Perceptual Axis Alignment Outperforms Statistical Decorrelation for Colour Transform Design in Image Compression

Read full text
Authors: Juan Manuel Martinez Castro
Cited by: 0

Abstract

"We compare the NTSC YIQ transform (1953) and the YCoCg transform (Malvar et al., 2008) for JPEG image compression on the 24-image Kodak benchmark — the same dataset from which YCoCg was derived. Using four metrics from PSNR to S-CIELAB, we show that the ranking between transforms reverses as metrics become more perceptually accurate. Under equalised quantisation at Q75, PSNR favours YCoCg; S-CIELAB favours YIQ by 14% on all 24 images. Under standard JPEG chroma quantisation with 4:2:0 subsampling, S-CIELAB favours YIQ by 30% on 23/24 images. The more accurately a metric models human colour vision, the larger the advantage of perceptual axis alignment over statistical decorrelation."

Fields of Study & Subjects

Advanced Data Compression Techniques Image and Video Quality Assessment Video Coding and Compression Technologies

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
15 March 2026
Language
EN
Document type
Preprint