Trellis quantization
The trellis quantization is an algorithm in the field of source coding to improve the compression of lossy image and video compression methods. Applications are, for example, the image compression JPEG 2000 and the video compression methods Xvid and x264 . In the case of video compression, the trellis quantization takes place after motion compensation by optimizing the coefficients of the discrete cosine transformation (DCT), in JPEG2000 by optimizing the coefficients of the wavelet transformation (DWT).
Trellis quantization reduces the value of some coefficients while restoring the value of others. This process can increase the quality, because the coefficients chosen by Trellis have the lowest rate-distortion ratio . Trellis quantization finds the optimal quantization for each block in order to maximize the peak signal-to-noise ratio (PSNR) relative to the bit rate . The efficiency depends on the input data and the compression method.
literature
- MW Marcellin, MA Lepley, A. Bilgin, TJ Flohr, TT Chinen, JH Kasner: An Overview of Quantization in JPEG-2000 . Ed .: Signal Processing: Image Communications. tape 17 , no. 1 , 2001, p. 73-84 .
Web links
- VirtualDub / Xvid guide mentioning Trellis quantization
- FFMPEGx option documentation
- Trellis explanation and pseudocode by the x264 author
Individual evidence
- ^ Jie Liang: Source Coding in Digital Communications: Trellis Coded Quantization. (No longer available online.) Simon Fraser University, 2013, archived from the original March 7, 2014 ; accessed on March 7, 2014 . Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice.
