Optimization of downsampling occurring before encoding images during compression

Patent metadata

Field	Value
Publication number	US-11265563-B2
Application number	US-201916599780-A
Country	US
Kind code	B2
Filing date	Oct 11, 2019
Priority date	Oct 12, 2018
Publication date	Mar 1, 2022
Grant date	Mar 1, 2022

Abstract

The disclosure lies in the field of coding image data, scalable in image resolution, such that one or more successive downsamplings are performed from a source signal of maximal image resolution to a minimal image resolution, in order to encode a base layer (L 0 ) corresponding to this minimal resolution, the base layer (L 0 ) serving as a reference for one or more enhancement layers (L 1 , L 2 ) to be used for decoding at a terminal having a screen of higher image resolution than the minimal image resolution. From said successive downsamplings, the base layer (L 0 ) and the one or more enhancement layers (L 1 , L 2 ) are constructed, and then an encoded bit stream is prepared in order to be transmitted, comprising data of the base layer and of the enhancement layer or layers. In particular, the downsampling step comprises the application of an adaptive low-pass filter to the image data from the downsampling.

First claim

The invention claimed is:

1. A method for coding image data, scalable in image resolution, comprising:
performing, based on a source signal of maximal image resolution, one or more successive downsamplings to a minimal image resolution, in order to encode a base layer corresponding to said minimal resolution, wherein the base layer is usable as a reference for one or more enhancement layers to be used for decoding at a terminal having a screen of higher image resolution than the minimal image resolution,
constructing the base layer and the one or more enhancement layers based on the one or more successive downsamplings,
preparing an encoded bitstream to be transmitted, wherein the encoded bitstream comprises data of the base layer and data of the one or more enhancement layers,
wherein at least one of the one or more successive downsamplings comprises:
performing an image pre-analysis to determine at least an importance criterion to be given to one or more regions of interest in image data of the source signal, the importance criterion being a texture complexity score of a region of interest measured by:
a determination of a depth map in the current image, in order to identify a plurality of superimposed planes, and in each of the identified planes, a detection and a quantification of blur,
assigning of a score to a region according to a depth of a plane in which this region is located, and a sharpness of an image in this plane,

selecting at least one of a plurality of predetermined types of adaptive filters based at least in part upon the determined importance criterion,
applying an adaptive low-pass filter of the selected type to the image data of the source signal to reduce a spectral folding effect of the at least one of the one or more successive downsamplings.

2. The method according to claim 1 , wherein the adaptive filter is of a finite impulse response type.

3. The method according to claim 2 , wherein the finite impulse response filter is of a type chosen among at least:
a Kaiser-Bessel window filter,
a bicubic filter,
a bilinear filter,
an interpolation filter.

4. The method according to claim 1 , wherein the type of adaptive filter is selected for an image or an image region according to at least one criterion among:
an image resolution before downsampling,
a frame rate of the source signal,
an image coding rate.

5. The method according to claim 1 , wherein a bicubic filter is selected for sharp foreground regions, while a bilinear filter is used for sharp background regions and/or fuzzy foreground regions, and otherwise an interpolation filter is used.

6. The method according to claim 1 , wherein a range of configurations defining coefficients of a Kaiser-Bessel window filter is chosen based on a criterion among an image resolution before downsampling and a frame rate of the source signal.

7. The method according to claim 6 , wherein the coefficients of the Kaiser-Bessel window filter are chosen according to a configuration selected within said range, based on other criterion among the image resolution before downsampling and the frame rate of the source signal.

8. The method according to claim 1 , wherein a bicubic filter is selected if a coding rate of the base layer is greater than a threshold, and otherwise a bilinear filter is selected.

9. The method according to claim 8 , wherein the threshold is 2 Mbits/sec.

10. An image data coding device, scalable in image resolution, comprising a processing circuit for implementing the method according to claim 1 .

11. A computer program comprising instructions stored on a non-transitory computer-readable medium for implementing the method according to claim 1 , when said instructions are executed by a processor of the processing circuit.

12. A method for coding image data, scalable in image resolution, comprising:
performing, based on a source signal of maximal image resolution, one or more successive downsamplings to a minimal image resolution, in order to encode a base layer corresponding to said minimal resolution, wherein the base layer is usable as a reference for one or more enhancement layers to be used for decoding at a terminal having a screen of higher image resolution than the minimal image resolution,
constructing the base layer and the one or more enhancement layers based on the one or more successive downsamplings,
preparing an encoded bitstream to be transmitted, wherein the encoded bitstream comprises data of the base layer and data of the one or more enhancement layers,
wherein at least one of the one or more successive downsamplings comprises: performing an image pre-analysis to determine at least an importance criterion to be given to one or more regions of interest in image data of the source signal, the image pre-analysis identifying areas in the image data having a contrast or texture complexity above or below a predetermined threshold, the importance criterion being a texture complexity score of a region of interest measured by:
a determination of a depth map in the current image, in order to identify a plurality of superimposed planes, and in each of the identified planes, a detection and a quantification of blur,
assigning of a score to a region according to a depth of a plane in which this region is located, and a sharpness of an image in this plane,

selecting at least one of a plurality of predetermined types of adaptive filters based at least in part upon the determined importance criterion,
applying an adaptive low-pass filter of the selected type to the image data of the source signal to reduce a spectral folding effect of the at least one of the one or more successive downsamplings.

Assignees

• Ateme

Inventors

• Parois Ronan

Classifications

• H04N19/80

  Details of filtering operations specially adapted for video compression, e.g. for pixel interpolation (H04N19/635, H04N19/86 take precedence) · CPC title

• H04N19/17

  the unit being an image region, e.g. an object · CPC title

• H04N19/33 Primary

  in the spatial domain · CPC title

• H04N19/167

  Position within a video image, e.g. region of interest [ROI] · CPC title

• H04N19/187

  the unit being a scalable video layer · CPC title