engleski

Lossless Predictive Algorithm for Medical Image Compression

A new lossless image compression algorithm is proposed. Algorithm is based on a highly adaptive predictor that uses blending context obtained using simple classification technique. This allows modeling of more complex image structures that occur often in medical images such as non--trivially oriented edges and the periodicity and coarseness of textures. The prediction step is followed by contextual error modeling which removes remaining statistical redundancies in the residual image data. Finally, error image is entropy coded. Results obtained with proposed compression method on a large set of medical images are very encouraging, beating several well known lossless compression methods. Algorithm can be efficiently applied for natural images as well.

medical image compression; predictive models

nije evidentirano