Font Size: a A A

Research Of High-speed Detection Method Of Embedded Small Parts

Posted on:2011-07-09Degree:MasterType:Thesis
Country:ChinaCandidate:D M LiuFull Text:PDF
GTID:2191330332469919Subject:Detection Technology and Automation
Abstract/Summary:PDF Full Text Request
In the industrial production, the method of manual detection to realize small parts detection has slow speed and short precision. And it would cause damage in the detection process. So this system requires a high-precision, real-time non-contact detection methods. On-line detection method based on vision inspection technology has faster detection speed and higher detection accuracy, uses non-contact automatic detection, saves manpower and reduces errors caused by human factors.Based on the above characteristics, this paper uses vision inspection technology to detect the appearance of small parts. Real-time image signal is captured by the high-speed CCD image sensor. Then this paper uses the DSP TMS320DM643 to realize image denoising, image enhancing, image rotating, edge extracting, identification and judgment. The result outputs to the control agencies through the control port.The main studies are as follows:1 By studying existing image processing techniques, this paper selects the median filter-ing to realize image denoising and gray-scale transformation to realize image enhancement. It highlights the interesting features and is more beneficial to observe small parts.2 The placement ways of the object of most of images gathered by high speed CCD do not lie in the horizontal position, and so the images are rotated to make the following operations easier. Firstly, the target area is located. Secondly, the smallest rectangle of the target image is found by the method of polar coordinate rotating, and then the area is rotated. This method can make the object place in the horizontal position.3 Contour tracking is used to realize edge extraction. Firstly, starting point is searched, from which the neighbor ones are searched to get the edge points. Secondly, the right edge point is the new starting point, and then another search for another edge point in the neighbors is raised, and so on. Finally, the chain code of object contour of the image is obtained.4 The chain code table is used to realize thread contour tracking and identification. Firstly, according to the features of small part thread, the corresponding laws of the standard chain code table are found. In the chain code table, the number of threads and chain code values of a single thread are determined. The similar principle of rising edge and falling edge of a standard single thread is used to judge a single thread, while the similar principle of the chain code total amount of rising edge and falling edge of each thread is used to judge whether the small part is qualified.5 In the different conditions, we carry out a large number of experiments for a variety of screws. The result shows that the method has higher reliability, higher efficiency, faster processing and better flexibility and can be satisfied in the high-speed detection system for different types of screws.
Keywords/Search Tags:DSP, Small Part, DSP/BIOS, On-line quality inspection
PDF Full Text Request
Related items