The Research Of Intelligent Video Surveillance Module Based On Image Recognition

In recent years, with the progress of the development of society, science and technology, surveillance technology is increasingly used to maintain social stability and playing the bigger role in home and national security systems. In the current surveillance technology related, most of the traditional video surveillance systems still require human personnel to monitor camera footage, etc., which caused a great waste of resources, and with the field of surveillance technology to be applied more widely, intelligent surveillance systems are in greater demand.This paper will do in-depth study of the intelligence related to technology and image processing technology to achieve the monitoring system, Intelligent video surveillance system to monitor the area in the human face recognition, as well as to identify the face after face image acquisition scaled to address the video content analysis monitoring personnel to identify difficult issues as the focus of this study。the company developed around TI ’s new digital signal processing board TMS320DM6446designed a module based intelligent video image recognition technology, and based on this video module architecture to build an intelligent video surveillance system solutions.The system’s hardware will use TMS320DM6446as the core, to meet the demand for intelligent video surveillance systems, respectively, video capture module, data storage module, video backend processing module, video transmission module, switch control module, and power supply modules related hardware design. The objective of the system is to achieve the best balance between performance and cost.The software will adopt image recognition principle and scaling technology, the higher the rate of Adaboost algorithm to identify and Seam Carving techniques major software framework designed the system, and the key function implementation is analyzed. After the system through software simulation designed related video processing can achieve relatively good face recognition and image scaling effects.