Multiple Description Based on Fractal

抄録

In this paper, I design and develop a multiple description (MD) method which is based on fractal image coding procedure. In the encoder of MD, the IFS generated mappings are separated into different parts and encoded into different descriptions so that, on each description, a subset of these mappings can be involved. Meanwhile, a desired amount of redundancy is inserted into each description such that a satisfactory reconstruction quality will be ensured. In MD decoder, the redundancy and the mappings in one description are exploited to recover the missed mappings in the other description when only one description is received. Compared with the referenced methods, the the proposed MD coder can achieve better and more robust performance under various packet loss ratio circumstance.