Optimized Pre-Compensating Compression

Yehuda Dar, Michael Elad, Alfred M. Bruckstein

Abstract

This paper presents a novel pre-processing framework designed to pre-compensate for display-induced artifacts, such as LCD motion blur, during the image and video compression process. By integrating display degradation models directly into the rate-distortion optimization via the Alternating Direction Method of Multipliers (ADMM), the proposed method enables standard codecs like HEVC to produce bitstreams that yield significantly higher end-to-end visual quality. Experimental results demonstrate substantial gains, with PSNR improvements of up to 15.42 dB for still images and 13.90 dB for video sequences at high bit-rates, while maintaining full compatibility with standard decoders.

Introduction

In traditional imaging pipelines, compression (e.g., JPEG or HEVC) is treated as a standalone task focused on minimizing the difference between the original source and the decompressed output. However, this approach ignores the final stage of the journey: the display device. Modern LCD screens often introduce "motion blur" due to their hold-type mechanism, where frames remain static until the next update.

The end-to-end imaging pipeline, highlighting the post-decompression stage where display-induced degradation occurs.

Standard compression does not account for these display-side artifacts. Simply sharpening an image before compression is often ineffective, as the compression process tends to remove the added high-frequency details. This research proposes a way to integrate knowledge of display degradation directly into the compression process, creating a "pre-compensated" bitstream.

Methodology

The authors model display degradation as a known linear operator $H$. The objective is to find a compressed representation $v$ such that the degraded version $Hv$ is as close as possible to the original input $x$, subject to bit-rate constraints.

To solve this optimization problem, the framework utilizes the Alternating Direction Method of Multipliers (ADMM). This iterative approach splits the problem into two distinct steps:

  1. Compression Step: A standard codec (e.g., HEVC) is applied to an auxiliary variable. This ensures the method is "plug-and-play" with existing standards.
  2. Deconvolution Step: A quadratic optimization that sharpens the image to counteract display blur while maintaining proximity to the compression output.

This process shifts the computational complexity to the encoder. The resulting bitstream remains standard-compliant, allowing decoders (such as those in smartphones or TVs) to remain simple and power-efficient.

Experimental Results

The approach was evaluated using standard benchmarks, including 18 images from TESTIMAGES, UCID, and BSDS500, as well as high-frame-rate video sequences ('Shields' and 'Stockholm') featuring horizontal motion.

PSNR performance vs bit rate for the 'Cards' dataset, comparing the proposed method against standard and deblurring baselines.

Key findings include:

  • Performance Gains: For still images with Gaussian blur, the method outperformed standard HEVC by up to 15.42 dB PSNR at high bit-rates.
  • Video Quality: On the 'Shields' sequence, a BD-PSNR gain of 13.90 dB was achieved over regular HEVC.
  • Synergy with High Bit-rates: The method is particularly effective when bandwidth is sufficient to preserve the delicate pre-compensation signals. At lower bit-rates, coarse quantization limits the effectiveness of the sharpening.
  • Implicit Regularization: Interestingly, the compression step acts as a natural regularizer, preventing common deconvolution artifacts like ringing or excessive noise.

Conclusion and Future Work

The proposed ADMM-based framework successfully integrates display characteristics into the compression pipeline, offering massive quality improvements for high-end viewing experiences. While the current model assumes a linear degradation operator and requires knowledge of the target display, it paves the way for display-aware encoding. Future research may explore non-linear display models and techniques to optimize for a broader range of display devices simultaneously.

Summary updated

Abstract

In imaging systems, following acquisition, an image/video is transmitted or stored and eventually presented to human observers using different and often imperfect display devices. While the resulting quality of the output image may severely be affected by the display, this degradation is usually ignored in the preceding compression. In this paper we model the sub-optimality of the display device as a known degradation operator applied on the decompressed image/video. We assume the use of a standard compression path, and augment it with a suitable pre-processing procedure, providing a compressed signal intended to compensate the degradation without any post-filtering. Our approach originates from an intricate rate-distortion problem, optimizing the modifications to the input image/video for reaching best end-to-end performance. We address this seemingly computationally intractable problem using the alternating direction method of multipliers (ADMM) approach, leading to a procedure in which a standard compression technique is iteratively applied. We demonstrate the proposed method for adjusting HEVC image/video compression to compensate post-decompression visual effects due to a common type of displays. Particularly, we use our method to reduce motion-blur perceived while viewing video on LCD devices. The experiments establish our method as a leading approach for preprocessing high bit-rate compression to counterbalance a post-decompression degradation.
BibTeX
@article{dar2017optimized,
  author = {Yehuda Dar and Michael Elad and Alfred M. Bruckstein},
  title = {{Optimized Pre-Compensating Compression}},
  journal = {IEEE Trans. Image Process.},
  year = {2017},
  doi = {10.1109/tip.2018.2845125},
}
Figure