The Design And Implementation Of The Kit Image Recognition System

With the improvement of science and technology,automation and intelligent equipment has been gradually applied to all aspects of the whole society.Pharmacy automation has reached a high level in western developed countries.And automatic pharmacy system has been put into use in our country to help with dispensing work.However,most of the current equipment can only complete the dispensing process automation,warehousing work still needs to be done manually,resulting in relatively low pharmacy efficiency and warehousing accuracy which can be higher.A kind of medicine box identification system based on the image feature is designed in this paper.It can solve the automation problem of the medicine warehousing and distribution by detecting the image characters of the carton packaging,which can greatly enhance the automation level of the pharmacy.According to the system requirements,the overall design scheme of the system is given.The requirements analysis and design of the lower computer control system is completed,including the selection of the lower controller and the motor,the design of communication protocol and the design of the lower computer control program.This paper also completed software design,including database and software.The features of the medicine box images are analyzed,and the images are identified by detecting the color features and shape features.A medicine box image detection and segmentation method based on the Harris corner is designed.A comparison test between K-Means and octree-based main color feature extraction algorithm is conducted.The more efficient octree algorithm is used to calculate the image main color features.A new improved shape context algorithm based on Harris corner is proposed to extract the shape feature of the carton image.The effectiveness of the algorithm is validated by experiments.The system test environment is built and the system is tested.The validity and usability of the system is verified.