| Budget Amount *help |
¥2,000,000 (Direct Cost: ¥2,000,000)
Fiscal Year 1996: ¥400,000 (Direct Cost: ¥400,000)
Fiscal Year 1995: ¥1,600,000 (Direct Cost: ¥1,600,000)
|
|---|
| Research Abstract |
In this study, a highly efficient loss-less image coding scheme is developed. Image coding is basically separated into two blocks. One is the model part and another is the entropy encoder. First, we proposed a three dimensional Markov model coding scheme as the model part of the loss-less image coding. In ourmodel, correlations between bit-planes are utilized. The computer simulation is performed for the images of 256 levels using the three dimensional Markov model of 12 reference elements and we can get results that this method can reduce the entropy by 8 percent at its maximum compared with the conventional Markov model of 1 reference pixel. Next, we proposed two entropy coding methods. the first one is concerned with the arithmetic coding. In our coding scheme, any augend state is transited into a same state independent of symbol values, by estimating probability of LPS to 1/ (1+2^Q). In this method, augend data are only calculated by shift operation using the prepared augend table. Simulation experiments from points of view in coding efficiency and the number of coding operations are performed using binary images. The second one is concerned with Variable-to-Fixed code (VF code), that is high speed decodable code. We revealed that the VF code is more efficient than the conventional Huffman code under some condition. By using above mentioned model and entropy coding scheme, the highly efficient loss-less image coding scheme is described.
|
