Perceptual Quality Study on Deep Learning Based Image Compression

Cheng, Zhengxue; Akyazi, Pinar; Sun, Heming; Katto, Jiro; Ebrahimi, Touradj

doi:10.1109/icip.2019.8803824

Cited by 31 publications

(25 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Comparison of six different image compression algorithms with a subjective quality test using high-resolution images shows that the learned compression optimized by MS-SSIM gives competitive results close to the efficiency of state-ofthe-art compression (Cheng, Akyazi, Sun, Katto, & Ebrahimi, 2019).…”

Section: A Mean Square Error (Mse)mentioning

confidence: 91%

Strategies in JPEG compression using Convolutional Neural Network (CNN)

Suman¹

2021

Preprint

View full text Add to dashboard Cite

Interests in digital image processing are growing enormously in recent decades. As a result, different data compression techniques have been proposed which are concerned mostly in minimization of information used for the representation of images. With the advances of deep neural networks, image compression can be achieved to a higher degree. This paper describes an overview of JPEG Compression, Discrete Fourier Transform (DFT), Convolutional Neural Network (CNN), quality metrics to measure the performance of image compression, and discuss the advancement of deep learning for image compression mostly focused on JPEG, and suggests that adaptation of model improve the compression.

show abstract

Section: A Mean Square Error (Mse)mentioning

confidence: 91%

Strategies in JPEG compression using Convolutional Neural Network (CNN)

Suman¹

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…When the value in it is greater than one, it states MS-SSIM optimized end-to-end compression model allocates more bits there than the MSE optimized one and vice versa. Considering MS-SSIM optimized image compression has much better perceptual quality than MSE optimized compression [12], adapting bits according to the importance map can probably lead to perceptual quality enhancement.…”

Section: Importance Map Generationmentioning

confidence: 99%

Knowledge Distillation From End-To-End Image Compression To Vvc Intra Coding For Perceptual Quality Enhancement

Yang

Liu

et al. 2021

2021 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

In the current hybrid coding schemes, mean-squared-error is widely used for the rate-distortion optimization, which leads to high peak signal-to-noise ratio but sub-optimal perceptual quality. Although human perception-related measures, like multi-scale structural similarity (MS-SSIM), have been proposed, plugging them into the hybrid coding schemes may be computationally expensive. Recently, end-to-end optimized image compression has demonstrated the advantage of perceptual quality-oriented optimization by simply changing the training loss function. Inspired by this, we propose to distill the "perceptual" knowledge from end-to-end image compression and use the knowledge to enhance the perceptual quality for Versatile Video Coding (VVC) intra coding. For an input image, we obtain the block-level bit allocation via end-to-end image compression, and use the bit allocation to adjust the quantization parameter of VVC intra coding. Being compatible to the VVC standard, our method achieves on average 9.32% BD-rate reduction on the Kodak image set when evaluated by MS-SSIM, compared to the VVC reference software.

show abstract

“…Enormous improvements in this field have been observed in the latest years through several contributions [5,6,7,8]. Quality analyses of learning-based image compression approaches are already available in the literature [9,10]. However, they lack an execution time analysis of the solutions, which is essential to understand the applicability of these novel methods.…”

Section: Introductionmentioning

confidence: 99%

Quality and Complexity Assessment of Learning-Based Image Compression Solutions

Dick¹,

Abreu

Grellert

et al. 2021

2021 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

This work presents an analysis of state-of-the-art learningbased image compression techniques. We compare 8 models available in the Tensorflow Compression package in terms of visual quality metrics and processing time, using the KO-DAK data set. The results are compared with the Better Portable Graphics (BPG) and the JPEG2000 codecs. Results show that JPEG2000 has the lowest execution times compared with the fastest learning-based model, with a speedup of 1.46× in compression and 30× in decompression. However, the learning-based models achieved improvements over JPEG2000 in terms of quality, specially for lower bitrates. Our findings also show that BPG is more efficient in terms of PSNR, but the learning models are better for other quality metrics, and sometimes even faster. The results indicate that learning-based techniques are promising solutions towards a future mainstream compression method.

show abstract

Perceptual Quality Study on Deep Learning Based Image Compression

Cited by 31 publications

References 16 publications

Strategies in JPEG compression using Convolutional Neural Network (CNN)

Strategies in JPEG compression using Convolutional Neural Network (CNN)

Knowledge Distillation From End-To-End Image Compression To Vvc Intra Coding For Perceptual Quality Enhancement

Quality and Complexity Assessment of Learning-Based Image Compression Solutions

Contact Info

Product

Resources

About