Perceptually lossless image compression

This paper presents an algorithm for perceptually lossless image compression. A compressed image is said to be perceptually lossless for a specified viewing distance if the reconstructed image and the original image appear identical to human observers when viewed from the specified distance. Our approach utilizes properties of the human visual system in the form of a perceptual threshold function model to determine the amount of distortion that can be introduced at each location of the image. Constraining all quantization errors to be below the perceptual threshold function results in perceptually lossless image compression. The compression system employs a modified form of the embedded zerotree wavelet coding algorithm to limit the quantization errors below the levels specified by the model of the threshold function. Experimental results demonstrate perceptually lossless compression of monochrome images at bit rates ranging from 0.4 to 1.2 per pixel at a viewing distance of six times the image height. These results were obtained using a simple, empirical model of the perceptual threshold function which included threshold elevations for the local brightness and local energy in neighboring frequency bands.