Counting Method Of Thin Steel-plates Based On CCD Detecting

Along with the high pace developing of modern industry, thin metal plates(such as cold-rolled plate, galvanized sheet, etc.) are wildly applied in fields of home appliance, vehicle, light industry and so on. The numbers of the laminated thin material is usually too large to count efficiently. Currently enterprises machining thin plates tend to use mechanical methods to obtain the quantity of thin plates in process of machining. But there are some defects of mechanical methods in practical application such as departing while stacking, and decimating defected plates and so on, which will lead to error counting numbers in contrast to real quantity. Manual counting when delivering is inefficient and prone to error, resulting to failure of reviewing or counting rapidly.Form the perspective of image processing, this thesis introduces methods of non-contact and automatic detection counting laminated plates using CCD(linear and area-array) sensor, and also designs specific programs counting thin plates. The preliminary test results show that they can be operated successfully in production practices.The main content of the thesis includes:(1) for the nessorry of data processing of CCD detecting, the noise generated in image processing and filtering methods, and binary segmentation were discussed and analyzed.(2) System scheme of which methods of thin laminated steel plates counting based on linear CCD, was proposed and analyzed, and were with system components. Due to Circumstances of spurious peaks appearing, real peaks or wave troughs missing, an adaptive peaks detecting method based on auto-correlation principle and Fourier transform was proposed, which turned counting the thin plates to detecting and counting the wave peaks. The test results show that the system can operate succesfully at the laminated steel-plates specifications low to 0.7mm.(3) System scheme of which methods of thin laminated steel-plates counting based on area-array CCD, was proposed and analyzed, and were with its system components.An OTSU improved algorithm within rough background separation and self-adaptive segmenting threshold corrected is proposed, which is based on CLAHE principle, dimensional mean-filtering and morphology optimization principle of expansion and corrosion. Due to the shortcomings on missing or error peaks detecting in parameter space of traditional Hough transform, an algorithm based on auto-correlation principle and feature segmentation is proposed. With lots of tests done and method continusly improved, the sysytem has already met the preliminary requirements of workshop.Experimental results demonstrate that the two methods above meet the need of counting the laminated steel-plates. The linear CCD counting method can be used for specifications low to 0.7mm and total thickness to 600 mm, and its repeating cycle is 23 pixels. While the specifications of the area-array CCD counting method can be as low as 0.5mm, and repeating cycle is 14 pixels.