Outline Detection And Brand Recognition Of Link-plate Based On Image Processing

Along with the enhancement of the industrial production level, today,the society demands more product as well as better product quality, and theautomated production is the inevitable trend. The component automaticdetection has developed along with the unceasing development ofAutomation and the intellectualized technology in the modernizationindustrial production, there are great meanings that it is not only saving thetime and the work force, but also improving production efficiency and theprecision of the detection.The link-plate of chain is struck from the striking board, so the outline oflink-plate is possibly not entire;moreover, there are some differences in thecircumference and inner hole between the two surface of the link-plate, suchas the round angular zone, the luminous zone and the crack zone. Thisdifference owing to the processing means will affect the quality of theassembled chain very much. And in order to avoid this bad effect, andenhance the fatigue strength, the resistance to abrade and the appearancequality, each production enterprise will use the technical method generally,which is to distinguish the different faces of link-plate first, and then toassemble. This paper uses the computer image processing technology torealize the detection of the outline and automatic brand recognition oflink-plate. And we have done some research in the following aspects:Firstly we discuss the image preprocessing algorithms, mainly includingthe removing blur caused by uniform liner motion and the image smoothingprocessing. In latter, it compares and analyzes the complexity of algorithmsand the removing noise effect of both the neighborhood mean value filterand median filter in detail. And the experiment proves that the median filternot only could be able to remove the noise effectively, but also maintain theedge information of the link-plate image better, enhance the interestingcharacteristic in the picture. It was more advantageous to carry on theobservation to the goal object, but there are some shortage that it is notsuited to process the image including many thin line and the details.The edge in the image is one of the most basic and also the mostimportant characteristics of the image. The edge detection is classicalresearch topics in the computer vision and image processing domain all thetime, but people have not found a perfect solution to such question until now.The edge detection is one of basic steps of image processing, the imageanalysis, the image recognition, the computer vision as well as human vision.The result's accuracy and reliability will affect the machine vision system tounderstand the objective world directly. First, it introduces the definition ofthe edge, the edge detection steps, requirements and the appraisal method.In the edge examination algorithm aspect, it mainly carries on thecomparison with all kinds of traditional algorithms (Roberts, Sobel, Prewitt,Laplacian, Krisch, canny and so on), elaborates the principle of eachalgorithm, and realizes them with the C + + Builder programming tool, andthen makes the analysis to the identical picture of static link-plate with thesemethods, compares the detection effect of each algorithm intuitionally.Regarding the fuzzy link-plate picture caused by the movement, anapproach to arc recognition based state machine was developed first time inthis paper. It has studied the arc recognition condition criterion, formulatedthe arc recognition state table and the next state operation table. Throughthe simulation with software programming, and the result shows that thismethod is practical. In this foundation, the arcs are grouped and matched tobe the shapes of link-plate, then we recognize the link-plate's outlineaccording to the shape characteristic of link-plate, the mutual position ofcorresponding arc, the concave-convex relations. In order to distinguishbetween the obverse and the reverso of link-plate, we take the bland as thedetection object, through examining the edge quantity in the link joint waistregion whether bigger than the certain threshold value. If bigger, we thoughtthat we have found the bland in the above side, namely has found theobverse side that contains the bland.For the sake of fulfilling the real time request of image processing, thispaper discusses how to realize the link-plate detection and recognitionsystem with hardware, which is mainly composed of FPGA 1K50 and ARMLPC2106. It introduces the application and the hardware knowledge aboutFPGA 1K50 and the ARM LPC2106 simply, mainly discusses how to realizeand debug the median filter algorithm, the edge detection algorithm and thearc recognition based on state machine with FPGA. We defined thecommunication agreement between FPGA and ARM, and finishedcommunication procedure between FPGA and the ARM, the ARM's startingcode and display procedure using serial interface.