Research On Azimuth Recognition Of Block Workpiece Based On Machine Vision

In today's world,the competition between the manufacturing industry become increasingly fierce,along with the increasing manual labor costs,high-quality products,and the expanding proportion of automation systems,so the manufacturers' needs for automated detection system are more and more urgent.Considering the needs for automated sorting and packaging systems,the thesis researched on machine vision technology which was used to identify the orientation of the workpiece online and automatically,so as to replace the manual detection mode.When the workpieces are transfered online,their placing positions must be identified in real-time,then adjusted to the same position and entering the next step of the installation.The thesis proposed to use three-dimensional measurement technology based on laser vision to capture three-dimensional shape features of the workpiece.By extracting the laser light trace information projected onto the surface of the piece,the three-dimensional contour of the workpiece information in different observation plans can be obtained,transfering the identification problem of workpiece orientation into light trace curve shape recognition.Which not only can greatly reduce the amount of data in image matching recognition process,but also the recognition rate is higher and the speed is faster.Before the identification,in order to suppress the speckle noise and obtain information about the skeleton curve,the image filtering and refined process were needed.So a fast filtering algorithm and trace curves thinning algorithm were given.Taking into account the offset of the workpiece in the production line,therefore,the crossed curves were needed to be separated,each would be registrated and identificated separately.In terms of matching algorithm,defining two new kinds of distance measurements,respectively,as a crude match and exact match.In order to improve the accuracy recognition,proposing a classification method that of degree of attention to evaluate the similarity of the templates.System software was written by Visual C ++ 6.0,and the algorithm was applied to the workpiece position identification system of related company.The system is not only of high correct rate and high speed,but also is insensitive to the angular offset and translation of the workpiece;when the angular offset is ±15° and the translation distance is in 10 mm that the identifying correct rate is 100% and the average recognition time is 14.45 ms / block,better meeting the requirements of real-time online testing production.