A perceptually-tuned block-transform-based progressive transmission image coder

Abstract

A perceptually-tuned block-transform-based progressive transmission image coder is presented. The novel embedded coder features a locally adaptive perceptual masking thresholder that computes, based on the contents of the original image, the maximum amount of noise energy that can be injected at each transform coefficient that results in perceptually distortion-free still images or sequences of images. The adaptive thresholder is used as a pre-processor to a block-based embedded image coder. Perceptually normalized block-transform coefficients less than their corresponding masking thresholds can be set to zero before the normal embedded quantization and bit plane coding step. The result is a visually-tuned embedded coder which is capable of transparent coding when all bit planes are encoded. Visually insignificant coefficients are never included in the compressed bitstream. If lower bit rates are desired, the decoder and possibly the encoder simply deal with the higher bit planes of the visually significant coefficients. Excellent reconstructed image quality without annoying blocking and ringing artifacts can still be obtained at low bit rates.