Research On Eligible Detection Method Of Syringe Needles Based On DSP

Defective needles such as flip and the hook may be produced during the production process of syringe needles.If these defective ones are used, which will do harm to people's physical and mental health. Therefore, some effective measures should be taken to detect defective ones. Currently, most syringe manufacturers mainly use artificial visual detection method to eliminate defective needles, however, because the syringe needles is very slender,the diameter of needle stem is no less than 1mm, workers'visual may get fatigue, which will cause omission and error detections. To improve the qualification rate of products, a quality detection method of syringe needles based on the experimental platform of TMS320DM642 development board is proposed, which combine the characteristics of syringe needles with machine vision identification method.The content are listed as follows:(1) Implementing image acquisition of syringe needles based on TMS320DM642. With CCD image sensors, microscopes, lighting, TDS642EVM multi-channel real-time image processor and analog TV, building a syringe needle detection hardware platform. Realizing needle image real-time acquisition and display using API functions in real-time operating system (DSP/BIOS) made by TI company.Utilizing backlight method to reduce the influence caused by external light source on image acquisition.(2) Researching image pre-processing methods of syringe needles. Using edge information to keep a good median filtering algorithm to eliminate the impact of noise on the detection algorithm, using Otsu to segment the needle out from the image background, using the biggest connected region method to extract needle tip to remove the interference of the pseudo-target, using the boundary tracking method to extract the contour of needles edge, using image correction and the tip part of the extraction methods, extracting accurately the contour of needlepoint edges, establishing foundation of accurate extraction for following characteristic information of needlepoint.(3) Studying the detection methods of the syringe needle eligibility. Firstly, to achieve needle flip detection, the extracted tip profile feature information is matched with eligible needle the feature library, which is established by using the border region moment invariants method. Secondly, in the local coordinate system of tip, the detection of the needles is carried out according to the maximum curvature of the two edge curves, which is fitted by least-square method. Finally, output of tip detection results are achieved through voice prompt module which is controlled by the GPIO port of TMS320DM642Through the detection experiment of eligible and defective needles, the results show that the method has a higher detection rate to defective needles and a lower error estimation rate to eligible needles, and it can be applied to the actual production.