US 2022/0101564 A1 · COMPRESSING DIGITAL IMAGES UTILIZING DEEP LEARNING-BASED PERCEPTUAL SIMILARITY

Abstract

Methods, systems, and non-transitory computer readable storage media are disclosed for utilizing deep learning to intelligently determine compression settings for compressing a digital image. For instance, the disclosed system utilizes a neural network to generate predicted perceptual quality values for compression settings on a compression quality scale. The disclosed system fits the predicted compression distortions to a perceptual distortion characteristic curve for interpolating predicted perceptual quality values across the compression settings on the compression quality scale. Additionally, the disclosed system then performs a search over the predicted perceptual quality values for the compression settings along the compression quality scale to select a compression setting based on a perceptual quality threshold. The disclosed system generates a compressed digital image according to compression parameters for the selected compression setting.

Claims20

Dependent claims

A non-transitory computer readable storage medium comprising instructions that, when executed by at least one processor, cause a computing device to:

generate, utilizing a neural network given an uncompressed digital image as input, a plurality of points indicating predicted compression distortions for a plurality of possible compressed versions of the uncompressed digital image;
determine a perceptual distortion characteristic curve for the uncompressed digital image from the plurality of points generated by the neural network;
select, based on the perceptual distortion characteristic curve, a compression setting that meets a perceptual quality threshold;
and
generate a compressed digital image utilizing the selected compression setting that meets the perceptual quality threshold.

2 more elements

A system comprising:

at least one computer memory device comprising an uncompressed digital image;
and
one or more servers configured to cause the system to:
generate, utilizing a neural network, a plurality of points indicating predicted compression distortions for a plurality of possible compressed versions of the uncompressed digital image by generating a predicted perceptual similarity value for each of the plurality of possible compressed versions of the uncompressed digital image;
determine a perceptual distortion characteristic curve for the uncompressed digital image from the plurality of points generated by the neural network by fitting the plurality of points to a curve across a compression quality scale;
select, based on the perceptual distortion characteristic curve, a compression setting that meets a perceptual quality threshold;
and
generate a compressed digital image utilizing the selected compression setting that meets the perceptual quality threshold.

5 more elements

Dependencies are inferred from the claim text.

Compressing digital images utilizing deep learning-based perceptual similarity

Life of this patent

Compressing digital images utilizing deep learning-based perceptual similarity

Life of this patent

Abstract

Claims20