Image compression using the Anamorphic Stretch Transform

A new method for image compression is introduced. It uses a new physics-based mathematical transformation that increases spatial coherency by causing feature-selective stretch. We show that by reshaping the image with a special class of transfer functions, more samples are allocated to sharp features, where they are needed, and less to coarse features, where they are redundant. The method achieves image space-bandwidth compression without feature detection or prior knowledge of the image.