Research Of Multimedia Record Identification System Based On Video Image

In recent years,computer vision has made great progress,video monitoring is to replace human monitoring an important monitoring method,it can complete video information collection,video playback,and query.Video monitoring has been widely applied in traffic management,industrial production,national defense security and so on,it can replace human do some simple repetitive work,but not to the collected images for automatic analysis and target recognition,the more complex operations require human intervention.And when the need of the large number of target monitoring,under the condition of rapid change,human monitoring and analysis has become very difficult.Now that ordinary video monitoring has many shortcomings,this paper puts forward intelligent multimedia record identification system based on video image,the system can be used to monitor and identify equipment on the change of light-emitting diodes(LED),voltage ammeter pointer and the digital thermometer,this system is known as multimedia record identification systemin this paper.According to the trigger signal,the system to collection and identification for the target of the device,this can effectively reduce the data error caused by human monitoring,and improve the work efficiency.The multimedia record recognition system consists of front-end acquisition identification and back-end server,among them,front-end acquisition identification including of light-emitting diodes identification and pointer and digital tube identification.The front-end part is to processing,analysis and recognition the collected image information,and then by establishing a local area network to send the results to the server,to query and analysis for operating personnel.In this paper,combining the characteristics of the unknown object,the multimedia record identification system to do the following research:(1)In order to achieve the test purpose,to design the overall structure of the system,and expound the corresponding recognition software design steps.(2)Because of system testing for time accuracy is higher,to analysis and research light-emitting diodes sequential detection algorithm and study the industrial computer through the API function to control the process of camera collecting data.And the noise reduction algorithm,position correction algorithm,the recognition algorithm are also studied.(3)According to the characteristics of the digital table and pointer table,analyzed the specific identification method and the technical difficulties.To introduce the necessary identify preprocessing steps in detail,and focuses on pointer table recognition based on angle and digital tube recognition method based onthreading method.