A system for distributed image acquisition, content-analysis and similarity retrieval

Emergence of mobile communications devices capable of acquiring high-resolution images and sensor devices with communication capabilities contributes to the increasing feasibility of remote screening of various medical and environmental conditions. In this work we present a system that supports remote acquisition of images and similarity retrieval from a centralized database after automatic content analysis. The system supports remote image acquisition by a device such as a smart phone, wireless PDA or digital camera, and communication via the Internet. While the system architecture is generic to accommodate various uses, we have implemented a prototype for the remote screening of skin lesion images. Accordingly, the system incorporates content analysis modules such as a segmentation module, shape analyzers and a shape-feature based similarity computation module. The whole process is automated: clients use a Web interface to submit a query image, which goes through an automated analysis procedure and produces an interactive report containing similar images and a risk analysis report. Immediate application areas include distributed detection and monitoring of environmental hazards, recognition of biochemical security threats and medical conditions.