Geometric Parameter Measurement And Shape Identification Of Glass Based On PDAD

With the development of technology and the popularity of computers, human’sdemanding for intelligent control has become increasingly high. With regard of the factoryglass detection control equipment, different shape identification and real-time parametersmeasurement of glass play a vital role in processing and intelligent control. To solve thisproblem aforementioned, both here and abroad have put a lot of manpower and materialresources in image research. Lots of equipments at home and abroad mainly use machinevision systems to acquire object image by a CCD or CMOS imaging device and utilizecertain image processing methods to process image and calculate parameters. However, theexpansion of this system is low, and the demanding for substrate background is high andcostly. Therefore, this system can not meet the requirements of modern industry. This paperproposed a fast image processing method, fast measurement of geometric parameters andshape recognition based on photodiode array detection (PDAD) to address this problem.In this article, a series of treatment for collected glass image from PDAD is made tofinally display image in real-time and quickly calculate geometric parameters of glass andcarry out effective shape recognition to achieve intelligent control in the background of glassdetection of factory large-scale equipment. Firstly, median filtering and gain correctionprocessing methods are used to address the instability of electronic device and theinconsistency of detection module when images are captured, which will display processingimage in real-time to achieve real-time display of the movement of the glass. Then, imageswith severe serrated edges caused by low detection accuracy are processed by usingsegmentation and binarization, connected domain detection, image edge detection, adaptivethreshold extreme corner detection and chipping judgment described by certain mathematicalmethod to quickly and accurately calculate relevant parameters of moving glass such as length, width, area, angle, chipping and so on, after the acquisition of whole glass image iscompleted. At last, the shape of glass can be identified quickly by using extreme weight andfast and efficient shape matching algorithm. Experimental results show that algorithmsproposed and software designed in this paper work well in filtering irrelevant goals,preserving interested target and have high efficiency, timeliness and accuracy, which willmeet requirements of modern industry testing.