FR Patent - Compression of Images and Videos

Patent Number: FR 1600516 - August 2, 2024
Publication Number: 3 049 799
Filing Date: March 29, 2016

Title:
Compression of images, sequences of images and videos using the Fast Fourier Transform and one-dimensional methods.

Abstract:
The present invention relates to the use of one-dimensional compression methods to efficiently compress certain images and videos, using the two-dimensional (2D) or one-dimensional (1D) Fast Fourier Transform (FFT).
The most widespread methods currently used to compress images use two-dimensional methods and two-dimensional transformations, in particular the discrete cosine transform (DCT), the discrete wavelet transform (DWT), the high efficiency video coding (HEVC), or even more rarely the 2D Fast Fourier Transform. Temporal redundancy reduction algorithms use motion-compensated prediction (or motion compensation).
To compress the images, we use the fast Fourier transform in 2D (k-space) or 1D (intermediate space) and apply methods already known (notably in audio compression) to lines in the frequency domain. For the videos, we apply line-adapted residual generation algorithms and additional lossless compressions that are efficient and compatible with parallel processing. Adequate use of phase precision allows for significant increases in compression rates.
These methods are intended to compress images and videos for storage or transmission, including those from medical imaging, such as magnetic resonance imaging (MRI) or computed tomography (CT), as well as those used in stereoscopy or multi-view.

Notes:
- This information is written for investors, developers and decision-makers.
- Additional articles will be posted from time to time to provide more details or to add supplements.