Research And Application Of Intelligent Monitoring And Recognition System Based On Embedded Terminal

In recent years,more and more embedded products with sensing and image recognition are used in wider fields.The problem arises: how to develop higher quality monitoring embedded products faster to meet the rapidly changing market needs?Based on the research of sensor technology and image processing,the thesis designed a general framework of intelligent monitoring and recognition system to support the monitoring of environment and personnel in different application.The framework consists of front-end embedded system and back-end service system.The front-end embedded system implements several parts of work: acquiring sensors' information based on multi-threading technology;using V4L2 framework to acquire video;compressing and transmissing video through H.264 encoding,and classification and recognition based on flower recognition algorithm.The back-end server system mainly implements displaying and storing sensor data,monitoring video,recognition results,or further classification and recognition of images.Taking the designed framework as the core and facing the application of flower intelligent monitoring and recognition,the thesis designs and implements a prototype system.The front-end of the system accomplishes three parts: acquiring environmental parameters such as temperature and humidity,carbon dioxide concentration and illumination;finishing the acquisition and compression of video;and using simple flower recognition algorithm to recognize flower categories.The video data is transmitted to the back-end server through UDP-based Socket network communication,and the sensor data and recognition results are transmitted by TCP.Considering the generality in Windows and Linux systems,the back-end server template is written by QT.Back-end server display and store environmental data,monitoring video and flower recognition returned from the embedded front-end.Finally,the embedded front-end and server back-end are tested.The experimental results show that the framework can support general intelligent monitoring and recognition applications.