Drug identification method based on improved SURF algorithm

Abstract

The invention discloses a drug identification method based on an improved SURF algorithm. The method comprises the steps that a drug bag image is acquired; constructing a Hessian matrix to generate interest points; constructing a scale space; positioning feature points; distributing main directions of the feature points; generating a feature point descriptor; carrying out initial extraction on thefeature points through an Euclidean distance, then screening the feature points through cosine similarity, if the cosine value ratio between two vectors is measured to be smaller than a set thresholdvalue, screening out the feature points, and otherwise, reserving the feature points; and finally, using template matching to complete drug auditing identification, i.e., selecting a template most similar to the detection image according to the matching number of the detected drug bag image and the template image matching points, wherein the type of the template is the type of the detection drugbag image. According to the method, the experimental environment during drug auditing and matching is comprehensively considered, the matching and recognition accuracy of drug bag features is high, and the method has good robustness and high matching speed in rotation and illumination.

Description

technical field [0001] The invention relates to the fields of machine vision and image recognition, in particular to a drug recognition method based on an improved SURF algorithm. Background technique [0002] The traditional drug identification review part often uses manual identification to classify drugs, and adopts a standardized management system and placement system to classify and place them to facilitate subsequent manual identification and improve efficiency when taking drugs. However, manual identification is time-consuming and labor-intensive, and the accuracy rate is often not high. [0003] With the popularization of intelligence, the application of drug identification codes has appeared. Each drug approval number is like a drug ID card. Drugs are represented by a batch number. Although the method of using the drug identification code has a high accuracy rate, the labor demand is still relatively large. In addition, in recent years, radio frequency identifica...

AI Technical Summary

Problems solved by technology

Although the method of using the drug identification code has a high accuracy rate, the labor demand is still relatively large

In addition, in recent years, radio frequency identification technology RFID has also had some applications in drug identification, but radio frequency identification technology has problems such as high label cost and labeling cost.

Images