Research On Real-time Monitoring System Of Film Thickness Curve Based On Image Processing

In current development of industrial technology and multidisciplinary crossover study,a PC or digital display equipment is widely used to collect,process or display a variety of data,such as production index parameters,condition monitoring and fault diagnosis of mechanical equipment,etc.Since most outside dimension data acquisition devices are in different interfaces which need corresponding application software and their data transmission is low,it makes real-time data collection and processing perform not well and that the system is not easy to integrate and cost highly.But a data acquisition system based on USB possesses some advanced characteristics,such as easy extension,high transmission rate,low power consumption,low cost,flexibility and good compatibility.It has certain generality for a automatic detection and control system relied on computer.So a high-performance data acquisition system based on USB is a important way to solve these problems,and it can largely improve the production efficiency and production quality.The study in this paper focuses on a image acquisition and processing system to a thin film extrusion control interface based on USB.The aim is to realize the automatic control of film thickness uniformity.There are the main procedures of image data acquisition and processing:First,image data acquisition.A USB module(VGA2USB)is used to capture the image data of a digital display device in real time,then a internal procedure of PC sends a request to receive data.If this application confirmed the collected data valid,immediately the device sends a appropriate response and starts transmitting data.It achieves the communication between a digital display device and a PC by using USB interface protocol of CY7C68013 A,and also it can be used to process and display the color image.Second,the image data processing.The specific work is to detect target foreground and extract corresponding target data and character recognition from target color image.1)Curve data extraction.In view of using the traditional image segmentation algorithm to extract target data,the extracted data has great intermittent.In this paper,after determining the target interface and intercepting the target area of the target interface,two group of RGB component threshold based on the color image is set up to preliminary extract target information,then performed image processing.The improved threshold segmentation method based on Otsu is used to extract the target data,then which is stored in the database.Compared with the original image target curve and the curve after using a single algorithm,it is proved that the segmentation algorithm in this paper improves the continuity and accuracy of the data extraction.2)Target character processing.Characters are divided into three parts through their Contour features,using Canny edge detection to locate total width of the binary English character blocks.According to the characteristic of each English character same width,the characters are segmented into single character by scanning per column line by line.Then the character samples are determined after those single characters are normalized.3)target character recognition.As to the low recognition rate and weak robustness of single template or single feature recognition algorithm,a advantaged single algorithm is used to compare tested character with the character template and then find out the most similar template as a recognition results.At the last,the experimental results show that the multiple template matching algorithm is more robust and has higher recognition rate,which can be applied to engineering control.Finally,a image data acquisition and processing interface is created to real-time monitor the thickness data,then control the thickness uniformity of extruded film.