Constructing colour image databases using BTC for efficient storage and effective retrieval

We introduce a simple image coding method, the block truncation coding (BTC) technique, as a novel approach to the construction of colour image databases. It is shown that BTC can not only be used to compress images, thus achieving storage efficiency, but the BTC codes can also be used directly to construct image features for effective image retrieval. From the BTC code we have developed an image feature termed the BTC colour co-occurrence matrix (BCCM) as an effective measure of image contents. Experimental results are presented to show that BCCM is comparable to state of the art techniques, such as color correlogram, in image retrieval.