Face Image Retrieval from Sparse Codeword's

Vandana Shinde, S.R.Durugkar

Abstract: The sizes of digital image collections have been changing since its progression. These images are increasing in the size and consists of human faces majority of times. Nowadays photos with people are becoming an area of interest for many users. Thus, many real time applications work on image retrieval techniques. These content-based image retrieval techniques work on a large database containing photos. The goal of the proposed system is to detect the human attributes automatically by utilizing semantic clues of face photos. This results into improved content-based image retrieval. This is done with the help of semantic codewords. We described two orthogonal approaches named as attribute-embedded inverted indexing and attribute-enhanced sparse coding for improved face retrieval in both offline and online modes.

Keywords: Face image, human attributes, content-based image retrieval, sparse coding, attribute embedded inverted indexing