An object processing unit divides each image frame of received image data into block areas, extracts an object from each image frame, and extracts, as metadata, the features of this object for each block area. A real-time detection processing unit compares the metadata extracted for each block area with a received real-time detection query, and detects an object which satisfies the real-time detection query. A post-retrieval processing unit compares the metadata for each block area stored in a metadata storage unit with a received post-retrieval query, and retrieves an object which satisfies the post-retrieval query.