| The quality detection of the drug is closely linked to the health ofpeople. Pays special attention to the quality detection of the drug is the keyfor enterprise foothold. But many domestic pharmaceutical factories stilladopt the method of light inspection for defect detecting of capsules. Rely onworkers using eyes to check the capsule one by one. Judging from color andsilhouette is whether qualified or not. This traditional sorting method is oflow efficiency, high miss rate, easy cause repeat pollution, will have a hugeimpact on the quality of capsules. The main drawback of detection system forcapsule shapes on the market at present is of Low detection speed, lowdetection accuracy and a single defect detection types.In response to these problems, a two-channel rapid detection system forcapsule shapes is presented in this paper. Comprehensive automatic detectionfor360°of medicinal hollow capsule, the two end faces and internal defectsis realized. The detection rate of the system is84000grains per hour and themaximum error detection rate is less than1%. This system is well applied inthe detection for medicinal hollow capsule shapes. The following researchhas been made in this paper.Firstly, capsule image feature database is established, capsule defectsare classified and the defect images are analyzed in this paper. The technicalindex of the detection system is put forward. The main research contents andchapters of this article are expounded.Secondly, the hardware structure of the system is designed, including theoverall design of the mechanical structure of the system, the design of thefully expanded surface of the hollow capsule, mechanical system acquisitionmodule design and automatic reject ing device design. A set of reliable andstable mechanical system is designed to ensure the normal transmission of thehollow capsules, the high quality acquisition of images and the normaloperation of the filter system. Then, the software structure of the system is designed by using C++language and modular design method, including the image signal acquisitionsystem, image signal processing system, real-time display system and controlsystem. Capsule images are analyzed and defects are identified by comb iningwith the image processing algorithms. Mainly through the double thresholdsegmentation and square block algorithm respectively to realize theextraction of capsule profile and capsule combination, through edge detectionand hole filling algorithm to realize detection of capsule defects.Finally, this system is tested at the scene of the production. Test resultsshow that capsule detection system has a good detection accuracy andstability at the speed of84000grains per hour and can effectively ident ify thedefects such as mixed color, deformation, holes, plum flower top and so on.The maximum error detection rate is less than1%. This system can well meetthe requirements of modern industrial automation testing. |
