Research Of Food Storage Moni-Toring And Recognition System Based On Enbedded

With the development of internet technology and embedded technology, intelligent ware-house system and smart home system are very popular in researches and all aspect of people’s life. Some developed countries in Europe and America have developed more mature intelli-gent warehouse and smart home system based on internet of things. In recent years, plenty of domestic researchers are working on that and also achieved good results. This article studies the food storage monitoring and recognition system which involves many technologies, in-cluding computer technology, image processing technology, network communication technol-ogy, electronic circuit technology, etc.The food storage monitoring and recognition system in this article is divided into client and server. PC machine worked as the client and the target board (ARM A9) worked as the server, them communicate through TCP/IP protocol. The server can be divided into four pro-cess, which respectively are image acquisition process, environmental information collection and control process, GSM alarm process and the Qt process; the client is mainly used to be a monitoring and identification tool.Video4Linux2 camera under linux kernel driver is used to collect images and sent to PC through Target board.Those images are the data sources of face detection and face recognition. The client adopt Qt design system login screen and display interface that include face detection and face recognition.Face detection and face recognition are implemented based on OpenCV vision library in this article. Face detection uses AdaBoost algorithm combined with Haar rectangular characteristics, the improved algorithm greatly improves the speed and accuracy of detection. Face recognition algorithm is PC A-LDA algorithm. Firstly, single LDA algorithm is with "small sample" shortcoming. Secondly, PCA is very good efficient in dimensionality reduction and feature extraction, but non-supervised learning and low precision. While PCA-LDA algorithm can better solve the problem of the above two algorithms. MO collect ware-house environment information, and then send it to the target board with a low power and short distance Zigbee wireless technology, the target board can also transfer command to con-trol MO module, MO makes the corresponding operation.GSM module uses the GPRS remote wireless transmission technology, it is able to send text messages to the mobile phone that the system set up, the wireless transmission technology of GPRS and Zigbee improve the reliability of the whole system.This article completed the embedded system platform, including the selection of hard-ware, the programming of software system, the transplant of linux system, and the transplan-tation of JPEG library and SqLite database.Test results show that the monitoring and recognition system based on ARM A9 target board, can monitor the whole warehouse momently. No matter if in similar background and condition of uniform illumination, face detection and face recognition accuracy is still high.