Design And Implementation Of TV Station Logo Recognition System Based On Smart TV

With the rapid development of science and technology,smart TV has become more and more popular.TV station logo as the main symbol of the TV channel,not only contains the TV program channel information,but also contains the program content information and program source information.If we can identify the TV station when TV users watching TV,not only can the content of the TV broadcast be monitored,but also the TV ratings can be effective counted.In addition,TV station logo identification technology enable users search channel more quickly.Existing logo recognition system,are collecting images in the TV side and recognizing images on the server.This approach may not work in real time because of the network transmission costs.If there are a large number of pictures need to be identified at the same time on the server,it will bring huge computational pressure to the server,affecting the performance of the server.In addition,the existing logo recognition system,often only to achieve the recognition function,and have not the statistical function of the recognition result.The platform identification system realized in this paper not only realizes the TV station logo identification on the smart TV,solves the real-time problem of the existing methods,but also realizes the real-time statistical analysis of the recognition results.The system implemented in this paper consists of two subsystems,one is the logo recognition subsystem and the other is the management platform subsystem.The logo recognition subsystem is C / S structure,its client is integrated smart TV with identification service,the main function is to identify the logo and the identification results passed to the server for data statistics.In this paper,the algorithm is a kind of image recognition algorithm based on supervised learning.SIFT feature extraction technology,word bag algorithm and SVM linear classifier technology are used to realize logo recognition.There are two main functions of the server.A function is to process the recognition results in real time and generate statistics.Another function is to send the configuration information of the identification service to the television terminal.The management platform subsystem is B/S structure,using Spring MVC framework technology and Ajax technology to achieve.The data managed by the management platform includes user-related data,terminal-related data,and statistical analysis-related data.The browser side of the management platform subsystem is used to provide an interface for displaying various data and for the user to modify,view and add the data.The server of the management platform subsystem implements the functions related to data management,including data query,data cascade deletion,data modification and other functions.In addition,the server also implements a number of functions for the protection of the system security,including logging,user authentication and other functions.After testing,the system implemented in this article,the various functional modules can work properly.The recognition accuracy of the algorithm is 81%,and the misrecognition rate is less than 1%.The data statistics module realizes the real-time and correct statistics of the recognition results,which is greatly improved compared with the traditional method of sampling statistics,correctness and real-time.In addition,the management platform of the various functional modules have been tested can work properly,so the background maintenance staff can easily manage system data.