| The research subject of this paper mainly came from the commercial project ofNASDAQ:YY: Intelligent Mobile Album. That solution has been proposed for this projectin this paper, using the core technology and architecture of the enterprise on the basis ofthe web album named Cloud Image, which combines techniques of Android development,face detection and classification, image classification. Meanwhile, it has been designed andimplemented an Android-based intelligent mobile album in this paper. The system providesfor Android mobile users with services such as recording baby growth, organizing photosby one key, online synchronization backup and sharing photos to social networks, which isan integrated application for managing Android mobile phone photos, which have a varietyof capabilities, such as baby face detection, classification by geographical information,cloud storage and quick sharing.The main work is in the following areas:(1)That is designed and implemented the function of extracting baby photosautomatically in user’s mobile album, using face detection and classification technologies,by the method detecting the containing baby photos from the image set, which helpsparents to make exclusive growth albums for their babies.(2)That is implemented the function of automatically extracting and classifying users’photos containing GPS information, using image classification technologies, which helpsusers to set up live tour album.(3)That is designed and implemented for Android-based Intelligent Mobile Album.And a good framework and interface has been designed to realize multi-functional smartphones album, which are realized of network photos backup, quick sharing photos to socialnetworking set, quick cleaning up duplicate, fuzzy photos and web browsing.The main contribution of this paper is to improve a number of existing technologiesand related items:(1)That is proposed the method using integratedly AdaBoost face detection, C1-Sfeature extraction and SVM classification to realize baby face detection, which is realizedof the accurate extraction of baby photos from complex image set.(2)That method has been proposed in this paper is scaling and grid-type distributionof geographic areas, which can be used to divide the life and travel photographsself-adaptable by extracting the GPS geographic information from them, so that the photoswhich are taken nearby could be collected in one album, and each photo album set ownsrelatively average number of photos. Currently, the function of each module of Intelligent Mobile Album system has beenimplemented, and the system can be downloaded on line in the Android Market. Everyonecan quickly and easily use the system for managing photos with multifunction on theAndroid client. An initial user study confirms that the developed Android application isvery useful in practice. Effective code of the system is about13,000lines. |
PDF Full Text Request