Query by sketch: An asymmetric sketch-vs-image retrieval system

This paper presents a novel query method in sketch-based image searching system. Semantically similar images can be retrieved via user-drawn sketches in large image collections. Firstly the system analyses the spatial distribution of user's sketch and computes the weight map. And then the system extracts features from the query sketch. The edge map descriptor and haar wavelet transform feature are used, which can match user-drawn sketch with the target image efficiently. To substantiate our approach, we build a retrieval system and allow the user to express a query as a simple sketch portraying ‘what’ she/he is looking for. We also show how our method can be used for the asymmetric sketch-vs-image retrieval efficiently.