Research On Tablet Vision Detection System Based On Machine Vision

In recent years,the rapid development of China’s pharmaceutical industry,in which tablets as an important part of the pharmaceutical industry,its quality is directly related to the patient’s life and health and safety.At present,many domestic enterprises through the manual to complete the tablets quality detection,which is time-consuming and can not guarantee the reliability of test results.Therefore,it is imperative to study the automatic detection system of tablets.The rapid development of machine vision detection technology provides an effective technical support for the nondestructive testing of tablets.Therefore,this paper designs a Tablet Detection System based on machine vision.In this paper,broken tablets,scratched tablets and stain tablets appeared in the production of tablets were taken as research objects,the working principle and testing procedure of the Tablet Detection System were expounded in detail,and the algorithms of tablets defect detection were emphatically studied.Tablet Detection System mainly consists of two parts:hardware devices and software algorithms.In the part of hardware devices,this paper sets up an image acquisition unit for the actual production and testing performance of the tablet,and made a detailed research and analysis on the camera,the light source and the lens components that make up the image acquisition unit,the final determination and choice were the MV-VS078SF-C model CCD camera,AFT-2514MP model of the lens aind the AFT-RL100W model ring LED light source,and the light source adopts the frontal vertical illumination to the test tablet.The image acquisition unit can obtain the image information of the upper surface and the side surface of the tablet,and the obtained image of the tablet is less noise.In the part of software algorithms,in order to solve the contradiction between the filtering of pepper and salt noise and the protection of image details,this paper presents a new detail protection filter algorithm based on fuzzy inference system---adaptive fuzzy median filtering algorithm.After the image was filtered by the new algorithm,the details of the image can still be preserved.In order to verify the detection accuracy of the system,we made a detailed introduction to the classic Zhang Zhengyou method,calibrated the camera with MATLAB stereo calibration kit,obtained the corresponding internal and external parameters of the camera,and calculated the pixel area of the tablets according to the actual diameter of the qualified tablets(theoretical value),compared with the actual software algorithm measured pixel area,verify the system to meet the detection accuracy requirements.In order to separate the defect information accurately,this paper proposes an improved fuzzy membership segmentation algorithm based on the traditional fuzzy theory image segmentation algorithm.The mean and variance were introduced into the membership function,and the ratio of the variance between classes and the variance within classes was taken as the objective function.The experimental verification show that the algorithm can extract the defect characteristics of the tablet accurately.Combined with the geometric characteristics of the tablet defects,the algorithm can accurately detect unqualified tablets.After testing,the success rate of detection of tablet fragments and tablet stains are more than 95%,verifying the suitability of the Tablet Detection System.