A common image classifier for classified still image coding scheme is designed based on Hausdoff-Measure
by which the image blocks can be grouped into four basic ciasses-Shade
Midrange
Edge and Mited. The introduction of fractal parameters improves the result of ciassification Meanwhile
under certain assumptions and with consideration of Visual Entropy and the characteristies of Human Visual System
two special categories
Key-point and Random-Texture
can be extracted from the fourth groupMixed
which makes possible optimal arrangement of image coding scheme and allocation of bitrate for coefficients in the meaning of subjective visual quality for decoded images.