pSQ - Emerging Trends in Image Processing, Computer Vision, and Pattern Recognition - page 59

No. 204151/2013, and LABEX Σ-LIM France, Coimbra Group Scholarship Programme granted
by University of Poitiers and Region of Poitou-Charentes, France.
1 Introduction
Digital image compression has been a research topic for many years and a number of image
compression standards have been created for different applications [ 1 , 2 ]. The JPEG2000 is in-
tended to provide rate-distortion and subjective image quality performance superior to exist-
ing standards, as well as to supply functionality [ 3 ]. However, JPEG2000 does not provide the
most relevant characteristics of the human visual system, since for removing information in
order to compress the image mainly information theory criteria are applied. This information
removal introduces artifacts to the image that are visible at high compression rates, because
of many pixels with high perceptual significance have been discarded. Hence, an advanced
model is necessary that removes information according to perceptual criteria, preserving the
pixels with high perceptual relevance regardless of the numerical information. The Chromatic
Induction Wavelet Model presents some perceptual concepts that can be suitable for it. Both
contrast band-pass filtering (CBPF) and JPEG2000 use wavelet transform. CBPF uses it in for
der to generate an approximation to how every pixel is perceived from a certain distance tak-
ing into account the value of its neighboring pixels. By contrast, JPEG2000 applies a perceptual
criterion for all coefficients in a certain spatial frequency independently of the values of its sur-
rounding ones. In other words, JPEG2000 performs a global transformation of wavelet coef-
icients, while CBPF performs a local one. CBPF atenuates the details that the human visual
system is not able to perceive, enhances those that are perceptually relevant, and produces an
approximation of the image that the brain visual cortex perceives. At long distances, the lack
of information does not produce the well-known compression artifacts; rather it is presented
as a softened version, where the details with high perceptual value remain (e.g., some edges).
2 Related Work: JPEG 2000 Global Visual Frequency
Weighting
In JPEG2000, only one set of weights is chosen and applied to wavelet coefficients according
to a particular viewing condition (100, 200, or 400 dpi's) with fixed visual weighting [ 3 , Annex
J.8]. This viewing condition may be truncated depending on the stages of embedding, in other
words at low bit rates, the quality of the compressed image is poor and the detailed features of
the image are not available since at a relatively large distance the low frequencies are percep-
tually more important. Table 1 specifies a set of weights which was designed for the luminance
component based on the contrast sensitivity function (CSF) value at the mid-frequency of each
spatial frequency. The viewing distance is supposed to be 4000 pixels, corresponding to 10 in.
for 400 dpi print or display. The weight for LL is not included in the table, because it is always
1. Levels 1, 2, …, 5 denote the spatial frequency levels in low- to high-frequency order with
three spatial orientations, horizontal, vertical , and diagonal .