ABSTRACT
Similar to this paper, some papers use edge information as feature for image retrieval. In [10] (Zhou et al. 2001), some information embedded in the structure of edges is extracted. In [7] (Banerjee et al. 2003), compactness of edge for image retrieval is used. In [3] (Ohashi et al. 2003), the edge based feature extraction method functioned by representing the relative position relationship between edge pixels. However, the extracted edge features in [10] (Zhou et al. 2001), [7] (Banerjee et al. 2003), [3] (Ohashi et al. 2003) do not contain the edge spatial position information which relative to key points, and they are not effective for image retrieval.
