| With the gradual maturity of image processing technology,embedded image processing systems have started to be used in the fields of face recognition,traffic detection,container detection and so on increasingly.However,most of these applications need to develop an independent system to support algorithm operation,which greatly causes waste of development cost and time.To solve the above problems,this thesis extracts the common requirements of embedded image processing applications and designs an embedded image processing platform for supporting different image processing tasks.The platform consists of an embedded front-end and a server back-end,and extracts the common modules involved in the image processing process,such as video capture based on Huawei SDC camera,data transmission based on the TCP socket and data storage.In addition,the platform reserves algorithm interfaces to meet the algorithm requirements of different application systems.Secondly,this thesis conducts research on container coding recognition and proposes container coding recognition algorithms based on Paddle OCR technology respectively according to the requirements.On this basis,the system code recognition algorithm module is designed to realize the front-end recognition and back-end recognition functions.Finally,with the designed platform and the container coding recognition algorithm as the core,a container coding recognition prototype system is implemented in this thesis.The front-end of the system mainly performs the following tasks: acquiring container video information,drawing frames from the video and saving the images;calling lightweight algorithms to perform container encoding recognition for the captured images;transmitting recognition results,original images and caching the transmitted data to the back-end of the system.Among them,in order to enhance the system recognition effect,there is a container using multiple cameras for video acquisition at the same time,and the front-end recognizes these images in parallel and uses voting to obtain a final front-end recognition result.The back-end of the system is based on SSM framework and relies on My SQL database for data management,whose functions include receiving,storing and managing the original images transmitted by the front-end of the system,processing the results,and encoding and recognizing the images again to check the recognition results of the front-end of the system to generate the final recognition results.In addition,a Web application-based client is used for user interaction,providing users with real-time information viewing,historical information viewing,local image encoding recognition and other functions.The test results show that the container coding recognition system based on the universal image processing platform has good reliability and meets the design requirements,which can meet the practical application requirements and has certain application value. |