CN-121985132-A - Chroma boosting of SDR and HDR display adaptation signals for SL-HDRx systems
Abstract
The invention provides a method comprising obtaining (90) a current RGB image, classifying (92) the colors of pixels of the current RGB image into a plurality of classes, for each color class, determining (94) data representing the color class including a primary luminance value representing the luminance of a color predominance in the class and determining (95) a value representing a chrominance gain representing a margin for increasing the chrominance component in the color class from the data representing the color class, and encoding (96) the primary luminance value and the value representing the gain corresponding to each class as metadata representing a saturation gain function in a bitstream, the function defining a color correction to be applied to the pixels from the luminance of the pixels.
Inventors
- D. Tuz
- M-J. Clatis
- C. Sell
Assignees
- 交互数字CE专利控股公司
Dates
- Publication Date
- 20260505
- Application Date
- 20210428
- Priority Date
- 20200505
Claims (20)
- 1. A method, comprising: receiving high dynamic range HDR image data associated with a current frame; Converting the HDR image data into standard dynamic range SDR image data to obtain a plurality of reference pixel values; Determining, for a plurality of color classes, a plurality of chroma saturation gain values to be applied to the plurality of reference pixel values; Generating a saturation gain function mapping luminance to chrominance gain for the plurality of color classes based on the plurality of chrominance saturation gain values, and The saturation gain function is encoded as metadata associated with the SDR image data.
- 2. The method of claim 1, further comprising: acquiring a chromaticity plane representing the color gamut, and The chromaticity plane is divided into a plurality of chromaticity sectors, wherein each of the plurality of chromaticity sectors corresponds to a respective one of the plurality of color categories.
- 3. The method of claim 1, further comprising, for pixels in the set of pixels of the current frame: deriving a luminance component from the HDR image data; applying tone mapping to the luminance component to obtain a tone mapped luminance component; Deriving a chrominance component from the HDR image data; applying a joint normalization and color correction to the chrominance components to obtain corrected normalized chrominance components, and The pixels are classified into one of the plurality of color categories using the tone-mapped luma component and the corrected normalized chroma component.
- 4. The method of claim 1, further comprising, for each of the plurality of color classes, determining a maximum allowed chroma value, wherein the maximum allowed chroma value is a chroma value that prevents clipping in a representation of the SDR image data.
- 5. The method of claim 4, further comprising, as part of the metadata, encoding a primary luminance value for each of the plurality of color categories and the maximum allowable chroma value determined for the respective color category.
- 6. The method of claim 1, wherein the SDR image data and the metadata are included in video data such that an SDR display renders the SDR image data and an HDR display reconstructs HDR content by applying the metadata to the SDR image data.
- 7. The method of claim 1, wherein the HDR image data remains as a base layer and the metadata is included with the HDR image data in video data configured as a decoder for converting the HDR base layer into SDR for SDR display.
- 8. The method of claim 1, wherein the current frame is included in a video sequence, and temporal filtering is applied to the information representative of the chroma saturation gain values based on information representative of chroma saturation gain values calculated for at least one frame of the video sequence preceding the current frame.
- 9. The method of claim 5, wherein temporal filtering is applied to information representative of the chroma saturation gain values and reinitialized at a beginning of the video sequence or upon identifying a scene cut in the video sequence.
- 10. The method of claim 1, further comprising, for each of the plurality of color categories: acquiring a histogram of luminance values of pixels of the current frame associated with the color class, and One of the following is selected as the primary luminance value: a luminance value corresponding to the maximum number of pixels in the luminance value histogram; A luminance value corresponding to a maximum chrominance energy for a histogram bin, wherein the chrominance energy for the bin is the product of the number of pixels in the bin and the maximum chrominance value found at the bin, and A luminance value corresponding to a maximum average chrominance energy for a histogram bin, wherein the average chrominance energy for the bin is the product of the number of pixels in the bin and the maximum chrominance value found at the bin.
- 11. An apparatus for video encoding, the apparatus comprising: a processor configured to: receiving high dynamic range HDR image data associated with a current frame; Converting the HDR image data into standard dynamic range SDR image data to obtain a plurality of reference pixel values; Determining, for a plurality of color classes, a plurality of chroma saturation gain values to be applied to the plurality of reference pixel values; Generating a saturation gain function mapping luminance to chrominance gain for the plurality of color classes based on the plurality of chrominance saturation gain values, and The saturation gain function is encoded as metadata associated with the SDR image data.
- 12. The apparatus of claim 11, wherein the processor is further configured to: acquiring a chromaticity plane representing the color gamut, and The chromaticity plane is divided into a plurality of chromaticity sectors, wherein each of the plurality of chromaticity sectors corresponds to a respective one of the plurality of color categories.
- 13. The apparatus of claim 11, wherein the processor is further configured to, for pixels in the set of pixels of the current frame: deriving a luminance component from the HDR image data; applying tone mapping to the luminance component to obtain a tone mapped luminance component; Deriving a chrominance component from the HDR image data; applying a joint normalization and color correction to the chrominance components to obtain corrected normalized chrominance components, and The pixels are classified into one of the color classes using the tone-mapped luminance component and the corrected normalized chrominance component.
- 14. The apparatus of claim 11, wherein the processor is further configured to determine, for each of the plurality of color categories, a maximum allowed chroma value, wherein the maximum allowed chroma value is a chroma value that prevents clipping in a representation of the SDR image data.
- 15. The apparatus of claim 14, wherein the processor is further configured to encode, as part of the metadata, a primary luminance value for each of the plurality of color classes and the maximum allowable chroma value determined for the respective color class.
- 16. The apparatus of claim 11, wherein the SDR image data and the metadata are included in video data such that an SDR display renders the SDR image data and an HDR display reconstructs HDR content by applying the metadata to the SDR image data.
- 17. The apparatus of claim 11, wherein the HDR image data remains as a base layer and the metadata is included with the HDR image data in video data configured as a decoder for converting the HDR base layer into SDR for SDR display.
- 18. The apparatus of claim 11, wherein the current frame is included in a video sequence, and temporal filtering is applied to information representative of chroma saturation gain values calculated for at least one frame preceding the current frame in the video sequence based on the information representative of the chroma saturation gain values.
- 19. The apparatus of claim 15, wherein temporal filtering is applied to information representative of the chroma saturation gain values and reinitialized at a beginning of the video sequence or upon identifying a scene cut in the video sequence.
- 20. The apparatus of claim 11, wherein the processor is further configured to, for each of the plurality of color categories: acquiring a histogram of luminance values of pixels of the current frame associated with the color class, and One of the following is selected as the primary luminance value: a luminance value corresponding to the maximum number of pixels in the luminance value histogram; a luminance value corresponding to a maximum chrominance energy for a histogram bin, wherein the chrominance energy of the bin is the product of the number of pixels in the bin and the maximum chrominance value found at the bin, and A luminance value corresponding to a maximum average chrominance energy for a histogram bin, wherein the average chrominance energy for the bin is the product of the number of pixels in the bin and the maximum chrominance value found at the bin.
Description
Chroma boosting of SDR and HDR display adaptation signals for SL-HDRx systems The application is a divisional application of patent application with the application number 202180036592.0, the application date 2021, the month 4 and the day 28 and the application name of 'chromaticity improvement of SDR and HDR display adaptation signals aiming at a SL-HDRx system'. 1. Technical field At least one of the present embodiments relates generally to the field of distribution of HDR video using a SL-HDRx system (x=1, 2, or 3). 2. Background art Recent advances in display technology began to allow for an extended dynamic range of color, brightness, and contrast in the image to be displayed. The term image here refers to image content which may be, for example, video or still pictures or images. High dynamic range video (HDR video) describes video with a dynamic range that is greater than the dynamic range of standard dynamic range video (SDR video). HDR video involves capturing, authoring, content/encoding, and displaying. HDR capture and display can present brighter white and darker black. To accommodate this, the HDR encoding standard allows for higher maximum brightness, and uses at least a 10-bit dynamic range (compared to 8 bits for non-professional video and 10 bits for professional SDR video) in order to maintain accuracy over this extended range. Although technically "HDR" strictly refers to the ratio between maximum luminance and minimum luminance, the term "HDR video" is also generally understood to mean a wide color gamut. Although many HDR display devices have emerged, as well as image cameras capable of capturing images with increased dynamic range, the amount of HDR content available is still very limited. There is a need for a solution that allows extending the dynamic range of existing content so that such content can be efficiently displayed on an HDR display device. Standard SL-HDR1 (ETSI TS 103 433-1 series, latest version v1.3.1) provides direct backward compatibility by using metadata that allows reconstructing the HDR signal from an SDR video stream that can be delivered using existing SDR distribution networks and services. SL-HDR1 allows HDR rendering on an HDR device and SDR rendering on an SDR device using a single layer video stream. Standard SL-HDR2 (ETSI TS 103 433-2 series, latest version v 1.2.1) is suitable for HDR devices. Standard SL-HDR2 allows transmission of ST-2084 (also known as PQ (perceptual quantizer) or HDR 10) streams, as well as metadata. When a stream is received by a device compatible with ST-2084 only and not with metadata, the latter ignores the metadata and displays the image without knowing all its technical details (depending on the device model and its processing capabilities, color rendering and tone scale detailed information may not be consistent with the original source). When a device supporting the ST-2084 format and metadata receives a stream, it will display an optimized image that best meets the intent of the content producer. Standard SL-HDR3 (ETSI TS 103 433-3 v 1.1.1) allows transmission of HLG (mixed logarithmic gamma) streams and metadata. The SL-HDR3 system comprises an HDR/SDR reconstruction block based on a SL-HDR2 HDR/SDR reconstruction block, i.e. it consists of a cascade of HLG to ST-2084 OETF (photoelectric transfer function) converters and SL-HDR2 HDR/SDR reconstruction blocks. OETF describes the action of a sensor to convert scene brightness into data. In the SL-HDRx system, the chromaticity of the SDR and HDR display-adapted signals can be tuned because the color correction adjustment variables are contained in the SL-HDRx metadata. These color correction adjustment variable metadata define a piecewise function, called SGF (saturated gain function), which modifies the default color correction function in any SL-HDRx process. The color correction depends on the luminance (Y component of the image signal), i.e. the color correction modifies the color (e.g. the U and V components) of the pixel in accordance with the luminance of said pixel. Typically, SGF metadata defines a maximum of six points in coordinates (SGF _x, SGF _y), SGF _x represents luminance, and SGF _y represents color correction at this luminance. sgf _x and sgf _y coordinates are values comprised between "0" and "255", for example. By default, SGF provides the same default color correction for each intensity value. This default color correction, which is typically defined empirically, results in neutral SDR and HDR display adaptation signals. A basic solution to increase the chromaticity of SDR and HDR display adaptation signals is to increase the chromaticity globally by making different color corrections for each of the luminance values. This solution has some limitations: only globally controlling the colors. Thus, if certain colors are already saturated enough, adding color correction to those colors will make them appear supersaturated; adding uncontrolled color correctio