Research On Image Location Algorithms Of Diamond Machine For Watchcases Based On Computer Vision

With the fast onward march of the development of technology, the industrial processesneed to be more mechanized, and the various sectors of society have continually developed alot of new automated production lines and utilized in plants Behind this scene, there is anotherfact in watch assembling that the procedure of diamond embedding depends largely on handwork currently. It is difficult to meet the need of scale manufacture and wastes labor resourceto some extent. Based on this case, a diamond embedding machine is study in this thesis,which can recognize the location of the grooves for diamonds in the watchcase and thepositions of diamonds in the delivery plate for the purpose of embedding the diamonds in thegiven place accurately. The obvious benefit of this machine for the watch production lays inraising production efficient and making the automatic manufacture possible.The basic process of diamond embedding includes locating diamonds and grooves,dispensing and picking up diamond, moving diamond, and then placing and setting them.According to the procedures, this thesis has designed a overall system solution for diamondembedding, which from both hardware and software. In terms of the hardware, a briefintroduction to the visual system in selecting light source, camera, lenses and other equipmenthas been proposed. The lighting system has been established and all the types of its device aredetermined eventually. On the software, the software interface is simply demonstrated. Thetheory study of the image pre-processing and location are the crucial section in this thesis.Also, the function of each part is made clear.This paper put much emphasis on the study of the image location algorithm, andproposes two different locating schemes according to the different features between thediamond image and grooves image. In the image pre-processing stage, the methods offiltering, morphology and edge detection are introduced. After this stage, the variousalgorithms of circular identification and center location are focused on, which includetemplate matching, watershed segmentation and Hough transform. On the basis of thesealgorithms, this thesis adopts the FFT-based image matching algorithm to locate the grooveson watchcase, and the Hough-based circular detection algorithm to locate the diamond on the plate. The two schemes are effectively shortening the positioning time on the basis ofaccuracy. All the experiments achieved good results.Finally, after the test of the location algorithms, the algorithm flow charts of the locationsystem are given. Apply the algorithms in the practical production and analysis theexperiment results.