Research And Realization Of Automatic Measurement For Bacteriostatic Ring And Recognition For Labels

Antibiotics are widely used in medical. In order to control the quality of drugs, antibiotics must be measured strictly. Now, the general method is microbe standardization, it discuss the potency of antibiotics by measuring the diameter of bacteriostatic ring automatically, and then recognize and record drug labels manually. This process is complicated and inefficient. Therefore, using digital image processing methods, the bacteriostatic ring measurement and label recognition system is presented in this paper. It not only realized the automatic measurement of bacteriostatic ring diameter, but also completed the automatic recognition of antibiotic labels, and wrote the results into database. The study includes image processing, character recognition, circle detection, database technology and so on.All label characters must be extracted firstly from the whole image for effective recognition, including gray-scale treatment, median filter, pills localization, image binarization, character rotation, segmentation and normalization. Label recognition is realized by the method of feature matching, which calculate Euclidean distance between character and template feature vector, the template with smallest distance is recognition result. Feature vector of character is made up of hole, grid, intersection and outline characteristic. In accordance with label features, characters are divide into English and number, and then are sent to different classifier for recognition. This method have high recognition rate.To reduce the influence of background and interference points, each bacteriostatic ring sub-image is extracted from original image. After preprocessing, edge detection and clearing interference points in sub-image, the radius of bacteriostatic ring can be detected accurately with least squares principle. This algorithm can apply to the situation of edge intersectant and edge incomplete, its accuracy achieve sub-pixel level. Finally, in VC development environment, ADO method is used to operate drug information database. According to label recognition results, the measured radius of bacteriostatic ring is written into a new table.Experimental results show that, with high precision, strong adaptability, and high degree of automation, the entire system has a wide range of applications in biomedical field.