Using multiple examples for content-based image retrieval

Query specification for content based image retrieval is typically accomplished through query-by-example paradigms, such as query-by-image and query-by-sketch. In some cases query-by-sketch can be difficult because of lack of sketching abilities, difficulty to detect distinguishing image features. Therefore, querying is performed through the query-by-image paradigm. However, this paradigm often fails since a single sample image rarely includes all and only the characterizing elements the user is looking for. The paper presents a system that supports query-by-image using multiple image examples, both positive and negative. The system also enables editing of examples so as to disregard those image features that are not relevant to the query.