Research And Implementation Of Billboard Recognition Algorithm Based On Smart Phone

Nowadays,outdoor advertising media,with its unique geographical location,has played a positive role in enterprise products,social image and the propaganda of culture.Under the influence of rising Internet multimedia,how to attract the attention of public concerns outdoor media industry.This thesis aims to design and implement a recognition system of enterprise billboard image,which is based on image recognition algorithm.According to the practical application,this paper puts forward and implements the algorithm flow.Based on this algorithm,a sign recognition system is set up.Firstly,thesis introduces the related technology and the basic flow of image recognition system,and proposes the design of billboard recognition system according to the user requirements.Secondly,in order to select the most appropriate preprocessing algorithms,thesis compares processing efficiency of noise with average filter,median filter and gauss filter respectively.Thirdly,by researching the properties of ORB algorithm,SIFT algorithm,SURF algorithm respectively and proposing the sample positioning method,thesis formulate a accurate recognition scenario for the billboard recognition system;Finally,basing on the architecture of C/S mode,the server of recognition system is deployed Spark cluster system and the client is deployed on iPhone.By a great deal of practical experience,this system could adapt to the complex background of actual application and response the request response within the allotted time.