SINGLE BIT-MAP BLOCK TRUNCATION CODING OF COLOR IMAGES

This paper describes a color-image data compression technique based on block truncation coding (BTC). BTC is first implemented to encode a weighted trichromatic image. The bit map generated is then used to compress each of the R, G, and B signals separately, removing the intercolor correlations. The weighted trichromatic signal is generated by properly weighting R, G, and B signals so that the reconstructed R, G, and B signals have minimum mean square error (MSE). Another approach using a bit-wise majority rule on the R, G, and B BTC bit maps is also discussed. The algorithm preserves the spatial details in the image content and has low computational complexity. An 8:1 compression can be achieved. The truncated data is further compressed using adaptive DCT. Differential coding and error feedback techniques are also implemented to reduce the bit rate and improve the MSE performance. The total compression ratio is about 12:1. By exploiting human visual system characteristics, a 14:1 compression of NTSC color TV components can be achieved.