High-fidelity coding for super high-definition images
The super high‐definition images (SHDI) expected to be prominent in the systems of the 21st century consist of approximately 4000 scan lines and have extremely high fidelity. This paper describes a coding scheme for still SHDI. Predictive coding is thought to be a coding technique capable of not spo...
Saved in:
| Published in: | Electronics & communications in Japan. Part 1, Communications Vol. 80; no. 3; pp. 47 - 57 |
|---|---|
| Main Authors: | , |
| Format: | Journal Article |
| Language: | English |
| Published: |
New York
Wiley Subscription Services, Inc., A Wiley Company
01.03.1997
|
| Subjects: | |
| ISSN: | 8756-6621, 1520-6424 |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The super high‐definition images (SHDI) expected to be prominent in the systems of the 21st century consist of approximately 4000 scan lines and have extremely high fidelity. This paper describes a coding scheme for still SHDI. Predictive coding is thought to be a coding technique capable of not spoiling the quality of SHDI images. Thus, an extrapolative/interpolative prediction scheme that has been proposed [1] for high‐efficiency coding of HDTV is used for intraframe coding of SHDI. The extrapolative prediction error is coded with a first‐order Huffman code. Comparatively, the prediction efficiency of interpolative prediction is superior to that of extrapolative prediction [2]. Furthermore, since the pixel correlations of the images used in this paper are extremely high [9], the probability of the 0 representative level of the quantizer is high. Thus, in this paper, the prediction error is converted to binary levels when interpolative prediction is used and it is proposed that a coding method based on a type of arithmetic code, the QM‐Coder [4], which is used in JPEG, be used to code the resulting binary information. When binary information is coded, the resulting conditional entropy of the reference pixels is shown to be rather constant; and, thus, the number of states can be reduced without an increase in the rate. Image coding simulation results show that the coding performance is improved with this method where extrapolative/interpolative prediction is used by 0.350 ∼ 0.458 bits/pel when compared with the case that uses a first‐order Huffman code. Accordingly, high‐fidelity images whose signal‐to‐noise ratio (SNR) is greater than 50 dB with rates less than at 1.653 bits/pel are obtained. © 1997 Scripta Technica, Inc. Electron Comm Jpn Pt 1, 80(3): 47–57, 1997 |
|---|---|
| Bibliography: | istex:1D6AFD6F71D1AAC8E6E8AD4B991DC3B68B5D334C ark:/67375/WNG-JZH4GG5C-2 ArticleID:ECJA5 |
| ISSN: | 8756-6621 1520-6424 |
| DOI: | 10.1002/(SICI)1520-6424(199703)80:3<47::AID-ECJA5>3.0.CO;2-6 |