Corpus ID: 204801243

ProxIQA: A Proxy Approach to Perceptual Optimization of Learned Image Compression

@article{Chen2019ProxIQAAP,
  title={ProxIQA: A Proxy Approach to Perceptual Optimization of Learned Image Compression},
  author={Liheng Chen and Christos George Bampis and Zhi Li and Andrey Norkin and Alan C. Bovik},
  journal={ArXiv},
  year={2019},
  volume={abs/1910.08845}
}
  • Liheng Chen, Christos George Bampis, +2 authors Alan C. Bovik
  • Published in ArXiv 2019
  • Computer Science, Engineering
  • The use of $\ell_p$ $(p=1,2)$ norms has largely dominated the measurement of loss in neural networks due to their simplicity and analytical properties. However, when used to assess the loss of visual information, these simple norms are not very consistent with human perception. Here, we describe a different "proximal" approach to optimize image analysis networks against quantitative perceptual models. Specifically, we construct a proxy network, broadly termed ProxIQA, which mimics the… CONTINUE READING

    Create an AI-powered research feed to stay up to date with new papers like this posted to ArXiv

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 54 REFERENCES

    Variational image compression with a scale hyperprior

    VIEW 5 EXCERPTS
    HIGHLY INFLUENTIAL

    End-to-end Optimized Image Compression

    VIEW 12 EXCERPTS
    HIGHLY INFLUENTIAL

    Learning Image and Video Compression Through Spatial-Temporal Energy Compaction

    VIEW 1 EXCERPT

    Efficient Nonlinear Transforms for Lossy Image Compression

    • Johannes Ballé
    • Computer Science, Engineering
    • 2018 Picture Coding Symposium (PCS)
    • 2018
    VIEW 1 EXCERPT

    Generative Adversarial Networks for Extreme Learned Image Compression

    VIEW 1 EXCERPT