Research Of Android Cotent-Based Image Retrieval System

With the development of smart phone, user’s need of the mobile is no longer satisfied with making a call or sending a message. By using smart phone’s photographed function and strong calculation processing ability, content-based image retrieval system in smart phone has broad prospects for application, such as commodity picture retrieval.This paper realized the Android content-based image retrieval system. The system uses CEDD feature descriptor. CEDD keeps a good balance between feature’s size and retrieval quality. This balance ensures the image retrieval system both efficiency and quality.The system uses OpenCV as the image development tools in Android. This thesis analyses two different way to use OpenCV library in Android. The system ultimately select OpenCV Android API for the development of methods, realized the extracting of CEDD feature. To retrieval the images, the system also implemented Tanimoto coefficient, Cosine similarity and Manhattan distance as feature matching method.This thesis introduces the implementation process of Android image retrieval system. The system consists of five modules, image input module, image pre-processing module, feature extraction module, feature matching module and result show module. User can take a picture or load one in SD card to retrieval image in the system. The system will gives five retrieval results as output.Issue tests and analyzes of the performance of the system. Result shows that the system has a good performance.