An HDR10 Image Hierarchical Encoding and Decoding Method, Device, and Storage Medium

Abstract

The present invention relates to an HDR10 image hierarchical encoding and decoding method, apparatus, and storage medium. The method includes receiving YUV420 video data frames and performing dynamic range segmentation on the pixels of each frame data, dividing each pixel into a standard dynamic range part SDR and a high dynamic range part HDR; respectively performing non-linear quantization calculations on the pixel value brightness of the standard dynamic range part SDR and the high dynamic range part HDR to obtain the brightness component of the base layer and the brightness component of the enhancement layer; based on the standard dynamic range part SDR and the high dynamic range part HDR, performing non-linear quantization calculations on the chrominance of each pixel value to obtain the chrominance residual value and the chrominance residual quantization value; performing hierarchical encoding and metadata encapsulation on the pixels of the frame data to obtain an HDR10 bitstream, and transmitting the HDR10 bitstream. The present invention realizes efficient encoding and transmission compatible with the HDR10 standard through a dynamic range segmentation algorithm, dual-segment non-linear quantization, hierarchical encoding, and dynamic metadata synchronization.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular, to an HDR10 image hierarchical encoding and decoding method, apparatus, and storage medium. Background Art

[0002] The core requirements of the HDR10 standard are 10-bit color depth (brightness range 0 - 10000 nits, conforming to the ST2084 standard) and including static metadata (MaxFALL / MaxCLL). Existing HDR image transmission technologies have the following technical pain points:

[0003] (1) Bit-depth compression distortion: Traditional 8-bit encoding causes loss of details in high-light / shadow areas and cannot meet the 10-bit color depth performance required by the HDR10 standard.

[0004] (2) Color gamut compression: Color deviation occurs during the conversion from the BT.709 color gamut (standard high definition) to the P3 color gamut (digital cinema).

[0005] (3) Linear quantization defect: Conventional linear quantization is prone to color level breakage in low-bandwidth scenarios, damaging the wide color gamut characteristics of HDR.

[0006] (4) Dynamic range loss: Existing hierarchical encoding schemes are not optimized for HDR characteristics, and the enhancement layer lacks a luminance component compensation mechanism.

[0007] (5) Poor hardware compatibility: Most hardware encoders only support 8-bit processing and it is difficult to directly implement 10-bit HDR data stream transmission.

[0008] Typical technical solutions such as the HLG / PQ curve can expand the dynamic range, but have the following deficiencies: They do not solve the problem of high-precision residual transmission under low bandwidth, lack a compatible design with existing encoding and decoding standards, and cannot achieve 10-bit reconstruction in an 8-bit encoding framework. Summary of the Invention

[0009] The present invention provides an HDR10 image hierarchical encoding and decoding method, apparatus, and storage medium, aiming to solve at least one of the technical problems existing in the prior art.

[0010] The technical solution of the present invention is an HDR10 image hierarchical encoding and decoding method, and the HDR10 image hierarchical encoding and decoding method includes the following steps:

[0011] S100. Receive YUV420 video data frames and perform dynamic range segmentation on the pixels of each frame data, dividing each pixel into a standard dynamic range part SDR and a high dynamic range part HDR;

[0012] S200. Non-linearly quantize and calculate the pixel value brightness of the standard dynamic range part SDR and the high dynamic range part HDR respectively to obtain the luminance component of the base layer and the luminance component of the enhancement layer ;

[0013] S300. Based on the standard dynamic range part SDR and the high dynamic range part HDR, non-linearly quantize and calculate the chrominance of each pixel value to obtain the chrominance residual value and the chrominance residual quantization value ;

[0014] S400. Perform hierarchical coding and metadata encapsulation on the pixels of the frame data to obtain the HDR10 bitstream, and transmit the HDR10 bitstream

[0015] Furthermore, the HDR1 / 0 image hierarchical encoding and decoding method further includes the following steps:

[0016] S500. Receive and interpret the HDR10 bitstream, and reconstruct the base layer data through inverse quantization and upsampling operations

[0017] S600. Calculate the reconstructed luminance and the reconstructed chrominance value , and fuse the enhancement layer

[0018] S700. Convert and output the YUV444 video data of HDR10 with a 10-bit depth

[0019] Furthermore, step S100 includes:

[0020] S110. Receive the YUV420 video frame data of the HDR10 standard with a 10-bit (bit) depth. Among them, the luminance Y and chrominance U / V of each pixel point in the video frame data are both [0, 1023];

[0021] S120. Based on the improved Logistic function and a preset threshold, obtain the standard dynamic range part SDR and the high dynamic range part HDR

[0022] In the step S120, the improved Logistic function is:

[0023]

[0024]

[0025] The preset threshold is 0.5. When < 0.5, this pixel point belongs to the standard dynamic range part SDR. When When it is ≥0.5, this pixel belongs to the high dynamic range part HDR.

[0026] Furthermore, in step S200,

[0027] When the pixel belongs to the standard dynamic range part SDR, the two-segment non-linear quantization luminance component is:

[0028]

[0029] When the pixel belongs to the high dynamic range part HDR, the two-segment non-linear quantization luminance component is:

[0030]

[0031] Wherein,

[0032]

[0033] is the 10-bit luminance value of the pixel of the video data frame, is the adjusted luminance value;

[0034] The luminance component of the base layer is:

[0035]

[0036] Wherein, round is the rounding function, and the luminance component of the base layer is an 8-bit unsigned integer, 0 ≤ ≤ 255;

[0037] The luminance component of the enhancement layer

[0038]

[0039] Wherein, round is the rounding function, clip is the upper and lower limit function, is the reconstructed luminance value of the base layer:

[0040]

[0041] The luminance component of the enhancement layer is an 8-bit signed integer, -128 ≤ ≤ 127.

[0042] Furthermore, step S300 includes:

[0043] S310, Downsample each pixel of the frame data to the YUV420 format to obtain an 8-bit non-linear quantization chrominance value ;

[0044] S320. Calculate the residuals of each pixel when converting from the P3 color gamut to 8-bit non-linear quantized chrominance values, obtaining chrominance residual values ; ;

[0045] S330. Convert the chrominance residual values to chrominance residual quantization values .

[0046] Furthermore, step S310 includes:

[0047] S311. Based on the transfer matrix, convert the color gamut of the video data frame from the P3 color gamut to the BT.709 color gamut. The transfer matrix is:

[0048]

[0049] where is the input P3 color gamut chrominance value. The P3 color gamut is in the YUV444 format with 10-bit precision, is the BT.709 color gamut chrominance value after color gamut conversion. The BT.709 color gamut is in the YUV420 format with 8-bit precision, is the 3×3 color gamut conversion matrix for converting from the DCI-P3 color gamut to the BT.709 color gamut using the ITU-R BT.2087 standard:

[0050]

[0051] S312. Downsample the YUV444 chrominance to YUV420 chrominance based on the anti-aliasing filtering method;

[0052] S313. Calculate the 8-bit non-linear quantized chrominance value based on the chrominance quantization formula. The chrominance quantization formula is:

[0053]

[0054] where round is the rounding function;

[0055] In step S320, the chrominance residual value is:

[0056]

[0057] where is the input P3 color gamut chrominance value, is the 8-bit non-linear quantized chrominance value;

[0058] In step S330, the chrominance residual quantization value is:

[0059]

[0060] Among them, round is the rounding function, and clip is the upper and lower limit functions.

[0061] Further, in step S400,

[0062] The layered coding includes base layer coding and enhancement layer coding.

[0063] The base layer encoding is based on the HEVC Main Profile encoding standard, encoding the input 8-bit depth YUV420 data, and its bit rate control mode is CBR;

[0064] The enhancement layer encoding is based on the HEVC Range Extension coding standard, which is used to encode the luminance component of the input 8-bit depth enhancement layer. and 8-bit depth residual quantization value Encoding is performed with the bit rate control mode being VBR.

[0065] Further, refer to Figure 4 In step S500, reconstructing the base layer includes:

[0066] S510, decode the HDR10 code stream and obtain the brightness component of the base layer ;

[0067] S520: Dequantize the brightness component of the base layer , get the base layer reconstruction brightness value ;

[0068] If the brightness component of the base layer <=116,

[0069]

[0070] If the brightness component of the base layer >116,

[0071] ;

[0072] S530, use Lanczos3 interpolation algorithm to perform chroma upsampling operation and restore to YUV444 format;

[0073] The step S600 includes:

[0074] S610, in The luminance component of the enhancement layer is superimposed , and the reconstructed brightness is obtained , where 0<= <=1023;

[0075]

[0076] S620. Reconstruct the chrominance value through the chrominance values in the BT.709 color gamut and the chrominance residual quantization value .

[0077]

[0078] The step S700 includes:

[0079] S710. Convert the luminance data and chrominance data into the color space of the display device;

[0080] S720. Calculate the display luminance based on the electro-optical transfer function ST2084 EOTF curve ;

[0081]

[0082] wherein, is the reconstructed luminance, is the maximum luminance value of the content.

[0083] Furthermore, the present invention also provides an HDR10 image hierarchical encoding and decoding device for performing the HDR10 image hierarchical encoding and decoding method. The HDR10 image hierarchical encoding and decoding device includes:

[0084] A dynamic analysis unit for dividing the pixels of each frame data into a standard dynamic range part SDR and a high dynamic range part HDR based on dynamic range segmentation;

[0085] A non-linear quantization unit for non-linearly quantizing the luminance and chrominance of the pixels of each frame data. The non-linear quantization unit is connected to the dynamic analysis unit;

[0086] A dual-enhanced layer encoding unit for performing hierarchical encoding operations. The dual-enhanced layer encoding unit is connected to the non-linear quantization unit;

[0087] An HDR metadata generation unit for generating encapsulated data and transmitting it;

[0088] A decoding unit for decoding and displaying the encapsulated data.

[0089] Furthermore, the present invention also provides a computer-readable storage medium, on which program instructions are stored. When the program instructions are executed by a processor, the HDR10 image hierarchical encoding and decoding method is implemented.

[0090] The beneficial effects of the present invention are:

[0091] The described HDR10 image hierarchical encoding and decoding method, device, and storage medium achieve efficient encoding and transmission compatible with the HDR10 standard through a dynamic range segmentation algorithm, two-segment non-linear quantization, hierarchical encoding, and dynamic metadata synchronization. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] Figure 1 It is the overall flowchart of the HDR10 image hierarchical encoding and decoding method.

[0093] Figure 2 It is a schematic diagram of the encoding process in the HDR10 image hierarchical encoding and decoding method.

[0094] Figure 3 It is a schematic diagram of the hierarchical encoding in the HDR10 image hierarchical encoding and decoding method.

[0095] Figure 4 It is a schematic diagram of the decoding process in the HDR10 image hierarchical encoding and decoding method.

[0096] Figure 5 It is a schematic diagram of the output process in the HDR10 image hierarchical encoding and decoding method.

[0097] Figure 6 It is a schematic diagram of the structure of the HDR10 image hierarchical encoding and decoding device. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0098] The following will clearly and completely describe the concept, specific structure, and technical effects generated by the present invention in combination with the embodiments and the drawings, so as to fully understand the purpose, solution, and effects of the present invention. It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

[0099] It should be noted that, unless otherwise specified, when a certain feature is referred to as "fixed" or "connected" to another feature, it can be directly fixed or connected to another feature, or indirectly fixed or connected to another feature. In addition, the up, down, left, right, top, bottom, etc. descriptions used in the present invention are only relative to the mutual positional relationship of the components of the present invention in the drawings.

[0100] In addition, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field of the present invention. The terms used in the description of the present specification are only for describing specific embodiments, rather than for limiting the present invention. The term "and / or" used herein includes any combination of one or more related listed items.

[0101] It should be understood that although the terms first, second, third, etc. may be used in this disclosure to describe various elements, these elements should not be limited to these terms. These terms are only used to distinguish elements of the same type from each other. For example, without departing from the scope of this disclosure, the first element may also be referred to as the second element, and similarly, the second element may also be referred to as the first element.

[0102] Referring to Figures 1 to 6 , in some embodiments, the technical solution of the present invention is an HDR10 image hierarchical encoding and decoding method. Referring to Figure 1 , the HDR10 image hierarchical encoding and decoding method includes the following steps:

[0103] S100. Receive YUV420 video data frames and perform dynamic range segmentation on the pixels of each frame data, dividing each pixel into a standard dynamic range part SDR and a high dynamic range part HDR;

[0104] S200. Perform non-linear quantization calculations on the pixel values of the standard dynamic range part SDR and the high dynamic range part HDR respectively to obtain the luminance component of the base layer and the luminance component of the enhancement layer ;

[0105] S300. Based on the standard dynamic range part SDR and the high dynamic range part HDR, perform non-linear quantization calculations on the chrominance of each pixel value to obtain the chrominance residual value and the chrominance residual quantization value ;

[0106] S400. Perform hierarchical encoding and metadata encapsulation on the pixels of the frame data to obtain an HDR10 bitstream and transmit the HDR10 bitstream.

[0107] The beneficial effects of the present invention are:

[0108] The HDR10 image hierarchical encoding and decoding method, device and storage medium realize efficient encoding and transmission compatible with the HDR10 standard through a dynamic range segmentation algorithm, two-stage non-linear quantization, hierarchical encoding and dynamic metadata synchronization.

[0109] Specifically, the present invention realizes efficient encoding and transmission compatible with the HDR10 standard through the following technological innovations:

[0110] (1) Dynamic range segmentation algorithm:

[0111] - Separate SDR / HDR content based on the S-shaped curve of the improved Logistic function,

[0112] - Adaptive brightness threshold selection (500 nit boundary) to retain 90% of the scene details.

[0113] (2)Hierarchical processing: Decompose 10-bit YUV420 data into:

[0114] - An 8-bit YUV420 base layer (including non-linear quantized main luminance),

[0115] - A luminance residual enhancement layer (compensating for highlight / shadow details),

[0116] - A chroma extension enhancement layer (retaining wide color gamut characteristics).

[0117] (3)Two-stage non-linear quantization:

[0118] - For dark areas (≤500 nit), use γ = 2.4 to enhance details,

[0119] - For bright areas (>500 nit), use γ = 0.6 to suppress quantization noise.

[0120] (4)Color gamut extension enhancement:

[0121] - Construct a P3→BT.709 color gamut difference matrix,

[0122] - Residual coding retains wide color gamut characteristics.

[0123] (5)Enhancement layer construction:

[0124] - Luminance residual layer: Extract luminance difference values beyond the 8-bit expression range,

[0125] - Chroma extension layer: Retain the difference information between the P3 color gamut and the BT.709 color gamut.

[0126] (6)Dynamic metadata synchronization:

[0127] - Embed MaxCLL / MaxFALL parameters in the SEI layer,

[0128] - Update luminance statistical information frame by frame.

[0129] Technical effects:

[0130] (1)Achieve high-fidelity reconstruction of 10-bit HDR data under an 8-bit coding framework.

[0131] (2)The dynamic range retention rate is increased by 40%. The reconstruction accuracy of HDR data under the 8-bit coding framework:

[0132] - PSNR ≥ 48 dB (0 - 1000 nit region),

[0133] - PSNR ≥ 42 dB (1000 - 10000 nit region).

[0134] (3)Bandwidth efficiency:

[0135] - 1080P at 60fps only requires 15Mbps (35% lower than the traditional solution).

[0136] (4) Color gamut restoration:

[0137] - P3 / BT.709 conversion with ΔE < 1.5 (JNCD < 0.5).

[0138] (5) Compatible with HEVC Main10 Profile / 8 bit yuv420 hardware encoding and decoding.

[0139] Further, the HDR10 image hierarchical encoding and decoding method further includes the following steps:

[0140] S500, Receive and interpret the HDR10 bitstream, and reconstruct the base layer data through inverse quantization and upsampling operations;

[0141] S600, Calculate the reconstructed luminance and the reconstructed chrominance values to fuse the enhancement layer;

[0142] S700, Convert and output 10-bit depth HDR10 YUV444 video data.

[0143] Further, referring to Figure 1 and Figure 2 Step S100 includes:

[0144] S110, Receive 10-bit (bit) depth YUV420 video frame data in the HDR10 standard, where the luminance Y and chrominance U / V of each pixel point in the video frame data are both [0,1023];

[0145] S120, Based on the improved Logistic function and a preset threshold, obtain the standard dynamic range part SDR and the high dynamic range part HDR.

[0146] In the step S120, the improved Logistic function is:

[0147]

[0148] where the steepness of the improved Logistic function is 0.005, and Y is the luminance of the pixel point;

[0149] The preset threshold is 0.5. When < 0.5, this pixel point belongs to the standard dynamic range part SDR. When >= 0.5, this pixel point belongs to the high dynamic range part HDR.

[0150] Specifically, the dynamic range is segmented for the input 10-bit HDR10 YUV420 data (Y ∈ [0, 1023], U / V ∈ [0, 1023]), and the improved Logistic function is applied to separate the dynamic range:

[0151]

[0152] Among them, the SDR part (L < 0.5): 0 - 500 nits, is directly mapped to 8 bits, and the HDR part / residual part (L ≥ 0.5): 500 - 10000 nits, generates a residual enhancement layer.

[0153] Parameter design: S-shaped curve The steepness (k = 0.005) of the curve ensures a clear boundary between the dark / bright parts at 500 nits.

[0154] Y0 = 500 nits, the SDR / HDR boundary, and the analysis of typical HDR10 scenarios. Based on the dynamic range segmentation mechanism of the improved Logistic function (k = 0.005 ± 10%), the linkage control between the 500-nit threshold and the HDR10 metadata.

[0155] In a specific embodiment, such as an urban night scene image,

[0156] Neon light area (Y = 600):

[0157]

[0158] Dark alley area (Y = 200):

[0159]

[0160] Furthermore, referring to Figure 1 and Figure 2 , in step S200,

[0161] When the pixel point belongs to the standard dynamic range part SDR, the two-segment non-linear quantization luminance component is:

[0162]

[0163] When the pixel point belongs to the high dynamic range part HDR, the two-segment non-linear quantization luminance component is:

[0164]

[0165] The segmented design with γ = 2.4 for the dark part / γ = 0.6 for the bright part, and the parameter optimization with α = 116.5 and β = 127.5;

[0166] Among them,

[0167]

[0168] is the 10-bit luminance value of the pixels of the video data frame, is the adjusted luminance value;

[0169] The luminance component of the base layer is:

[0170]

[0171] Among them, round is the rounding function, and the luminance component of the base layer is an 8-bit unsigned integer, 0 ≤ ≤ 255;

[0172] The luminance component of the enhancement layer

[0173]

[0174] Among them, round is the rounding function, and clip is the upper and lower limit function, is the reconstructed luminance value of the base layer:

[0175]

[0176] The luminance component of the enhancement layer is an 8-bit signed integer, -128 ≤ ≤ 127.

[0177] Specifically, the two-segment non-linear quantization is:

[0178]

[0179]

[0180]

[0181] Output 8-bit Y component (0 - 255)

[0182] The piecewise quantization parameter (γ = 2.4 / 0.6) and the segmentation threshold are complementary:

[0183] The dark area (Y ≤ 500 nit) is stretched by a high γ value to match the dark vision sensitivity of the human eye

[0184] The bright area (Y > 500 nit) is compressed by a low γ value to suppress the noise visibility caused by the too rapid increase of the quantization step

[0185] Parameter Design:

[0186] γ = 2.4: Dark part gamma value, matching the dark vision response of the human eye (MSCNSSIM increased by 12%)

[0187] γ = 0.6: Bright part gamma value, suppressing high - light quantization noise (PSNR increased by 5.2dB)

[0188] α = 116.5 (ensuring Q(Y)=116 when Ynorm = 0.5))

[0189] β = 127.5 (ensuring the output range is 0 - 255))

[0190] Specifically, the base layer is generated as follows:

[0191]

[0192] Where Q(Y_norm) is a two - segment quantization function, corresponding to the non - linear mapping result after dynamic segmentation

[0193] Specifically, the enhancement layer is generated as follows:

[0194]

[0195] Where: is the reconstructed luminance value of the base layer, RY is an 8 - bit signed integer (-128~+127), and is stored in the enhancement layer bitstream.

[0196] Referring to Table 1, each parameter is calculated as follows:

[0197] Table 1

[0198] parameter calculation formula calculation process result Y_10bit input value 600 Y_norm Y_10bit / 1023 600 / 1023≈0.586 0.586 Q(Y_norm) 127.5×(0.586-0.5)^0.6 + 128 127.5×0.086^0.6 + 128≈160.3 160.3 Y_base round(Q(Y_norm)) round(160.3) 160 Y_rec_base [(160-128) / 127.5]^1 / 0.6 + 0.5 (32 / 127.5)^1.667 + 0.5≈0.586 600 R_Y round((600 - 600)×0.5) 0 0

[0199] Furthermore, referring to Figure 1 and Figure 2 , step S300 includes:

[0200] S310. Downsample each pixel of the frame data to the YUV420 format to obtain 8 - bit non - linear quantized chrominance values ;

[0201] S320. Calculate the residual of each pixel converted from the P3 color gamut to the 8 - bit non - linear quantized chrominance value to obtain the chrominance residual value ;

[0202] S330. Convert the chrominance residual value to the chrominance residual quantization value .

[0203] Furthermore, referring toFigure 1 and Figure 2 , step S310 includes:

[0204] S311. Based on the transfer matrix, convert the color gamut of the video data frame from the P3 color gamut to the BT.709 color gamut. The transfer matrix is:

[0205]

[0206] wherein, is the input P3 color gamut chromaticity value, and the P3 color gamut is in the YUV444 format with 10-bit precision, is the BT.709 color gamut chromaticity value after color gamut conversion, and the BT.709 color gamut is in the YUV420 format with 8-bit precision, is the 3×3 color gamut conversion matrix for converting the DCI-P3 color gamut to the BT.709 color gamut using the ITU-R BT.2087 standard:

[0207]

[0208] In a specific embodiment, P3 red [0.680, 0.320]:

[0209]

[0210]

[0211] S312. Based on the anti-aliasing filtering method, downsample the YUV444 chromaticity to YUV420 chromaticity;

[0212] In a specific embodiment, anti-aliasing filtering is used:

[0213] # Example of using a 5-tap filterkernel = [0.05, 0.25, 0.4, 0.25, 0.05] # Low-pass filter coefficientsfiltered_U = convolve2d(U_plane, kernel, mode='same')downsampled_U =filtered_U[::2, ::2] # 2x2 downsampling

[0214] Technical key point: Filter first and then downsample to avoid high-frequency information aliasing and retain edge information (such as using an adaptive filter).

[0215] S313. Calculate the 8-bit non-linear quantization chromaticity value based on the chromaticity quantization formula , and the chromaticity quantization formula is:

[0216]

[0217] Among them, round is the rounding function;

[0218] In terms of parameter design, offset +0.1: to protect the chroma of the dark part; divisor 1.2: to avoid high-saturation overflow; γ = 0.45: to match the chroma perception of human eyes.

[0219] In step S320, the chroma residual value is:

[0220]

[0221] Among them, is the input chroma value in P3 color gamut, is the 8-bit non-linear quantization chroma value; 0.07 is the empirical threshold of the maximum difference between P3 and 709 color gamuts.

[0222] In step S330, the chroma residual quantization value is:

[0223]

[0224] Among them, round is the rounding function and clip is the upper and lower limit function.

[0225] In a specific embodiment, an example of a red neon light:

[0226] 1) Original P3 chroma: (x = 0.680, y = 0.320)

[0227] 2) Converted to BT.709: (x = 0.630, y = 0.340)

[0228] 3) Residual calculation:

[0229]

[0230] Furthermore, referring to Figure 1 and Figure 3 , in step S400,

[0231] The hierarchical coding includes base layer coding and enhancement layer coding,

[0232] The base layer coding is based on the HEVC Main Profile coding standard to encode the input 8-bit depth YUV420 data, and its bitrate control mode is CBR;

[0233] The enhancement layer coding is based on the HEVC Range Extension coding standard for the input 8-bit depth luminance component of the enhancement layer and the 8-bit depth chroma residual quantization value Perform encoding with a VBR (Variable Bit Rate) rate control mode.

[0234] In a specific embodiment,

[0235] Base layer encoding:

[0236] Input: 8-bit YUV420 data (Y_base, C_709)

[0237] Encoding standard: HEVC Main Profile

[0238] Rate control: CBR (Constant Bit Rate) mode (target bit rate 12 Mbps)

[0239] Enhanced layer encoding:

[0240] Input: Y residual (R_Y 8-bit) + UV residual (R_UV, 8-bit)

[0241] Encoding standard: HEVC Range Extension

[0242] Rate control: VBR mode (peak bit rate 3 Mbps)

[0243] Metadata encapsulation:

[0244] <HDR10_Metadata>

[0245] <maxcll> 1000< / maxcll>

[0246] <maxfall> 300< / maxfall>

[0247] <masteringdisplay>

[0248] <red x="0.680" y="0.320" / >

[0249] <green x="0.265" y="0.690" / >

[0250] <blue x="0.150" y="0.060" / >

[0251] < / masteringdisplay>

[0252] < / HDR10_Metadata>

[0253] In a specific embodiment, in dynamic metadata synchronization, the SEI layer metadata encapsulation format and frame-by-frame update mechanism are involved.

[0254] Furthermore, in step S500, reconstructing the base layer includes:

[0255] S510. Decode the HDR10 bitstream to obtain the luminance component of the base layer ;

[0256] S520. Inverse-quantize the luminance component of the base layer , to obtain the reconstructed luminance value of the base layer ;

[0257] If the luminance component of the base layer <= 116,

[0258]

[0259] If the luminance component of the base layer > 116,

[0260] ;

[0261] S530. Perform chroma upsampling operation using the Lanczos3 interpolation algorithm and restore it to the YUV444 format;

[0262] In a specific embodiment, the neon pixels are:

[0263]

[0264]

[0265]

[0266] Different regions Result:

[0267]

[0268] The step S600 includes:

[0269] S610. On Overlay the luminance component of the enhancement layer to obtain the reconstructed luminance , where 0 <= <= 1023;

[0270]

[0271] S620. Reconstruct the chroma value and the chroma residual quantization value through the BT.709 color gamut chroma value

[0272]

[0273] In a specific embodiment, the neon pixels are:

[0274]

[0275]

[0276]

[0277] Referring to Figure 5 , the step S700 includes:

[0278] S710. Convert the luminance data and chrominance data into the color space of the display device;

[0279] S720. Calculate the display luminance based on the electro-optical transfer function ST2084 EOTF curve ;

[0280]

[0281] where, is the reconstructed luminance, is the maximum luminance value of the content.

[0282] The ST2084 EOTF (Electro-Optical Transfer Function) is a standard released by SMPTE (Society of Motion Picture and Television Engineers) in 2014, which is used to define the luminance mapping of HDR (High Dynamic Range) videos. The ST2084 EOTF is a non-linear transfer function. Based on the human visual perception of banding artifacts, it can achieve no visible banding artifacts under 12-bit encoding. It allows the display device to accurately convert the luminance information in the signal into the actual light output, thus achieving a higher luminance range and richer detail performance. This standard aims to create video content with a higher luminance range, rather than simply increasing the overall luminance. In this way, the ST2084 EOTF can present a consistent visual effect on different display devices.

[0283] The luminance range of the ST2084 EOTF: The ST2084 EOTF can represent a luminance range from 0.0001 nits to 10000 nits. This enables it to support very high luminance peaks, thus performing excellently when displaying high-light details. Comparison with SDR: Compared with the gamma curve used in traditional SDR (Standard Dynamic Range) videos, both the curve shape and luminance range of the ST2084 EOTF are significantly different. The luminance range of SDR is usually around 100 nits, while that of the ST2084 EOTF is as high as 10000 nits. Reference white light: Initially, the ST2084 defined the reference white light (nominal diffuse white light) as around 100 nits. However, after evaluation, it was found that better visual effects can be obtained when the reference white light is adjusted to about 200 nits.

[0284] In a specific embodiment, first convert it into the color space of the display device (such as Dolby Vision), then apply the ST2084 EOTF curve, and finally output 10-bit HDR10 YUV444 data.

[0285] Referring to Table 2, for the HDR10 image hierarchical encoding and decoding method described above, the test data (the "SunsetAtPier" sequence) is as follows:

[0286] Table 2,

[0287] index the present invention traditional solution PSNR(0 - 1000nit) 48.2 dB 42.7 dB decoding delay 33.2 ms 41.8 ms gamut coverage 97% P3 89% P3

[0288] The HDR10 image hierarchical encoding and decoding method described above can achieve a chromaticity reduction with ΔE < 1.5 at a bitrate of 15 Mbps, and the color gamut coverage is increased to 97%.

[0289] Furthermore, referring to Figure 6 , the present invention also provides an HDR10 image hierarchical encoding and decoding apparatus for performing the HDR10 image hierarchical encoding and decoding method described above. The HDR10 image hierarchical encoding and decoding apparatus includes:

[0290] A dynamic analysis unit for dividing the pixels of each frame data into a standard dynamic range part SDR and a high dynamic range part HDR based on dynamic range segmentation;

[0291] A non-linear quantization unit for non-linearly quantizing the luminance and chrominance of the pixels of each frame data. The non-linear quantization unit is connected to the dynamic analysis unit;

[0292] A dual-enhanced layer encoding unit for performing hierarchical encoding operations. The dual-enhanced layer encoding unit is connected to the non-linear quantization unit;

[0293] An HDR metadata generation unit for generating encapsulated data and transmitting it;

[0294] A decoding unit for decoding and displaying the encapsulated data.

[0295] Furthermore, the present invention also provides a computer-readable storage medium having program instructions stored thereon. When the program instructions are executed by a processor, the HDR10 image hierarchical encoding and decoding method is implemented.

[0296] Chinese interpretation:

[0297] P3 color gamut: DCI-P3 color space, covering approximately 45% of the CIE1931 color gamut, with a red wavelength of 630 nm, a green wavelength of 532 nm, and a blue wavelength of 467 nm.

[0298] BT.709 color gamut: ITU-R standard, covering 35.9% of the CIE1931 color gamut, used for SDR video.

[0299] ST2084 (PQ): The perceptual quantization curve defined by the SMPTE standard, supporting logarithmic encoding from 0.0001 to 10,000 nits.

[0300] MaxCLL (Maximum Content Light Level): The maximum content light level value (such as 1500 nit).

[0301] MaxFALL (Maximum Frame Average Light Level): The peak average light level of the frame.

[0302] As described above, these are only the preferred embodiments of the present invention. The present invention is not limited to the above-mentioned embodiments. As long as the same means are used to achieve the technical effects of the present invention, any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present disclosure shall be included within the scope of protection of the present disclosure. Within the scope of protection of the present invention, various different modifications and changes can be made to its technical solutions and / or implementation manners.

Claims

1. An HDR10 image hierarchical encoding and decoding method, characterized in that The described HDR10 image hierarchical encoding and decoding method includes the following steps: S100. Receive YUV420 video data frames and perform dynamic range segmentation on the pixels of each frame data, dividing the pixels into a standard dynamic range part SDR and a high dynamic range part HDR; S200, non-linearly quantize and calculate the pixel value brightness of the standard dynamic range part SDR and the high dynamic range part HDR respectively to obtain the brightness component of the base layer and the brightness component of the enhancement layer ; S300. Non-linearly quantize and calculate the chrominance of each pixel value based on the standard dynamic range part SDR and the high dynamic range part HDR to obtain a chrominance residual value and a chrominance residual quantization value ; S400. Perform hierarchical encoding and metadata encapsulation on the pixels of the frame data to obtain an HDR10 bitstream, and transmit the HDR10 bitstream; Among them, in step S400, the hierarchical encoding includes base layer encoding and enhancement layer encoding, The base layer encoding is based on the HEVC Main Profile encoding standard, encodes the input 8-bit depth YUV420 data, and its bitrate control mode is CBR; The enhancement layer is encoded based on the HEVC Range Extension coding standard for the luminance component of the input 8-bit depth enhancement layer and the 8-bit depth chrominance residual quantization values for encoding, and its bitrate control mode is VBR.

2. The HDR10 image hierarchical encoding and decoding method according to claim 1, wherein The described HDR10 image hierarchical encoding and decoding method further includes the following steps: S500. Receive and decode the HDR10 bitstream, and reconstruct the base layer data through inverse quantization and upsampling operations; S600, calculating and reconstructing luminance and reconstructing chrominance values , and fusing the enhancement layer; S700. Convert and output 10-bit depth HDR10 YUV444 video data.

3. The HDR10 image hierarchical encoding and decoding method according to claim 1, characterized in that, Step S100 includes: S110. Receive 10-bit depth YUV420 video frame data in the HDR10 standard, where the luminance Y and chrominance U / V of each pixel point in the video frame data are both in [0, 1023]; S120. Based on the improved Logistic function and a preset threshold, obtain the standard dynamic range part SDR and the high dynamic range part HDR; In the described step S120, the improved Logistic function is: ； Among them, the steepness of the improved Logistic function is 0.005, and Y is the luminance of the pixel point; The preset threshold is 0.

5. When < 0.5, this pixel belongs to the standard dynamic range part SDR. When >= 0.5, this pixel belongs to the high dynamic range part HDR.

4. The HDR10 image layer-by-layer encoding and decoding method according to claim 1, characterized in that, In step S200, When the pixel belongs to the standard dynamic range part SDR, the two-segment non-linear quantization luminance component is as follows: ； When the pixel belongs to the high dynamic range part HDR, the two-segment non-linear quantization luminance component is as follows: ； Among them, ； is the 10-bit luminance value of the pixels of the video data frame, is the adjusted luminance value; The luminance component of the base layer is as follows: ； Among them, round is the rounding function, and the luminance component of the base layer is an 8-bit unsigned integer, 0 ≤ ≤ 255; Luminance component of the enhancement layer is as follows: ； where round is the rounding function and clip is the upper and lower limit function, is the reconstructed luminance value of the base layer: ； Luminance component of the enhancement layer is an 8-bit signed integer, -128 ≤ ≤ 127.

5. The HDR10 image hierarchical encoding and decoding method according to claim 1, wherein Step S300 includes: S310, downsample each pixel of the frame data into the YUV420 format to obtain 8-bit non-linearly quantized chrominance values ; S320. Calculate the residuals of each pixel when converting from the P3 color gamut to the 8-bit non-linear quantization chromaticity values to obtain chromaticity residual values ; ; S330. Convert the chrominance residual value into a chrominance residual quantization value .

6. The HDR10 image hierarchical encoding and decoding method according to claim 5, characterized in that Step S310 includes: S311. Based on the transfer matrix, convert the color gamut of the video data frame from the P3 color gamut to the BT.709 color gamut, and the transfer matrix is: ； Among them, is the chromaticity value of the P3 color gamut. The P3 color gamut is in the YUV444 format with 10-bit precision. is the chromaticity value of the BT.709 color gamut after color gamut conversion. The BT.709 color gamut is in the YUV420 format with 8-bit precision. is the 3×3 color gamut conversion matrix for converting the DCI-P3 color gamut to the BT.709 color gamut using the ITU-R BT.2087 standard: ； S312. Based on the anti-aliasing filtering method, downsample the YUV444 chrominance to YUV420 chrominance; S313. Calculate the 8-bit non-linear quantized chrominance value based on the chrominance quantization formula , where the chrominance quantization formula is: ； Among them, round is the rounding function; In step S320, the chrominance residual value is: ； Among them, is the input chromaticity value of the P3 color gamut, is the 8-bit non-linear quantization chromaticity value; In step S330, the chrominance residual quantization value is as follows: ； Among them, round is the rounding function, and clip is the upper and lower limit function.

7. The HDR10 image hierarchical encoding and decoding method according to claim 2, characterized in that In the described step S500, reconstructing the base layer includes: S510. Decode the HDR10 bitstream to obtain the luminance component of the base layer ; S520, the luminance component of the inverse quantization base layer , to obtain the reconstructed luminance value of the base layer ; If the luminance component of the base layer <= 116, ； If the luminance component of the base layer > 116, ； S530. Use the Lanczos3 interpolation algorithm to perform chrominance upsampling operation and restore it to the YUV444 format; The described step S600 includes: S610. On superimpose the luminance component of the enhancement layer to obtain the reconstructed luminance , where 0 <= <= 1023; ； S620. Reconstruct the chrominance value through the chrominance values of the BT.709 color gamut and the chrominance residual quantization value ; ; ； The described step S700 includes: S710. Convert the luminance data and chrominance data into the display device color space; S720. Calculating the display brightness based on the ST2084 EOTF curve of the electro-optical transfer function ; ； wherein, is the reconstructed luminance, is the maximum luminance value of the content.

8. An HDR10 image hierarchical encoding and decoding device for performing the HDR10 image hierarchical encoding and decoding method according to any one of claims 1 to 7, characterized in that, The described HDR10 image hierarchical encoding and decoding device includes: A dynamic analysis unit for dividing the pixels of each frame data into a standard dynamic range part SDR and a high dynamic range part HDR based on dynamic range segmentation; A non-linear quantization unit that performs non-linear quantization on the luminance and chrominance of the pixels of each frame data, and the non-linear quantization unit is connected to the dynamic analysis unit; A dual-enhanced layer coding unit for performing hierarchical coding operations, the dual-enhanced layer coding unit being connected to the non-linear quantization unit; An HDR metadata generation unit for generating encapsulated data and transmitting it; A decoding unit for decoding and displaying the encapsulated data.

9. A computer-readable storage medium having program instructions stored thereon, characterized in that, When the program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images