performance;and it integrates a large number of mature IP cores,which can effectively implement various hardware-based image processing operations.This paper makes full use of the characteristics of FPGA real-time processing,and studies an online detection system based on FPGA.The implementation of the tool online detection system includes both hardware and software.Altera's Cyclone IV E series FPGA chip is used as the main control chip in the hardware,which completes image acquisition,data signal storage control,image pre-processing and transmission.The image acquisition module uses OV7670 image sensor to realize the tool collection work,design the sequential circuit and use Modelsim to simulate and analyze.The storage control module uses DDR2 SDRAM to realize the storage and conversion of data signals,and design and verify on the Quartus II.The image transmission module uses Gigabit Ethernet to complete the high-speed transmission of video signals,and uses Wireshark to capture the integrity of the signal transmission.In the image preprocessing module,when the color conversion is completed,the bilateral filtering and the IP core of the Sobel operator are designed to complete the noise processing of the tool image and the extraction of the contour region of interest,which provides a better sample for defect detection.In the software part,the defect detection software developedbased on the mature Halcon algorithm is selected to realize the defect detection of the tool.In this paper,the automatic detection and analysis of the common point defects and scratch defects are carried out.The comparison of the images before and after the inspection,and the description of the relevant detection coefficients,and the calculation of the tool according to the defect degree and the defect range of each tool rate.If the defect rate of the tool point or scratch defect exceeds20%,it indicates that the tool has serious defects,which is not suitable for use and should be eliminated.Finally,the two samples were tested for defects.The defect rate of spot defect detection in sample one is 8.59%,which is within the allowable range;its scratch defect defect rate is 3.13%,so the tool can be used for subsequent use.The defect rate of sample two under scratch defects reached 20.36%.Although the point defect of the tool is only 8.89%,it still does not meet the use standard,so it should be eliminated.In this study,a set of online defect detection system based on FPGA was built.The system transmits data stably,and the detection speed is fast,and the shape of the tool can be successfully extracted.It provides detailed analysis data for its spot and scratch defects,and outputs reference suggestions for the tool to continue to use,which provides a reliable way to further improve the detection of tools in industrial production.This system has certain reference value for on-line detection of parts in industrial production.