Video indexing and retrieval for archeological digital library, CLIOH

Abstract

With the ever-increasing amount of digitally archived libraries that are being collected, new techniques are needed to organize and search these collections, retrieve the most relevant selections, and effectively reuse them. This helps a user find contents of interest in faster and more precise fashion than searching a single track. This paper introduced a video indexing and retrieval system for an archeological database, CLIOH (Cultural Digital Library Indexing our Heritage), using wavelet best basis and self-organizing neural networks. Texture similarity matching provides the functionality of video retrieval by comparing the Euclidean distance of encoded wavelet quadrature tree structures generated from probe texture icon and gallery texture icons. Experimental result using video sequences drawn from the CLIOH database proves the feasibility of our approach.