Design And Implementation Of A Image Classification Tagging System Based On Cloud Environment

With the development of the Internet, and the availability of image capturingdevices, the size of multi media collection is increasing rapidly. Content-based Imageand video retrieval become an active subject of research.This topic is based on the ground of the project of “The internet application ofimage(video) classification based on cloud computing”, which is corporation project ofChina Telecom Research Guangzhou and Information Processing Laboratory. Thepurpose of this project is to develop the application of image classification on internet,combing with cloud computing. The current paradigm of image classification consistsof manually collecting a large training set of good exemplars of the desired objectcategory; training a classifier on them and then evaluating it on novel images, possiblyof a more challenging nature. The size and quality of training set can heavily influencethe classification accuracy.We present those solutions to conquer the problem mentionedabove: first, we develop the platform based on cloud computing to crawl the raw dataimage from WWW; second, we develop a training image extractor based on pLSA(Probabilistic Latent Semantic Analysis), help users to get training set imageseffectively. At last, we integrate the SVM model to our system, which can be used toclassify or annotate novel images. According to our experiment, our system achieves theacquirement of the project..