High dynamic range display effect optimization method based on human visual characteristics

Abstract

The invention discloses a high dynamic range display effect optimization method based on human visual characteristics, and the method comprises the steps: carrying out the standardized nonlinear coding and decoding and HDR post-processing of an SDR image, and obtaining the linear brightness distribution of an HDR image after the SDR image is converted; linear brightness distribution of the HDR image is used as input, EOTF and brightness normalization are carried out, a glare spread function filter convolution kernel is calculated, retinal HDR image perception simulation is carried out, a retinal contrast distribution result is obtained, an evaluation index of a contrast peak signal-to-noise ratio is established, and high dynamic range display effect optimization is carried out. Theoretical support is provided for the HDR adjustment strategy conforming to the human visual characteristics, and practical support is provided for improving the display quality of the display device.

Description

Technical Field

[0001] This invention relates to the field of information display technology, and in particular to a method for optimizing high dynamic range display effects based on the characteristics of human visual perception. Background Technology

[0002] With the rapid development of display technology, High Dynamic Range (HDR) display has become a key technology for enhancing visual experience. Compared with Standard Dynamic Range (SDR) display, HDR can present a wider brightness range, richer colors, and higher contrast, thus more realistically reproducing the natural world as perceived by the human eye. However, current mainstream HDR content generation relies heavily on the physical parameters of hardware devices or simple brightness mapping, and its processing flow fails to fully consider the core perceptual characteristics of the human visual system (HVS). The human eye is not a linear light intensity sensor; its perception of brightness is non-linear and is affected by optical phenomena such as glare and scattering during actual viewing. When existing technologies directly convert SDR content to HDR, they often only focus on expanding the numerical brightness range, ignoring the rationality and optimization of this conversion result at the level of human visual perception. This can lead to problems such as loss of detail, color distortion, or decreased visual comfort in the generated HDR images.

[0003] Traditional conversion methods lack a theoretical model based on human visual perception to guide the conversion from SDR to HDR, resulting in a disconnect between the conversion process and the final viewing effect. The high-brightness light from HDR displays produces glare and optical diffusion effects when it enters the eye, altering the actual image on the retina. Existing technologies rarely simulate this crucial physical process, making it impossible to accurately predict and optimize the final viewing experience of HDR content. A key technical challenge is how to synergistically enhance color while expanding dynamic range, ensuring both vividness and visual comfort under HDR. Therefore, the industry urgently needs to establish a conversion theory that aligns with visual perception, thereby achieving a leap from "accurate reproduction of physical signals" to "optimization of visual perception effects." Summary of the Invention

[0004] This invention provides a method for optimizing high dynamic range display effects based on the characteristics of human vision. It provides theoretical support for HDR adjustment strategies that conform to the characteristics of human vision, and provides practical support for improving the display quality of display devices.

[0005] This invention provides a method for optimizing high dynamic range display effects based on human visual characteristics, comprising the following steps: The SDR image is subjected to standardized nonlinear encoding and decoding and HDR post-processing to obtain the linear brightness distribution of the HDR image after the SDR image is converted; Using the linear brightness distribution of HDR images as input, EOTF (Electro-Optical Transfer Function) and brightness normalization are performed to calculate the glare spread function filter convolution kernel. Then, retinal HDR image perception simulation is performed to obtain the retinal contrast distribution results. The peak contrast signal-to-noise ratio is established as an evaluation index to optimize the high dynamic range display effect.

[0006] Optionally, in one embodiment of the present invention, the linear brightness distribution of the HDR image after SDR image conversion by performing normalized nonlinear encoding / decoding and HDR post-processing on the SDR image includes: Standardized nonlinear encoding and decoding is achieved through static PQ decoding, color gamut conversion, and static PQ encoding of SDR images; Saturation adjustment, HSV color enhancement, and brightness boosting are performed on HDR images with standardized nonlinear codecs to obtain the linear brightness distribution of the HDR image.

[0007] Optionally, in one embodiment of the present invention, static PQ decoding of an SDR image includes: The input 8-bit SDR image is normalized to a floating-point range of [0,1], and EOTF decoding is performed using the PQ curve in the ITU-R BT.2100 standard to obtain the linear luminance of each pixel in physical units. The decoding formula is as follows: ; Where L is the physical brightness of each pixel, and m1, m2, c1, c2, and c3 are the standard coefficients of PQEOTF. L max This is the maximum brightness limit for PQEOTF.

[0008] Optionally, in one embodiment of the present invention, color gamut conversion includes: SDR images use the BT.709 color gamut. To be consistent with HDR display devices, the linear brightness image after SDR image decoding is mapped to the BT.2020 color gamut through a fixed transformation matrix. The input for color gamut conversion is the decoded linear brightness image of the BT.709 color gamut, and the output is the linear brightness image of the BT.2020 color gamut.

[0009] Optionally, in one embodiment of the present invention, static PQ encoding includes: The linear luminance image in the BT.2020 color gamut is non-linearly mapped and encoded using PQOOTF and OETF. OOTF is a technology that converts linear light from natural scenes into SDR-like light for screen display. OETF maps SDR-like normalized luminance values ​​to non-linear code values ​​to compress dynamic range and simulate human eye perception characteristics. It can cover a display luminance range of 0.0001 to 10000 nits in a 10-bit signal. The mapping formula is as follows: ; Among them, input x The output y is the SDR-like normalized luminance value, and the output y is the PQ encoded value. m1, m2, c1, c2, and c3 are the standard coefficients of PQEOTF.

[0010] Optionally, in one embodiment of the present invention, saturation adjustment, HSV color enhancement, and brightness enhancement are performed on the HDR image with standardized nonlinear codec to obtain a linear brightness distribution of the HDR image, including: An HDR encoded image is obtained by standardizing and nonlinearly encoding / decoding the SDR image. The HDR encoded image is then converted to the HSV color space, and the V of the luminance channel and the H and S of the chrominance channels are extracted. The V of the luminance channel is normalized and used as an enhancement factor to weight and adjust the saturation S. The processed H and S channels are then merged with the original luminance layer and converted back to the RGB color space to output an RGB101010 HDR encoded image. The linear luminance value is then recovered using the PQROTF decoding function.

[0011] Optionally, in one embodiment of the present invention, the glare spread function filter convolution kernel is simulated and calculated using the glare spread function filter equation in the CIE standard, and the formula is: ; Wherein, RLUminance is the relative brightness formed on the retina by the glare source. The viewing angle between the scattering pixel and the receiving pixel. Age The age of the observer, p For the observer's iris pigmentation.

[0012] Optionally, in one embodiment of the present invention, retinal HDR image perception simulation includes: An idealized HDR image stimulus scene is constructed to simulate the real-perceived image received by the human retina. A linear luminance map decoded by PQ-EOTF is used to represent the actual light energy distribution of the input stimulus in physical space, with a luminance range covering from 10. -3 Up to 10 4 nits, which conforms to the dynamic brightness distribution characteristics of HDR content.

[0013] Optionally, in one embodiment of the present invention, the evaluation index of peak contrast signal-to-noise ratio is established by objectively quantifying the perceived impact of glare diffusion on image display quality in the human visual system: ; Where ContrastPSNR is the peak contrast-to-noise ratio, M is the number of rows, and N is the number of columns. C ( i , j )for i OK j Contrast of column pixels.

[0014] This invention presents a high dynamic range (HDR) display optimization method based on human visual characteristics. Addressing the lack of a theoretical model for SDR to HDR conversion based on human visual perception in traditional methods, this invention provides a more comprehensive and applicable HDR display optimization method that meets human visual perception requirements. Based on the basic structure of the human visual system and the principle of brightness perception, an experimental platform is constructed, innovatively establishing retinal HDR image perception based on the optical glare diffusion function (HVS). Through observer visual perception experiment ratings, experimental images, and simulation results, the invention intuitively reveals the interference effect of optical glare diffusion on the retinal image formation process under HDR conditions. This emphasizes the importance of introducing HVS perceptual modeling in HDR image quality assessment, providing a theoretical foundation for this mechanism to subsequently perform perception-driven image enhancement, quality evaluation, and even subjective comfort modeling. According to the method of this invention, when using mobile display devices such as smartphones, only the type of displayed image needs to be detected to implement high dynamic range display adjustment strategies.

[0015] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0016] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart illustrating a high dynamic range display effect optimization method based on human visual characteristics according to an embodiment of the present invention; Figure 2 This is a flowchart of the HDR algorithm according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the PQEOTF curve according to an embodiment of the present invention. Figure 4 This is a flowchart of HDR retinal contrast quantization based on the human eye glare diffusion function, according to an embodiment of the present invention. Figure 5(a) and (b) are schematic diagrams of the normalized glare diffusion function convolution kernel and two-dimensional cross-sectional views of the convolution kernel, respectively, according to embodiments of the present invention. Figure 6 (a) is the grayscale image of the input image in this embodiment of the invention; (b) is the linear brightness image after PQ-EOTF decoding; (c) is the scene brightness contrast image; (d) is the retinal perceived brightness distribution map; (e) is the comparison curve of one-dimensional scene brightness contrast (black solid line) and retinal contrast (red dashed line). Figure 7 (a) is the grayscale image of the input image in this embodiment of the invention; (b) is the linear brightness map after PQ-EOTF decoding; (c) is the scene brightness contrast image; (d) is the brightness distribution map perceived on the retina. Figure 8 This is an actual effect diagram of the comparison curve between one-dimensional scene brightness contrast (solid black line) and retinal contrast (dashed red line) in an embodiment of the present invention. Detailed Implementation

[0017] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0018] This invention aims to improve image clarity and visual realism on mobile terminals by processing sRGB 8-bit images into HRGB 10-bit images for output. Simultaneously, it addresses the issue of glare diffusion in HDR images, where high brightness stimuli spread to adjacent retinal areas via scattering and optical blurring within the eye, causing bright areas to "contaminate" dark areas. Therefore, this invention constructs a glare diffusion mechanism model based on the human visual system and quantifies the HDR retinal contrast effect. This invention not only provides solid theoretical support for developing HDR display adjustment strategies that conform to human visual characteristics but also offers practical solutions and technical support for significantly improving the ultimate visual quality of consumer and professional display devices.

[0019] Figure 1 This is a flowchart of a high dynamic range display effect optimization method based on human visual characteristics, according to an embodiment of the present invention.

[0020] like Figure 1 As shown, the high dynamic range display effect optimization method based on the characteristics of human visual perception includes the following steps: Step 1: Perform normalized nonlinear encoding and decoding and HDR post-processing on the SDR image to obtain the linear brightness distribution of the HDR image after the SDR image is converted; Step 2: Using the linear brightness distribution of the HDR image as input, perform EOTF and brightness normalization, calculate the glare spread function filter convolution kernel, and perform retinal HDR image perception simulation to obtain the retinal contrast distribution results. Establish the peak contrast signal-to-noise ratio evaluation index to optimize the high dynamic range display effect.

[0021] In one embodiment of the present invention, the linear brightness distribution of the HDR image converted from the SDR image is obtained by performing normalized nonlinear encoding / decoding and HDR post-processing on the SDR image, including: Standardized nonlinear encoding and decoding is achieved through static PQ decoding, color gamut conversion, and static PQ encoding of SDR images; Saturation adjustment, HSV color enhancement, and brightness boosting are performed on HDR images with standardized nonlinear codecs to obtain the linear brightness distribution of the HDR image.

[0022] In step 1, the sRGB SDR image undergoes HDR PQ encoding, color gamut transformation, and HSV (Hue, Saturation, Value) saturation enhancement to ultimately simulate the linear brightness distribution of the HDR image.

[0023] This invention establishes a conversion process from SDR-RGB images to HDR images using linear luminance value distribution, and introduces a color enhancement method to ultimately simulate the linear luminance distribution of HDR images. The process mainly includes the following stages: pixel-level perceptual quantizer (PQ) based decoding, color gamut mapping, HDR image encoding, HSV space color enhancement, and decoding to obtain the linear luminance distribution, such as... Figure 2 As shown.

[0024] In the static PQ decoding of SDR images, the input image is normalized and PQ EOTF decoding is performed. Specifically, the input 8-bit sRGB image is normalized to a floating-point range of [0,1], and the PQ curve in the ITU-R BT.2100 standard is used for EOTF (Electro-Optical Transfer Function) decoding to obtain the linear luminance (in nits) of each pixel in physical units. The decoding formula is as follows: ; Where m1=0.1593017578125; m2=78.84375; c1=0.8359375; c2=18.8515625; c3=18.6875; these constants are the standard coefficients of PQEOTF, used to precisely define the nonlinear mapping curve; L max=10000 indicates that the maximum brightness limit of PQEOTF is 10000 nits (DolbyVision). The PQEOTF curve is as follows: Figure 3 As shown in (A), (B), and (C).

[0025] Regarding the BT.2020 color gamut conversion, sRGB images use the BT.709 color gamut. To ensure consistency with HDR display devices, the decoded linear light image is mapped to the BT.2020 color gamut using a fixed 3×3 transformation matrix. This color gamut mapping ensures consistency in subsequent image display and avoids color distortion. The input is the decoded BT.709 color gamut RGB linear light brightness image (in nits), and the output is the BT.2020 color gamut RGB linear light brightness image (in nits). The color gamut conversion matrix is ​​as follows: .

[0026] Regarding PQ OETF decoding and dynamic normalized brightness, to facilitate the subsequent storage, encoding, and display of HDR images, the linear light image of the BT.2020 color gamut is non-linearly mapped and encoded using PQOOTF (Opto-Optical Transfer Function) and OETF (Opto-Electronic Transfer Function). OOTF converts linear light from natural scenes into SDR-like screen display light; the specific formula is as follows: ; in, F D To display the reference light value, E G represents the linear light value of the scene. 709 [E] is the electro-optical conversion function, G 1886 [] represents the inverse function of the electro-optical conversion function; ; in, The linear light value of the scene after electro-optic conversion; ; OETF maps SDR-like normalized luminance values ​​to non-linear code values ​​to compress dynamic range and simulate human eye perception. It can cover a display luminance range of 0.0001 to 10000 nits in a 10-bit (1024 levels) signal. Specifically, it is expressed as follows: ; Among them, input x Output SDR-like normalized luminance values y This is a PQ encoded value.

[0027] The image is then normalized and enhanced to the [0,1] range to fully utilize the output dynamic range, and quantized to a 10-bit integer format (RGB101010) to meet the requirements of the HDR display interface.

[0028] Furthermore, after constructing the HDR encoded image, HSV color enhancement is performed to further enhance the image's color performance and address the issue of color fading in bright areas. This ensures that the image maintains a high dynamic range while presenting more saturated and realistic colors on HDR display devices. First, the HDR image is converted to the HSV color space, and the luminance channel (V) and chrominance channels (H, S) are extracted. Second, the luminance channel is normalized and used as an enhancement factor to weight and adjust the saturation (S). The enhancement coefficient is set to 2.6 to control the overall saturation enhancement intensity. Finally, the processed H and S channels are merged with the original luminance layer and converted back to the RGB color space, outputting an RGB101010 HDR encoded image.

[0029] For linear brightness extraction of HDR images, after HDR image processing, to simulate a true brightness distribution consistent with the physical display, an RGB101010 HDR encoded image is input, and the linear brightness value is recovered using the PQROTF decoding function. After decoding, the brightness is calculated using BT.2020 weights, and the specific calculation expression is as follows: ; Where L is physical brightness, R is the red grayscale of the three primary colors, G is the green grayscale of the three primary colors, and B is the blue grayscale of the three primary colors.

[0030] Thus, the linear light distribution matrix of each pixel is obtained, realizing high-fidelity reconstruction from standard sRGB image to HDR linear light image. Combined with color enhancement technology, it provides a unified and high-quality HDR image foundation for visual perception research, HDR display optimization and image processing algorithms.

[0031] In step 2, the glare spread function filter convolution kernel is calculated, and the spatial distribution of the retinal image is calculated based on the optical glare spread function convolution kernel.

[0032] like Figure 4 As shown, the flowchart of HDR retinal contrast quantization based on the human eye glare spread function mainly calculates the convolution kernel of the optical glare spread function, considering light scattering in the human eye, i.e. glare, which is very important for accurately analyzing the visual perception effect of HDR images received by the human visual system (HVS).

[0033] This invention refers to the image received by the observer through the HVS as a "retinal image," which is the sum of the scene brightness and the light scattered to each pixel. The amount scattered to each pixel is the sum of the glare from all other pixels. Each glare contribution depends on the brightness of a distant pixel and the angular spacing between the scattering and receiving pixels.

[0034] The next step is to simulate and calculate the filter kernel using the Glare Spread Function (GSF) filter equation from the CIE standard. This kernel will then be used for convolution with the retinal input, where the formula is: ; Wherein, RLUminance is the relative brightness formed on the retina by the glare source. The viewing angle between the scattering pixel and the receiving pixel. Age The age of the observer, p For observer iris pigmentation. This formula measures equivalent veil glare in relation to relative illuminance energy. In simulations and subsequent subjective experiments, this invention uses black eye pigmentation. p=0， Age=25 .

[0035] The simulated scene parameters are a mobile phone display screen with a display area size of 69 (W) mm * 154 (H) mm, each pixel size of 0.0572917 mm, and a viewing distance of 260 mm. Based on the viewing angle between each pixel, the normalized glare diffusion function at a viewing distance of 260 mm is as follows: Figure 5 As shown in (a), Figure 5 (b) depicts its two-dimensional cross-section, with the output dimensions represented in logarithmic color levels.

[0036] To further analyze the role of optical glare in the perceived sharpness of HDR images, this invention constructs an idealized HDR image stimulation scene to simulate the real perceived image received by the human retina. Figure 6 (a) is a schematic diagram of the input image, which consists of 5×5 white rectangles of different gray levels arranged on a pure black background. Its pixel encoding adopts the RGB101010 format, which can fully express the gray level distribution under high dynamic range. Figure 6 (b) shows the linear luminance map after PQ-EOTF decoding, presenting the actual light energy distribution of the input stimulus in physical space, with a luminance range covering 10. -3 Up to 10 4 nits, which conforms to the dynamic brightness distribution characteristics of HDR content.

[0037] This invention uses 10 colors for equivalent comparison. Specifically, the equivalent brightness of the colored stimulus is equal to the brightness of the reference grayscale stimulus that produces the same pupil diameter.

[0038] Figure 6 (c) further presents the scene brightness contrast of the input image. By calculating the ratio of local maxima to minima in the spatial domain, the contrast distribution of the scene in physical space is obtained. It can be observed that there are obvious step-like brightness variations between the rectangular targets, and the contrast exhibits a relatively regular decreasing trend in the vertical direction, demonstrating the characteristics of brightness gradients in the stimulus design. In contrast, Figure 6 Figure (d) shows the perceived brightness distribution on the retina after considering the glare diffusion function of the human eye's optical system. This figure, labeled "RetinalContrast," clearly reflects the significant impact of light energy diffusion caused by glare on perceived contrast: the halo of bright areas diffuses into low-brightness areas, resulting in blurred edges and reduced local contrast, especially noticeable between the center and low-brightness targets.

[0039] To quantitatively compare the difference between scene contrast and perceived contrast. Figure 6 Figure (e) plots the luminance-contrast curve along the central cross-section, where the black bars represent the normalized scene luminance and the red dashed line represents the simulated retinal contrast. As can be seen from the figure, although the luminance-contrast in the original image exhibits a significant spatial gradient distribution, the retinal contrast is compressed due to glare diffusion, especially with a marked decrease in the peak contrast near the bright rectangles, while the contrast in low-brightness areas is further weakened. This phenomenon indicates that in an HDR display environment, simply increasing brightness or contrast does not equivalently improve perceived sharpness; the glare perception mechanism of the visual system plays a crucial role in the final subjective evaluation of the image.

[0040] After verifying the glare perception mechanism using idealized stimulus images, to further explore the applicability and manifestation of this mechanism in natural images, this invention selects a complex scene image containing high-frequency stripe textures and multi-level brightness regions as test material, such as... Figure 7 As shown in (a), this image depicts multiple zebras in natural lighting conditions. The stripe structure exhibits high spatial frequency and strong contrast edge features, while the overall image has a wide dynamic range of brightness, exhibiting typical HDR characteristics. Figure 7 (b) magnifies a portion of the image to focus on the high-contrast edges of the zebra's back, which is used in subsequent contrast cross-sectional analysis.

[0041] Figure 7 (c) shows the scene brightness-to-contrast distribution of the HDR image under conditions of imperceptible interference. Through local contrast calculations based on spatial domain brightness differentiation, it can be observed that the zebra's torso and stripe boundary regions exhibit numerous high-intensity contrast structures, especially in the foreground region where the contrast amplitude is more concentrated, reflecting the scene's rich spatial frequencies and illumination levels. However, when glare simulation based on the diffusion function is introduced into the human eye's optical system, Figure 7 The retinal perceived contrast distribution map (d) clearly shows the phenomena of "blurred boundaries" and "weakened contrast". Specifically, at the zebra outline with high contrast in the original image, the perceived contrast decreases significantly; while in areas with gradual changes in brightness, such as between the sky and the grass background, the perceived contrast is further compressed. This indicates that in complex scenes, glare diffusion has a universal effect on the perception of multi-scale contrast edges in the image, and the degree of influence varies with local brightness and edge frequency.

[0042] To quantitatively present this perceptual bias, Figure 8 Provided from Figure 7 The luminance-contrast cross-sectional analysis curve is shown in (b) with the dashed cross-section. The solid black line represents the normalized physical scene luminance-contrast, and the dashed red line represents the corresponding retinal perceived contrast. It can be observed that at each fringe edge, although the original scene luminance-contrast exhibits a highly sharp "cliff-like" change, the retinal perceived contrast shows a clear blunting trend, with its peak amplitude generally lower than the original contrast and exhibiting spatial expansion. This edge-perceived blurring caused by glare diffusion not only leads to the loss of structural details but also affects the subjective evaluation of image sharpness by the human eye, especially in HDR display devices.

[0043] To objectively quantify the impact of glare diffusion on the perceived sharpness of image display quality, this paper proposes a contrast peak signal-to-noise ratio (Contrast PSNR) evaluation metric, defined by the following formula: ; In summary, the experimental images and simulation results intuitively reveal the interference of optical glare diffusion on the retinal image formation process under HDR conditions, highlighting the importance of introducing HVS perceptual modeling in HDR image quality assessment, and providing a theoretical basis for this mechanism to model subsequent perception-driven image enhancement, quality evaluation, and even subjective comfort.

[0044] The highly modular fragment processing architecture of this invention provides underlying support for the subsequent integration of deep learning models (such as super-resolution and image restoration) or visual perception modulation mechanisms (such as CSF weighting and HDR content-aware tonemapping), and can be widely applied in practical scenarios such as smart terminal display, head-mounted display imaging, and low-power image processing.

[0045] Glare diffusion in the human visual system (HVS) is a crucial characteristic that profoundly impacts HDR perception. High-brightness stimuli can spread to adjacent retinal areas through scattering and optical blurring within the eye. This invention constructs a retinal HDR perception model based on the CIE McCann and Vonikakis retinal contrast standard calculated intraocularly. The model takes the linear brightness distribution of an HDR image as input and outputs a retinal contrast array. The system simulates and analyzes the contrast distribution of HDR images on the human retina, providing theoretical support and technical basis for subsequent HDR improvements.

[0046] This invention's method for optimizing high dynamic range (HDR) display effects based on human visual characteristics establishes a theoretical foundation for glare diffusion mechanisms and HDR display based on the perceptual characteristics of the human visual system (HVS). It determines the conversion process from SDR-RGB images to HDR linear brightness value distribution and introduces color enhancement methods to ultimately simulate the linear brightness distribution of HDR images. This invention innovatively establishes a linear brightness distribution based on pixel-level perceptual quantizer (PQ) decoding, color gamut mapping, HDR image encoding, and HSV space color enhancement. Considering the glare and image perception of light entering the human eye in HDR displays, this invention calculates the convolution kernel of the optical glare diffusion function and simulates retinal HDR image perception based on the optical glare diffusion function. This invention provides theoretical support for HDR adjustment strategies that conform to human visual characteristics and practical support for improving the display quality of display devices.

[0047] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0048] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0049] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

Claims

1. A method for optimizing high dynamic range display effects based on human visual characteristics, characterized in that, Includes the following steps: The SDR image is subjected to standardized nonlinear encoding and decoding and HDR post-processing to obtain the linear brightness distribution of the HDR image after the SDR image is converted; Using the linear brightness distribution of HDR images as input, EOTF and brightness normalization are performed, the glare spread function filter convolution kernel is calculated, and retinal HDR image perception simulation is performed to obtain retinal contrast distribution results. An evaluation index of peak contrast signal-to-noise ratio is established to optimize the high dynamic range display effect.

2. The method according to claim 1, characterized in that, The linear brightness distribution of the converted HDR image is obtained by performing normalized nonlinear encoding / decoding and HDR post-processing on the SDR image, including: Standardized nonlinear encoding and decoding is achieved through static PQ decoding, color gamut conversion, and static PQ encoding of SDR images; Saturation adjustment, HSV color enhancement, and brightness enhancement are performed on HDR images with standardized nonlinear codecs to obtain the linear brightness distribution of the HDR image.

3. The method according to claim 2, characterized in that, Static PQ decoding of SDR images includes: The input 8-bit SDR image is normalized to a floating-point range of [0,1], and EOTF decoding is performed using the PQ curve in the ITU-R BT.2100 standard to obtain the linear luminance of each pixel in physical units. The decoding formula is as follows: ； Where L is the physical brightness of each pixel, and m1, m2, c1, c2, and c3 are the standard coefficients of PQEOTF. L max This is the maximum brightness limit for PQEOTF.

4. The method according to claim 2, characterized in that, Color gamut conversion includes: SDR images use the BT.709 color gamut. To be consistent with HDR display devices, the linear brightness image after SDR image decoding is mapped to the BT.2020 color gamut through a fixed transformation matrix. The input for color gamut conversion is the decoded linear brightness image of the BT.709 color gamut, and the output is the linear brightness image of the BT.2020 color gamut.

5. The method according to claim 2, characterized in that, Static PQ encoding includes: The linear luminance image in the BT.2020 color gamut is non-linearly mapped and encoded using PQOOTF and OETF. OOTF converts linear light from natural scenes into SDR-like screen display light; OETF maps SDR-like normalized luminance values ​​to non-linear code values ​​to compress dynamic range and simulate human eye perception characteristics. It can cover a display luminance range of 0.0001 to 10000 nits in a 10-bit signal. The mapping formula is as follows: ； Among them, input x The output y is the SDR-like normalized luminance value, and the output y is the PQ encoded value. m1, m2, c1, c2, and c3 are the standard coefficients of PQEOTF.

6. The method according to claim 2, characterized in that, Saturation adjustment, HSV color enhancement, and brightness boosting are performed on HDR images with standardized nonlinear codecs to obtain the linear brightness distribution of the HDR image, including: An HDR encoded image is obtained by standardizing and nonlinearly encoding and decoding an SDR image. The HDR encoded image is then converted to the HSV color space, and the V of the luminance channel and the H and S of the chrominance channels are extracted. The V of the luminance channel is normalized and used as an enhancement factor to weight and adjust the saturation S. The processed H and S channels are then merged with the original luminance layer and converted back to the RGB color space to output an RGB101010 HDR encoded image. The linear luminance value is then recovered using the PQROTF decoding function.

7. The method according to claim 1, characterized in that, The glare spread function filter convolution kernel is calculated using the glare spread function filter equation from the CIE standard. The formula is as follows: ； Wherein, RLUminance is the relative brightness formed on the retina by the glare source. The viewing angle between the scattering pixel and the receiving pixel. Age The age of the observer, p For the observer's iris pigmentation.

8. The method according to claim 1, characterized in that, Retinal HDR image perception simulation includes: An idealized HDR image stimulus scene is constructed to simulate the real-perceived image received by the human retina. A linear luminance map decoded by PQ-EOTF is used to represent the actual light energy distribution of the input stimulus in physical space, with a luminance range covering from 10. -3 Up to 10 4 nits, which conforms to the dynamic brightness distribution characteristics of HDR content.

9. The method according to claim 1, characterized in that, To objectively quantify the perceived impact of glare diffusion on image display quality in the human visual system, the peak signal-to-noise ratio (PSNR) evaluation metric was established as follows: ； Where ContrastPSNR is the peak contrast-to-noise ratio, M is the number of rows, and N is the number of columns. C ( i , j )for i OK j Contrast of column pixels.

Images