Entropy based adaptive image compression procedure

Abstract

There are several different methods to compress digital signals, datasets or images, each of them with special characteristics. The procedure introduced here is an entropy based adaptive compression method, its main feature is the hierarchical structure and entropy based compression. Wavelet transform of a dataset could be arranged to a binary tree like structure, which is the starting point of the proposed compression method. Degree of complexity of an image varies, different parts have different information content. Detailed regions require more thorough description than relaxed, predictable parts. To accomplish this task a new type of variable is introduced, resembling the entropy of a particular area of an image. The higher the value of this variable indicating entropy, the higher is the importance of the area's detail coefficients.