Regular Article
Design and Evaluation of an Entirely Psychovisual-Based Coding Scheme

https://doi.org/10.1006/jvci.2001.0489Get rights and content

Abstract

In this paper a new psychovisual-based coding scheme is proposed. The analysis and the quantization stages, the two main functions which determine the performances of a coding scheme, are based on the human visual system properties. In the first stage, a filter bank decomposes images into subimages of perceptual significance when a contrast transformation is applied. Analytic cortex filters have been used because they provide an accurate modelization of visual receptive fields. The choice of subbands lies on psychovisual experiments led in the laboratory. It was found that visual information is processed through 17 channels. In the second stage the use of the local band-limited contrast yields very interesting properties concerning the quantization. A scalar and vector quantization have been considered. In this latter case the vector's construction methodology preserves the main properties of the human visual system about perception of quantization impairments and takes into account the masking effect due to interaction between subbands with the same radial frequency but with different orientations. The vector's components are the local band limited contrasts Cij (m, n) defined as the ratio between the luminance Lij at the point (m, n), which belongs to the radial subband i and angular sector j and the average luminance at this location. Hence the vector's dimension depends on the orientation selectivity of the chosen decomposition. The low pass subband, which is nondirectional is scalar quantized. A methodology for automatic subsampling matrix design was also developed. The performance have been evaluated on a set of images in terms of peak SNR, true bit rates, and visual quality. For the latter, no impairments are visible at a distance of four times the height of the used high quality TV monitor. The SNRs are about 6 to 8 dB under the ones of classical subband image coding schemes when producing the same visual quality. Another particularity of this approach, due to the use of the local band limited contrast, lies in the structure of the reconstruction image error which is found to be highly correlated to the structure of the original image.

References (44)

  • P. Whittle

    Increments and decrements: luminance discrimination

    Vision Res.

    (1986)
  • F. Kingdom et al.

    A model for contrast discrimination with incremental and decremental test patches

    Vision Res.

    (1991)
  • W.B. Pennebaker et al.

    JPEG still image data compression standard

    (1993)
  • A.B. Watson

    DCTune: A technique for visual optimization of DCT quantization matrices for individual images

    Soc. Inform. Display Digest Technical Papers

    (1993)
  • A.B. Watson et al.

    Visibility of wavelet quantization noise

    IEEE Trans. Image Process.

    (1997)
  • R.J. Safranek et al.

    A perceptually tuned sub-band image coder with image dependent quantization and post quantization data compression

    Proc. IEEE ICASSP, Glasgow, Scotland

    (1989)
  • R.E. Van Dyck et al.

    Subband/VQ coding in perceptually uniform color spaces

    Proc. IEEE ICASSP, San Francisco, CA

    (1992)
  • D.J. Heeger et al.

    A computer model of human retinal visual processing: effect of compressive nonlinearity on spatial frequency filters

    Proc. Conf. on Pattern Recognition, Montreal, Canada.

    (1984)
  • F.L. Kooi et al.

    Spatial localisation across channels

    Vision Res.

    (1987)
  • R.M. Shapley et al.

    Spatial frequency analysis in the visual system

    Annual Review of Neuroscience

    (1985)
  • E.T. Davis et al.

    Shifts in perceived spatial frequency of low-contrast stimuli: data and theory

    J. Opt. Soc. Amer.

    (1986)
  • Cited by (7)

    • Retinal model-based visual perception: Applied for medical image processing

      2016, Biologically Inspired Cognitive Architectures
      Citation Excerpt :

      The author in his work developed tone mapping algorithm to produce sharper images, but the accuracy was not too good. Hafiz, Alnajjar, and Alnajjar (2010) proposed a model from mammalian retina based on dynamic mask inspired by the neuron connection of retina towards better robotic vision Senan, Saadane, and Barba (2001) developed an HVS model for information coding application. The study involved modeling of visual cortex that is subjected to the high-level information processing and did not concentrate on what happens at the retina level, i.e., low-level processing.

    View all citing articles on Scopus
    View full text