Halftoning-based Block Truncation Coding feature on scrambled images

This paper investigates the usability of Halftoning-based Block Truncation Coding (HBTC) feature for image retrieval. It assumes that all images in database are stored in scrambled/encrypted format. Firstly, an image feature descriptor is derived from the scrambled/encrypted image. This image feature is subsequently converted into the binary representation to achieve fast similarity measurement. The Hamming distance measures the similarity between the query and target images in the bitwise manner. As documented in experimental result, the proposed method gives a promising result on the scrambled/encrypted image retrieval. It demonstrates the superiority of HBTC feature in dealing with scrambled/encrypted image.