Skip to main navigation Skip to search Skip to main content

Intra-prediction algorithm of the H.264 standard for aerial area CCD camera

  • Ran Feng Zhang*
  • , Xian Peng Hao
  • , Zeng Ming Lv
  • , Shang Li Han
  • , Jin Li
  • *Corresponding author for this work
  • CAS - Changchun Institute of Optics Fine Mechanics and Physics
  • University of Chinese Academy of Sciences

Research output: Contribution to journalArticlepeer-review

Abstract

In order to improve the coding efficiency for intra-prediction in H.264, we propose a novel intra-prediction mode of the H.264 standard for aerial area CCD camera in this paper. The idea of the proposed intra-prediction mode is that each pixel within a block is predicted by a weighted sum of its neighbors, according to an N-th order linear predictor. The weights are obtained from reconstructed data in the neighbor blocks which are selected by l1-norm-based approach. Therefore, the decoder doesn't need to convey the weights to the decoder, which can be very time consuming. Finally, the proposed intra-prediction mode is implemented in H.264 reference software JM17.6. The experimental results show that the proposed intra prediction algorithm has better rate-distortion (RD) performance. Compared with the intra prediction for regular H.264 standard, the proposed method could improve the average peak signal to-noise ratio (PSNR) by 0.57-0.17 dB for images, 0.05-0.14 dB for CIF-resolution sequences, 0.01-0.05 dB for 720p50-resolution sequences, while the bit-reductions are 9.71-3.94%, 1.23%-1.52%, 0.25%-0.47%, respectively. It is suitable for the application in aerial area CCD camera.

Original languageEnglish
Pages (from-to)1558-1564
Number of pages7
JournalGuangdianzi Jiguang/Journal of Optoelectronics Laser
Volume25
Issue number8
StatePublished - 15 Aug 2014
Externally publishedYes

Keywords

  • Area CCD camera
  • H.264 compression
  • Least-squares prediction
  • Video compression

Fingerprint

Dive into the research topics of 'Intra-prediction algorithm of the H.264 standard for aerial area CCD camera'. Together they form a unique fingerprint.

Cite this