Image rendering system and driving method thereof

Abstract

The present disclosure provides an image rendering system and a driving method thereof, and relates to the technical field of display. The driving method comprises the following steps: a plurality of physical sub-pixels on a display panel are divided into at least two pixel groups according to spatial visibility in a main viewing area, and different pixel groups correspond to different sub-viewing areas in the main viewing area for picture display; wherein the physical sub-pixels, which emit light beams converging to the same sub-viewing area, are located in the same pixel group; a sampling grid is divided in a picture display range corresponding to at least one sub-viewing area; a sampling point is selected in the sampling grid for sampling, and image rendering is performed based on the sampling result. In this way, the 3D picture edge jaggy caused by prism component light splitting and the dispersion of physical sub-pixels corresponding to the same sub-viewing area can be effectively improved, and the smoothness of the image geometric edge can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of display technology, and in particular to an image rendering system and its driving method. Background Technology

[0002] Current 3D display devices mainly achieve stereoscopic vision by configuring light-splitting structures such as prisms in front of the display panel to direct the light rays of different entity sub-pixels to different spatial directions, so as to ensure that the observer's left and right eyes can receive image information with parallax respectively.

[0003] However, the geometric edges of 3D images observed by the human eye often appear jagged, affecting the quality of 3D display. Summary of the Invention

[0004] To address the aforementioned technical issues, this disclosure provides an image rendering system and its driving method, which aims to reduce or eliminate jagged edges in 3D images and improve display quality.

[0005] In a first aspect, this disclosure provides a driving method for an image rendering system, including: Multiple physical sub-pixels on the display panel are divided into at least two pixel groups according to their spatial visibility in the main view area. Different pixel groups correspond to different sub-view areas in the main view area for displaying the image. Among them, multiple physical sub-pixels whose emitted light beams converge to the same sub-view area are located in the same pixel group. Divide the screen display area corresponding to at least one sub-view area into a sampling grid; Sampling points are selected within the sampling grid for sampling, and image rendering is performed based on the sampling results.

[0006] Secondly, this disclosure provides an image rendering system, driven by the driving method provided in the first aspect, the image rendering system comprising: The first module is used to divide multiple physical sub-pixels on the display panel into at least two pixel groups according to their spatial visibility in the main viewing area. Different pixel groups correspond to different sub-view areas in the main viewing area for displaying the image. Among them, multiple physical sub-pixels whose emitted light beams converge to the same sub-view area are located in the same pixel group. The second module is used to divide the screen display area corresponding to at least one sub-view area into a sampling grid; The third module is used to select sampling points within the sampling grid for sampling and to render the image based on the sampling results.

[0007] Thirdly, this disclosure provides yet another image rendering system, including: The first module is used to divide multiple physical sub-pixels on the display panel into at least two pixel groups according to their spatial visibility in the main viewing area. Different pixel groups correspond to different sub-view areas in the main viewing area for displaying the image. Among them, multiple physical sub-pixels whose emitted light beams converge to the same sub-view area are located in the same pixel group. The second module is used to divide the screen display area corresponding to at least one sub-view area into a sampling grid; The third module is used to select sampling points within the sampling grid for sampling and to render the image based on the sampling results.

[0008] The technical solution provided in this disclosure has the following advantages compared with the prior art: In 3D display systems, due to the presence of hardware beam-splitting structures (such as prism components), multiple entity sub-pixels sent to the same sub-viewing area (corresponding to a single eye of the observer) typically exhibit a discontinuous and discrete distribution in the physical space of the display panel. This solution, through the aforementioned driving method, can group these entity sub-pixels that are simultaneously visible within the same sub-viewing area of ​​the main viewing area into the same pixel group, and uniformly divide the display area corresponding to these pixels into sampling grids for oversampling rendering. This approach breaks the limitation of traditional anti-aliasing that only samples within a single entity pixel, enabling oversampling calculation within a continuous spatial grid oriented towards monocular or binocular vision. This effectively improves the jagged edges of 3D images caused by beam splitting by prism components and the discreteness of entity sub-pixels corresponding to the same sub-viewing area, thus enhancing the smoothness of image geometric edges.

[0009] In this disclosure, the division of the sampling grid and the selection of sampling points are based on the display area of ​​the screen corresponding to the sub-view area (i.e., the virtual display surface constructed by the same group of entity sub-pixels), rather than directly aligning with the entity sub-pixels on the display panel. Selecting sampling points within this macroscopic sampling grid for sampling calculation allows the image rendering system to fully consider the spatial visual transition between adjacent entity sub-pixels within the same sub-view area when performing image assignment. This makes the distribution of sampling points more closely match the actual continuous contour of the 3D scene in a specific view area, which is beneficial for improving the rendering refinement of image geometric edges and enhancing the smoothness of 3D vision. Attached Figure Description

[0010] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0011] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a schematic diagram of a 3D head-up display system provided in an embodiment of the present invention; Figure 2 The diagram shown illustrates a mapping relationship between some entity sub-pixels of the display panel and the main viewing area in the image rendering system provided by this embodiment. Figure 3 The image shows a schematic diagram of one arrangement of physical sub-pixels in a display panel; Figure 4 The diagram shown is a schematic representation of the arrangement of virtual sub-pixels in a sub-viewport of an image rendering system provided in an embodiment of this disclosure. Figure 5 The diagram shows a sampling grid division of virtual sub-pixels in a sub-view area; Figure 6 The diagram shown is a flowchart of a driving method for an image rendering system provided in an embodiment of this disclosure; Figure 7 The diagram shows a process for selecting sampling points within a sampling grid. Figure 8 The diagram shown is a schematic diagram of another arrangement of virtual sub-pixels in a sub-viewport of the image rendering system provided in the embodiments of this disclosure; Figure 9 The diagram shows a flowchart for calculating the driving data of the corresponding entity sub-pixels based on the sampling information. Figure 10 The diagram shown is a flowchart for obtaining sampling information at each sampling point; Figure 11 The diagram shown illustrates a correspondence between a virtual reference camera and a sub-view area. Figure 12 The diagram shows the relative positions of physical sub-pixels and prism components on the display panel. Figure 13 The diagram shows a process of selecting sampling points within a sampling grid for sampling and rendering an image based on the sampling results. Figure 14 The diagram shows another workflow for selecting sampling points within a sampling grid and rendering the image based on the sampling results. Figure 15 The diagram shows another workflow for selecting sampling points within a sampling grid and rendering the image based on the sampling results. Figure 16 The diagram shows another schematic of sampling grid division for virtual sub-pixels in a sub-view area; Figure 17The diagram shows another schematic of sampling grid division for virtual sub-pixels in a sub-view area; Figure 18 The diagram shows a correspondence between the human eye and sub-viewing regions. Figure 19 The diagram shows a flowchart for dividing the screen display area into sampling grids corresponding to at least one sub-view area; Figure 20 The diagram shown is a structural schematic of an image rendering system provided in an embodiment of this disclosure. Detailed Implementation

[0013] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0014] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0015] Current 3D display devices mainly achieve stereoscopic vision by configuring light-splitting structures such as prisms in front of the display panel to direct the light rays of different entity sub-pixels to different spatial directions, so as to ensure that the observer's left and right eyes can receive image information with parallax respectively.

[0016] Due to the physical limitations of this hardware architecture, when an observer is in a specific viewpoint, the set of pixels that the human eye can perceive is actually a number of discrete pixels that cross physical adjacency boundaries and are spatially distributed. In the existing image rendering system, the lack of a customized rendering mechanism for this set of spatially visible pixels results in noticeable jagged edges in the final projected or displayed 3D image, becoming a core technical defect affecting the quality of 3D displays.

[0017] Before describing the technical solution provided by the embodiments of the present invention, the applicable scenarios and system architecture of the technical solution are first illustrated. This technical solution can be applied in the field of 3D display, such as in automotive 3D head-up display (HUD) technology or other 3D display products.

[0018] Figure 1 This is a schematic diagram of a 3D head-up display system provided in an embodiment of the present invention. Figure 1As shown, the in-vehicle system includes an image processor (not shown), an image source, and an optical system. The image processor acquires 3D scene information at preset viewpoint sampling points and renders it into a 2D image. The rendered 2D image is then transmitted to the image source. Light emitted from the image source strikes the optical system (such as a reflective optical system), is reflected and magnified by the system, and then enters the windshield of the vehicle, ultimately reflecting off the windshield and reaching the driver's eyes.

[0019] The backward extensions of light rays converge to form a virtual image surface, and the direction of light rays perceived by the driver's eyes appears to originate from this virtual image surface. Two two-dimensional images are presented on the virtual image surface for the left and right visual areas to receive, respectively, with a parallax shift corresponding to spatial depth between the two images. After the driver's eyes receive the two-dimensional images in their respective visual areas, the brain fuses them, thus perceiving a 3D virtual image with stereoscopic depth.

[0020] Figure 2 The diagram shown illustrates a mapping relationship between some entity sub-pixels of the display panel and the main view area in the image rendering system provided by this embodiment. Figure 3 The image shows a schematic diagram of one arrangement of physical subpixels in a display panel. Figure 4 The diagram shown is a schematic representation of the arrangement of virtual sub-pixels in a sub-viewport of an image rendering system provided in an embodiment of this disclosure. Figure 5 The diagram illustrates the sampling grid division of virtual sub-pixels within a sub-viewport. It should be noted that... Figure 2 The mapping relationship between some entity sub-pixels P1 in display panel 00 and the main viewing area V is illustrated only. All entity sub-pixels in display panel 00 are not illustrated. The mapping relationship between other entity sub-pixels P1 and the main viewing area V can be referred to [the relevant documentation]. Figure 2 The arrangement of the physical sub-pixels P1 in the display panel can be referenced. Figure 3 . Figure 4 The diagram only illustrates the arrangement of some virtual sub-pixels P2 within a sub-viewport V0, and does not limit the actual number of virtual sub-pixels contained in the sub-viewport. Optionally, in some embodiments, the arrangement of other sub-pixels, the number of virtual sub-pixels contained in the sub-viewport, etc., can refer to the above figure and remain consistent with it. To clearly illustrate the division of the sampling grid, Figure 5 In this embodiment, only the center point 90 and sampling point 10 of the virtual sub-pixel are shown; the complete virtual sub-pixel P2 is not shown. The arrangement of the complete virtual sub-pixel P2 can be found in [reference needed]. Figure 4 . Figure 6 The diagram shown is a flowchart of a driving method for an image rendering system provided in an embodiment of this disclosure.

[0021] Please refer to Figures 2 to 6This disclosure provides a driving method for an image rendering system, including: Step S1: Divide multiple physical sub-pixels P1 on the display panel 00 into at least two pixel groups Z0 according to their spatial visibility in the main viewing area V. Different pixel groups Z0 correspond to different sub-viewing areas V0 of the main viewing area V for display. Among them, multiple physical sub-pixels P1 whose emitted light beams converge to the same sub-viewing area V0 are located in the same pixel group Z0.

[0022] It should be noted that, in order to clearly demonstrate the correspondence between entity sub-pixel P1 and sub-view area V0, Figure 2 The same pattern fill is applied to the physical sub-pixels P1 located in the same pixel group Z0, and the same pattern fill is applied to the sub-view area V0 formed by the convergence of the light beams of the physical sub-pixels P1 in the same pixel group Z0. The emitted light color of the physical sub-pixels P1 in the same pixel group Z0 is not limited. Different pattern fills are applied to the physical sub-pixels P1 located in different pixel groups Z0, and different pattern fills are also applied to the different sub-view areas V0.

[0023] Step S2: Divide the screen display area corresponding to at least one sub-view area V0 into sampling grids W0; Step S3: Select sampling point 10 within the sampling grid W0 for sampling, and perform image rendering based on the sampling results.

[0024] In 3D displays, due to the presence of light-splitting structures (such as prisms), the distribution of physical sub-pixels P1 at the hardware level does not correspond one-to-one with the image ultimately seen by the human eye. This solution completes the mapping from the physical space of the display panel to the visual space perceived by the human eye through three steps.

[0025] In step S1, the discrete entity sub-pixels P1 on the screen are categorized and grouped according to the principle that which entity sub-pixels P1 can be mapped to the same sub-view area V0. This reconstructs, in terms of spatial topology, the originally non-adjacent and separated entity sub-pixels P1 in the same group on the display panel into a virtual screen dedicated to the sub-view area V0. The display unit on this virtual screen corresponds to the virtual sub-pixel P2. This embodiment only uses a main view area V including 5 sub-view areas V0 (sub-view areas V1 to V5) as an example for illustration, but this disclosure is not limited to this. In some other embodiments of this disclosure, a main view area V may also include other numbers of sub-view areas V0.

[0026] In step S2, the sampling grid division is no longer limited to the physical boundary of a tiny single entity sub-pixel P1, but instead, a new set of macroscopic sampling grids W0 is divided for the virtual screen display range corresponding to the sub-view area V0.

[0027] In step S3, multiple sampling points 10 are directly selected within the redefined macroscopic sampling grid W0 for oversampling calculation, and the image is rendered based on the calculation results.

[0028] As mentioned earlier, in 3D display systems, due to the presence of beam-splitting structures (such as prisms), multiple entity sub-pixels P1 sent to the same sub-viewing area V0 (corresponding to the observer's single eye) exhibit a discontinuous and discrete distribution in the physical space of the display panel. This solution, through step S1, divides these entity sub-pixels P1 that are simultaneously visible within the same sub-viewing area V0 in the main viewing area V into the same pixel group Z0, and through steps S2 and S3, uniformly divides the sampling grid W0 within the display area corresponding to these pixels for oversampling rendering. This approach breaks the limitation of related technologies where anti-aliasing only samples within a single entity pixel, enabling oversampling calculation within a continuous spatial grid oriented towards monocular or binocular vision. This effectively improves the jagged edges of the 3D image caused by prism beam splitting and the discreteness of entity sub-pixels P1 corresponding to the same sub-viewing area V0, thus enhancing the smoothness of the image's geometric edges.

[0029] In this disclosure, the division of the sampling grid W0 and the selection of sampling points 10 are based on the screen display range corresponding to the sub-view area V0 (i.e., the virtual display surface constructed by the same group of entity sub-pixels P1), for example, referring to... Figure 4 and Figure 5 Instead of directly aligning with the physical sub-pixel P1 on the display panel, sampling point 10 is selected within the macroscopic sampling grid W0 for sampling calculation. This allows the image rendering system to fully consider the spatial visual transition between adjacent physical sub-pixels P1 within the same sub-view area V0 when performing image rendering. This makes the distribution of sampling point 10 more closely match the actual continuous contour of the 3D scene in a specific view area, which is beneficial to improving the rendering precision of the image's geometric edges and enhancing the smoothness of the 3D vision.

[0030] This scheme groups physical sub-pixels P1 according to their spatial visibility and performs gridded sampling within the display range of the sub-viewing area V0. Regardless of the arrangement of the physical sub-pixels P1 on the panel, as long as they meet the physical condition that the emitted light beams converge to the same sub-viewing area V0, this scheme is applicable. This helps improve the adaptive matching capability of this driving method in different display panels or complex optical projection systems, and has good versatility.

[0031] Please continue to refer to this. Figure 2 In one optional embodiment of this disclosure, at least two virtual sub-pixels P2 within the same sub-view area V0 are at least derived from physical sub-pixels P1 in non-adjacent blocks on the display panel; or, at least two physical sub-pixels P1 within the same pixel group Z0 are spaced apart by physical sub-pixels P1 in other pixel groups Z0.

[0032] This embodiment defines the physical distribution spatial topological characteristics of the pixel group Z0 divided in step S1. In the hardware structure, such as on a 3D display panel with beam-splitting components like prisms, the light emitted by each physical sub-pixel is deflected after passing through the beam-splitting structure and projected directionally to different viewpoints in space. Therefore, in order for the light rays that finally converge in a specific sub-viewing area V0 (e.g., the left-eye or right-eye viewing area) to form a relatively complete parallax image, the pixel group Z0 corresponding to the sub-viewing area V0 is not a continuous and compact area on the display panel. Instead, it must present a discrete and jumpy arrangement. That is, within the pixel group Z0 corresponding to the same sub-viewing area V0, at least two physical sub-pixels P1 are not adjacent in the physical space of the display panel, and are interspersed and spaced by physical sub-pixels P1 belonging to other pixel groups Z0 in the horizontal or vertical direction.

[0033] Please refer to Figure 2 and Figure 3 The physical sub-pixels P1 on the display panel are arranged periodically in red, green, and blue, such as red physical sub-pixel P11, green physical sub-pixel P12, and blue physical sub-pixel P13. However, due to the tilted beam splitting of the upper prism, the physical sub-pixels P11 that can be projected onto sub-viewing areas V0 may be the red physical sub-pixel P11 in the 1st column, the green physical sub-pixel P12 in the 5th column, and the blue physical sub-pixel P13 in the 9th column. Between them, there are physical sub-pixels P1 projected onto other sub-viewing areas V0. This scheme logically maps these physically dispersed physical sub-pixels P1 to... Figure 4 Virtual sub-pixels P2 are arranged continuously in the middle.

[0034] In this embodiment, at least two physical sub-pixels P1 within pixel group Z0 have a discrete distribution characteristic that is spaced apart from physical sub-pixels P1 in other pixel groups Z0, so that the grouping rules at the software-driven level match the discrete optical path in physical space brought about by the prism beam-splitting structure. This can prevent 3D ghosting, misalignment, and visual crosstalk caused by blindly classifying physically adjacent physical sub-pixels P1 as the same viewpoint, which helps to ensure the physical purity of the virtual screen provided to a single eye in the spatial optical path.

[0035] Because it is clear that the physical sub-pixels P1 within the same pixel group Z0 are not physically adjacent, the aforementioned step S2, when constructing the sampling grid W0, can therefore no longer be limited to the microscopic region of a single physical sub-pixel P1, transforming the physically separated but visually continuous physical sub-pixels P1 into a continuous sampling matrix in visual space. This helps to eliminate the visual abruptness and jagged geometric contours caused by the non-continuous arrangement of physical sub-pixels P1 corresponding to continuous virtual sub-pixels P2 in the sub-view area V0.

[0036] Figure 7 The diagram shows a process for selecting sampling point 10 within the sampling grid W0. Please refer to it. Figure 2 , Figure 5 and Figure 7 In one optional embodiment of this disclosure, step S3 above, selecting sampling points 10 within the sampling grid W0 for sampling, and performing image rendering based on the sampling results, includes: Step S31: Select at least two sampling points 10 within the sampling grid W0, and obtain the sampling information at each sampling point 10, including pixel grayscale values; Step S32: Calculate the driving data or compensation information of the corresponding entity sub-pixel P1 based on the sampling information, and provide the driving data or compensation information to the corresponding pixel group Z0 for image rendering.

[0037] This embodiment refines step S3. First, within the sampling grid W0 defined within the virtual screen display area, instead of simply collecting single-point values, at least two or more sampling points 10 are collected (i.e., oversampling is performed). The sampling information at these sampling points 10 is obtained, with the core carrier being the original pixel grayscale values ​​mapped to the viewport by the three-dimensional virtual scene. Subsequently, the data from these multiple sampling points 10 are aggregated and mathematically transformed using a preset algorithm to calculate the final executable signal for the specific entity sub-pixel P1 controlled by the sampling grid W0. This signal can be represented as driving data that directly drives its light emission (such as grayscale voltage values ​​used to change the deflection of liquid crystal molecules or OLED current), or as compensation information that corrects on the reference image (such as compensation coefficients used to eliminate edge color difference and brightness unevenness). Finally, this data is output to the corresponding pixel group Z0.

[0038] Because at least two sampling points 10 grayscale values ​​are extracted within the sampling grid W0 and converted into driving data or compensation information, the driving signal supplied to the entity sub-pixel P1 is no longer a single-point value, but rather includes the grayscale gradient changes within the pixel's visual coverage area. This significantly reduces visual flicker caused by spatial frequency aliasing in the final 3D image, allowing the entity sub-pixel P1 to carry smoother edge grayscale transition information when emitting light, thereby improving the visual smoothness of 3D image edges. Simultaneously, the compensation information also enhances the ability to correct hardware defects such as uneven brightness.

[0039] Please refer to Figure 5 In one optional embodiment of this disclosure, within the display range corresponding to at least one sub-view area V0, the sampling point 10 is located outside the center point 90 of the corresponding virtual sub-pixel P2. Figure 5The circular fill structure in the diagram represents the center point of the corresponding virtual sub-pixel P2. It should be noted that in this disclosure, the center point 90 of virtual sub-pixels P2 of the same color uses the same graphic fill as the corresponding virtual sub-pixel P2.

[0040] This embodiment defines the relative arrangement of sampling points 10 in spatial geometry. Please refer to... Figure 5 On the virtual screen corresponding to sub-view area V0, each virtual sub-pixel P2 has its corresponding center point 90 (if the virtual sub-pixel P2 is a rectangle, the center point 90 can be represented as the intersection of the two diagonals of the rectangle; if the virtual sub-pixel P2 is a circle, the center point can be represented as the center of the circle). When multiple sampling points 10 are selected inside the sampling grid W0, these sampling points 10 do not coincide with the center point of the virtual sub-pixel P2, but are scattered around the center point.

[0041] This scheme places sampling points 10 around the center point 90, enabling these sampling points 10 to accurately capture the contour edge changes of 3D objects that cross the outer perimeter of the virtual sub-pixel P2. When the edge of an object crosses the outer perimeter of the virtual sub-pixel P2, some sampling points 10 may fall inside the object, while others may fall outside, thus obtaining a comprehensive grayscale in a transitional state. This significantly improves the sampling accuracy of the system at object boundaries, avoids the one-sidedness of grayscale obtained by sampling only a single center point, and thus significantly improves the visual step-like effect of edges.

[0042] Because sampling point 10 is off-center and located on the periphery, its spatial position is closer to the visually adjacent virtual sub-pixel P2. During oversampling calculations, this helps ensure that the driving data or compensation information corresponding to the final calculated entity sub-pixel P1 naturally includes the color and light trends of other virtual sub-pixels P2 in visual space. For the 3D image perceived by the human eye after beam splitting, this makes the transition between pixels appear smoother and more natural, helping to alleviate the hard edges caused by 3D beam splitting.

[0043] Please refer to Figure 8 , Figure 8 The diagram shown illustrates another arrangement of virtual sub-pixels P2 in a sub-viewport V0 within an image rendering system provided in this embodiment of the present disclosure. Figure 5 The difference lies in the number of sampling points 10 in the sampling grid W0. In one optional embodiment of this disclosure, the number of sampling points 10 selected in the sampling grid W0 is N, where N≥3 and N is an integer, and the distance between each of the N sampling points 10 and the center point of the corresponding virtual sub-pixel P2 is equal. Thus, the N sampling points 10 all fall on the circumference of a virtual concentric circle centered on the center point of the virtual sub-pixel P2.

[0044] In real-time 3D graphics rendering (such as in-vehicle HUD navigation screens), the geometric edges of 3D objects, lane line outlines, arrows, or character lines are highly variable and tilted in any direction towards the 2D plane. If the spatial distribution of sampling points 10 is asymmetrical (e.g., long lateral distribution distance and short longitudinal distribution distance), the system will only be able to improve edge jaggedness in specific directions, while edges with other deflection angles are prone to jagged residue.

[0045] This scheme positions all sampling points 10 on the same virtual concentric circle, ensuring a greater balance in the spatial sampling frequency of the sampling grid W0 in the two-dimensional plane across all directions (horizontal, vertical, and tilt angles). This helps ensure that regardless of the angle at which high-frequency geometric edges in the 3D scene cross the virtual sub-pixel P2, the system can obtain an equal and balanced contour capture probability, thereby helping to eliminate jaggedness caused by directional bias and achieving a smooth, isotropic improvement in all directions.

[0046] When performing oversampled data fusion (e.g., merging the grayscale values ​​of N points into a single output grayscale value), one approach involves the graphics processor assigning a distance weight to each sampling point 10 based on its spatial distance to the center point. If the distances from each point to the center are unequal, the image rendering system needs to perform high-frequency non-linear distance weight lookup or multiplication operations for each pixel in each frame. In this scheme, since all sampling points 10 are equidistant from the center point, this means that from a physical perspective of the spatial contribution to the pixel, the distance weight coefficients of these N points for the virtual sub-pixel P2 can be logically set to be equal (i.e., the fusion weight of each point is a constant 1 / N). This geometric symmetry facilitates the direct reduction of complex weighted average calculations to basic mean calculations, simplifying the algorithm and reducing the power consumption of real-time anti-aliasing drivers.

[0047] Please continue to refer to this. Figure 8 In one optional embodiment of this disclosure, based on the premise that the distances between the N sampling points 10 and the center point 90 of the corresponding virtual sub-pixels P2 are equal, the distances between any two adjacent sampling points 10 are equal along the circumferential direction of the sampling grid W0. That is, along the circumferential direction of the sampling grid W0, the arc length or straight-line distance between any two adjacent sampling points 10 is equal. For example, when N=4, the four sampling points 10 occupy the four vertices of the square (or the four quadrants of the circle), achieving a symmetrical arrangement of the N sampling points 10.

[0048] In practical applications of in-vehicle 3D HUDs or naked-eye 3D dynamic content displays, the angle of entry of the geometric boundaries of 3D objects (such as curved lane lines, turning arrows, etc.) onto the 2D projection plane is completely random and dynamically changes in real time. If the sampling points 10 on the circumference are not uniformly distributed (e.g., the point matrix is ​​dense in some angle intervals and sparse in others), then the image rendering system will experience significant fluctuations in the capture probability and accuracy of anti-aliasing calculation when facing object boundaries with specific deflection angles. This manifests as some inclined surfaces being smooth while others still have residual jagged edges. This solution sets the circumferential distance between any two adjacent sampling points 10 to be equal, making these N sampling points 10 exhibit rotational symmetry within the 360° space of the circumference. This helps to ensure that the sampling grid W0 can achieve almost equal capture probability of high-frequency geometric signals in any spatial angle, which helps to eliminate directional discrimination in the algorithm and ensures that the image can obtain almost the same level of anti-aliasing smoothing improvement when facing horizontal, vertical, diagonal, or arbitrarily slightly tilted 3D contour boundaries, eliminating visual jagged flicker caused by viewpoint rotation.

[0049] Figure 9 The diagram shown illustrates a process for calculating the driving data for the corresponding entity sub-pixels based on sampling information. Please refer to it. Figure 8 and Figure 9 In one optional embodiment of this disclosure, the driving data for the corresponding entity sub-pixel is calculated based on the sampling information, including: Step S321: Obtain the grayscale values ​​at at least two sampling points 10 selected within the sampling grid W0, as the sampling information at each sampling point 10; Step S322: Calculate the mean or weighted average of the grayscale values ​​at each sampling point 10 to obtain the driving data of the corresponding entity sub-pixels.

[0050] This embodiment specifically defines the method for converting multi-point oversampling information into driving data for entity sub-pixels P1 in the above steps. The system first obtains the original scene grayscale values ​​at each sampling point 10 within the sampling grid W0. Then, two optional algorithms can be used for calculation: one is mean calculation, where the grayscale values ​​of all sampling points 10 are added together with equal weights and then divided by the total number of sampling points N to obtain the driving data for the entity sub-pixels. The other is weighted average calculation, where different sampling points 10 are assigned different weight coefficients based on their specific visual weight position or their proximity to specific geometric boundaries, and a weighted average is calculated to finally obtain the fused grayscale value, which serves as the driving data for the entity sub-pixels P1.

[0051] Due to hardware beam splitting, 3D displays easily amplify high-frequency signal abrupt changes into visible jagged edges. By mathematically averaging or weighted averaging the gray levels of multiple peripheral sampling points 10 within the sampling grid, the abrupt edges of the image are effectively smoothed and extended at the sub-pixel perception level. This helps the brightness or color of the entity sub-pixel P1 to exhibit a continuous and gradual transition when it finally emits light, thus significantly improving the harshness of edges in a single human eye.

[0052] Furthermore, the weighted average calculation allows designers to dynamically adjust the weighting coefficients based on the actual visual experience. For example, if sharper edges are desired, the weight of sampling point 10 closer to the center can be increased; if stronger anti-aliasing and smoother edges are desired, the weight of sampling point 10 closer to the periphery can be increased. This allows the driving system to have better algorithmic adaptive adjustment flexibility when facing different display content (such as text navigation text requiring high sharpness, and stereoscopic 3D animation requiring high smoothness), which helps to preserve the effective detail clarity of the image to the greatest extent while eliminating jagged edges.

[0053] Figure 10 The diagram shown is a flowchart for obtaining sampling information at each of the 10 sampling points. Figure 11 The diagram shown illustrates a possible correspondence between the virtual reference camera 20 and the sub-view area V0. Please refer to it. Figure 2 , Figure 8 , Figure 10 and Figure 11 In one optional embodiment of this disclosure, obtaining sampling information at each of the 10 sampling points includes: Step S311: Set up M spatially discrete virtual reference cameras 20 in the main view area V, and acquire the reference view images corresponding to each virtual reference camera 20. M is greater than or equal to the number of sub-view areas V0 contained in the main view area V. Step S312: Calculate the relative distance between sampling point 10 and each virtual reference camera 20; Step S313: Determine the virtual reference camera 20 with the smallest relative distance as the target reference camera, and take the pixel grayscale value at the geometrically corresponding position of the sampling point 10 in the reference view image corresponding to the target reference camera as the sampling information at the sampling point 10.

[0054] In this embodiment, M discrete virtual reference cameras 20 are set within the main viewing area V. These virtual reference cameras 20 are not physical cameras, but can be understood as shooting points. These virtual reference cameras 20 can render M complete, high-quality two-dimensional reference view images using a conventional and efficient perspective projection method.

[0055] When the system needs to know the grayscale value of a specific sampling point 10 within the macroscopic sampling grid W0, it does not need to re-render. Instead, it calculates the 3D coordinates or projection direction of the sampling point 10 in space and the 3D geometrical relative distance (or angular difference) between it and the M preset virtual reference cameras 20. It then identifies the virtual reference camera 20 with the closest relative distance and defines it as the target reference camera. Finally, it retrieves the pixel grayscale value at the corresponding position geometrically aligned with the sampling point 10 from the already rendered reference viewpoint image of the target reference camera and assigns it directly to the sampling point 10.

[0056] In real-time 3D graphics, frequently switching rendering matrices to perform independent 3D scene geometry clipping and projection calculations for a massive number of discrete sampling points 10 can quickly overload the processor in the image rendering system. This solution transforms the computationally intensive real-time rendering into a lightweight two-dimensional lookup table (e.g., finding the grayscale value corresponding to the position of sampling point 10 in the corresponding image) and distance comparison problem by determining the virtual reference camera 20 with the smallest relative distance to sampling point 10 as the target reference camera. This significantly saves the parallel computing power of the GPU or graphics processor and reduces system power consumption. Simultaneously, it enables the entire anti-aliasing driving method to run at high frame rates in real-time on low-computing-power, low-power automotive microcontrollers or ordinary display chips, facilitating the application of 3D anti-aliasing technology in mass-produced automotive HUDs.

[0057] Please refer to Figure 11 In one optional embodiment of this disclosure, the number of virtual reference cameras 20 is the same as the number of sub-view areas V0 in the main view area V, and the virtual reference cameras 20 are located on the center line ZX of the display range of the corresponding sub-view area V0, with the center line ZX perpendicular to the display surface corresponding to the display range. Thus, each virtual reference camera 20 represents the view from the very center of the main view of the corresponding sub-view area V0.

[0058] Since the virtual reference camera 20 is preferably positioned on the vertical center line of the display range of the corresponding sub-view area V0, this means that the grayscale information borrowed by each sub-view area V0 originates from the parallax image of the main viewpoint of that sub-view area V0. When the human eye experiences slight movement or optical path deviation within the sub-view area V0, the sampling point 10, by referencing the corresponding grayscale value in the reference viewpoint image of the nearest virtual reference camera 20, effectively prevents image misalignment caused by the sampling point 10 blindly borrowing data across viewpoints, thereby helping to ensure the stereoscopic geometric consistency of the image within a single eye.

[0059] Please refer to Figure 4 and Figure 5In one optional embodiment of this disclosure, a sampling grid W0 is divided within the display area corresponding to at least one sub-view area V0, including: virtual sub-pixels P2 of the same color are not located in the same sampling cell W1 of the same sampling grid W0.

[0060] This embodiment defines the color distribution logic within the sampling grid W0 in step S2. In the display panel, color display is achieved through spatial color mixing of sub-pixels with different primary colors such as red, green, and blue. After spatial visibility mapping, the virtual sub-pixel P2 reconstructed within the display area of ​​the sub-view zone V0 also carries different emitted light colors.

[0061] In this scheme, within any independent sampling cell W1 of the sampling grid W0, two or more virtual sub-pixels P2 with the same emitted light color are not allowed. For example, if a sampling cell W1 already contains a red virtual sub-pixel P21, then other positions within that cell can only contain green virtual sub-pixels P22, blue virtual sub-pixels P23, or other virtual sub-pixels of different colors; a second red virtual sub-pixel P21 cannot appear.

[0062] In 3D display, due to the anisotropy of light paths, if the sampling grid W0 is improperly divided, multiple virtual sub-pixels P2 of the same emitted light color may be grouped into the same sampling grid W1. During the subsequent oversampling fusion step S3, these pixels of the same color will share the same local sampling point 10 grayscale contribution, potentially causing an abnormal amplification of the primary color energy in that area. Conversely, the primary color may be severely lacking in adjacent sampling grids W1, resulting in noticeable local color stripes and color shifts. This solution addresses this by placing virtual sub-pixels P2 of the same color within the same sampling grid W1. This spatial geometric division achieves spatial uniformity and dispersion of each primary color within the macroscopic sampling unit, ensuring that each sampling grid W1 possesses a complete and balanced color perception reference. This, in turn, improves the color space uniformity of the 3D virtual image and eliminates color shift noise.

[0063] Please refer to Figure 2 , Figure 4 and Figure 5 In one optional embodiment of this disclosure, step S2, dividing the screen display area corresponding to at least one sub-view area V0 into sampling grids W0, includes: Multiple sampling grids W1 arranged in an array are divided within the display area corresponding to the sub-view area V0 to form a sampling grid W0. Each sampling grid W1 in the sampling grid W0 contains three virtual sub-pixels P2 with different emitted light colors. The virtual sub-pixels P2 are mapped to the corresponding physical sub-pixels P1 in the pixel group Z0 based on spatial visibility.

[0064] Thus, each sampling cell W1 has an identical internal pattern structure, containing exactly three virtual sub-pixels P2 with distinct emitting colors. In a standard RGB full-color display system, these three distinctly colored virtual sub-pixels P2 can be represented, for example, as a red virtual sub-pixel P21, a green virtual sub-pixel P22, and a blue virtual sub-pixel P23. These three virtual sub-pixels P2 are mapped, according to the spatial visibility mapping rule in step S1, to the physical sub-pixels P1 within different pixel groups Z0 on the display panel that can project light onto that sub-viewing area V0.

[0065] Because each sampling cell W1 in the array contains a neatly arranged set of virtual sub-pixels P2 representing three different light-emitting colors, each sampling cell W1 physically possesses the ability to independently synthesize pure white light and a full-color gamut image. This array-like spatial arrangement helps ensure high spatial consistency of brightness and chromaticity across the entire space when displaying large areas of bright, white, or highly saturated solid-color backgrounds on a 3D display, thus helping to suppress macroscopic white balance drift caused by the beam-splitting structure.

[0066] Figure 12 The diagram shown illustrates the relative positions of the physical sub-pixel P1 and the prism assembly 30 on the display panel. Please refer to it. Figure 12 and Figure 5 In one optional embodiment of this disclosure, the image rendering system further includes a prism assembly 30 disposed on the light-emitting side of the display panel; the display panel includes a pixel unit PD, the pixel unit PD includes three solid sub-pixels P1 of different colors, and the solid sub-pixels P1 in the same pixel unit PD are arranged along the first direction D1; The sampling grid W1 is a quadrilateral region with length Px and width Py, where Px=P and Py=x / sin(β); Wherein, P is the prism period of the prism assembly 30, x is the width of a single pixel unit PD on the display panel in the first direction D1, and β is the tilt angle of the extension direction of the prism 31 in the prism assembly 30 relative to the second direction D2, wherein the second direction D2 is perpendicular to the first direction D1 and parallel to the plane of the display panel.

[0067] This embodiment can be applied to 3D planar display products that do not include an imaging system. This embodiment illustrates the relationship between the physical geometric dimensions of the sampling grid W1 and the arrangement of the physical sub-pixels P1 in the display panel and the arrangement of the prism assembly 30.

[0068] A pixel unit PD in the display panel includes three physical sub-pixels P1 arranged along a first direction D1. The three physical sub-pixels P1 in the same pixel unit PD emit light of different colors. The width of a pixel unit PD along the first direction D1 is x. A prism assembly 30 (i.e., a beam-splitting structure) is provided on the light-emitting surface of the display panel, which has a specific physical period P (prism pitch, i.e., the distance between the center lines of two adjacent prisms 31) perpendicular to the extension direction of the prism 31. At the same time, in order to reduce moiré patterns, the extension direction of the prism 31 is tilted at an acute angle β relative to the second direction D2.

[0069] When dividing the sampling grid W0 in the aforementioned step S2, the sampling grid W1 in the virtual space is a quadrilateral structure. Its sampling period length Px in the horizontal direction is equal to the physical period P of the prism 31, and its sampling period width in the vertical direction follows the trigonometric function proportional relationship determined by the pixel unit PD width x and the tilt angle β, that is, Py=x / sin(β).

[0070] In 3D displays, if there is a slight spatial phase deviation between the width of the sampling grid W1 and the physical pitch of the prism, after beam splitting and amplification, it will produce large, alternating bright and dark flickering stripes on the screen, severely degrading image quality. This solution sets the horizontal sampling period Px to be equal to the physical period P of the prism 31, and sets the vertical period Py to the pixel correlation step size x / sin(β) after tilt angle correction. This ensures that the anti-aliasing sampling grid W1 is geometrically aligned with the refracted light field distribution of the prism 31 in spatial distribution. This helps avoid moiré interference caused by spatial high-frequency signal aliasing and helps ensure image purity.

[0071] The size of the sampling grid W1 calculated using the above mathematical formula is precisely the smallest and most stable independent virtual pixel unit that the 3D display device can present in visual space. This means that if the sampling grid is divided into smaller units, due to the physical limitations of hardware beam splitting, the extra sampling points 10 cannot be separated for different viewpoints; if it is divided into larger units, the highest resolution of the hardware cannot be fully utilized. Therefore, the aforementioned size formula helps the image rendering system find the most accurate balance point, minimizing the number of sampling grids W1 while ensuring maximum elimination of edge jaggedness, thus reducing the redundant rendering burden on the image rendering system.

[0072] Please continue to refer to this. Figure 12In one optional embodiment of this disclosure, the image rendering system further includes a prism assembly 30 disposed on the light-emitting side of the display panel and an imaging system (not shown in the figure) disposed in the light-emitting direction of the prism assembly 30; the display panel includes a pixel unit PD, the pixel unit PD includes three physical sub-pixels P1 of different colors, and the physical sub-pixels P1 in the same pixel unit PD are arranged along the first direction D1; The sampling grid W1 is a quadrilateral region with length Px and width Py, where Px = A × P and Py = Ax / sin(β); Where A is the magnification of the imaging system, P is the prism period of the prism assembly 30, x is the width of a single pixel unit PD on the display panel in the first direction D1, and β is the tilt angle of the extension direction of the prism 31 in the prism assembly 30 relative to the second direction D2, wherein the second direction D2 is perpendicular to the first direction D1 and parallel to the plane of the display panel.

[0073] This embodiment can be applied to 3D display products that include an imaging system. For example, an in-vehicle 3D HUD head-up display system with a freeform surface mirror.

[0074] In a 3D HUD head-up display system, the light rays from the subpixels on the display panel (image source) are split by the prism 31 and cannot directly enter the human eye. They must be magnified by a reflection or projection optical imaging system with a specific magnification A (which is usually large, such as several to tens of times), and finally projected onto a virtual image surface several meters in front of the car's windshield.

[0075] Due to the significant spatial magnification effect of optical imaging systems, pixel sizes and prism periods that are originally micrometer-scale at the display panel are magnified many times over on the virtual image surface. Therefore, in step S2, when dividing the sampling grid W0, this scheme multiplies the dimensions of the horizontal period Px and the vertical period Py by the magnification factor A of the imaging system, thereby directly dividing the macroscopic array sampling grid W0 on the final virtual image surface spatial scale.

[0076] In a 3D HUD, the driver doesn't see a small display screen, but rather a macroscopic virtual image several meters or even further ahead. On this distant, wide-field-of-view virtual image surface, any tiny physical pixel break will be magnified by the optical magnification A, resulting in exceptionally coarse and glaring jagged edges. This solution directly shifts and converts the sampling grid W0 onto the virtual image surface for size division, meaning that anti-aliasing sampling is performed on the endpoint surface perceived by the driver's naked eye. This helps ensure that the selected sampling point 10 accurately corresponds to the actual geometric boundary of the 3D virtual navigation scene (such as AR lane lines) on the virtual image surface, thus significantly improving the coarse jagged edges on the virtual image surface and making the projected 3D markings smoother.

[0077] This embodiment embeds the key optical parameter of the imaging system, namely the magnification A, as a core factor directly into the geometric calculation formula of the sampling grid W0, enabling the image rendering driving method to possess cross-platform adaptability. Regardless of how the optical system of the automotive HUD iterates or the magnitude of the magnification A, the graphics rendering system only needs to adjust this parameter to automatically adapt and generate an accurate sampling grid W0, without needing to redefine a new algorithm. This enhances the universal value of the solution in markets such as automotive.

[0078] Please refer to Figure 5 or Figure 8 In one optional embodiment of this disclosure, when sampling point 10 is selected inside sampling grid W1, the position of sampling point 10 is located inside the geometric boundary mapped by multiple virtual sub-pixels P2 in the corresponding sampling grid W1 within the display range of the screen.

[0079] Although the aforementioned implementation indicates that the sampling points 10 are preferably scattered around the center point, this solution further limits the specific spatial range of the sampling points 10. The selected sampling points 10 are located inside the overall external geometric boundary of the sampling grid W1, which is composed of multiple (e.g., three) virtual sub-pixels P2 and mapped as a whole within the display area of ​​the screen.

[0080] This scheme sets the activity range of sampling point 10 within the entire sampling grid W1 (i.e., the macroscopic boundary containing multiple virtual sub-pixels P2 of red, green, and blue), allowing at least two sampling points 10 to flexibly and symmetrically span the four corners or edges of the sampling grid. When the boundary of an object in the 3D scene cuts through this sampling grid W1, these sampling points 10, which span the internal boundaries of sub-pixels but are within the overall boundary of the grid, can capture the actual grayscale situation with higher spatial resolution. This is beneficial for providing more global sub-pixel level energy distribution information for the entity sub-pixels mapped by the sampling grid during the rendering grayscale calculation in step S3, thereby effectively improving the problems of local image discontinuity and geometric edge jaggedness caused by 3D beam splitting.

[0081] Since the sampling point 10 is selected within the entire geometric boundary of the sampling grid W1, it provides a sufficiently wide geometric degree of freedom for its spatial arrangement. The sampling point 10 can be arranged in a regular polygonal ring within the entire grid, such as at equal intervals or angles, without being forced to crowd into a small monochrome sub-pixel area. This more reasonable spatial discrete distribution within the overall grid boundary helps the sampling point 10 to evenly cover the grayscale variation trends of each color channel within the grid, making the sampling data more uniform and continuous in spatial dimension. This helps prevent high-frequency sampling noise caused by excessive clustering and narrow spatial positions of the sampling point 10, thereby improving the smoothness and warmth of edge colors and brightness transitions in the 3D image.

[0082] Figure 13 The diagram illustrates a process of selecting sampling point 10 within the sampling grid W0 for sampling and rendering the image based on the sampling results. Please refer to the diagram. Figure 8 and Figure 13 In one optional embodiment of this disclosure, sampling points 10 are selected within the sampling grid W0 for sampling, and image rendering is performed based on the sampling results, including: Step S31-1: Divide the sampling grid W1 based on the first target virtual sub-pixel PM1. The first target virtual sub-pixel PM1 is located in the middle position inside the sampling grid W1. Set and select at least two sampling points 10 for the first target virtual sub-pixel PM1 and obtain the sampling information at each sampling point 10. Step S31-2: Based on the sampling information, calculate the first driving data or first compensation information of the entity sub-pixel P1 corresponding to the first target virtual sub-pixel PM1; Step S31-3: Provide the first driving data or the first compensation information to the physical sub-pixel P1 corresponding to the first target virtual sub-pixel PM1 in the sampling grid W1.

[0083] In the method provided in this embodiment, when dividing the sampling grid W0, a virtual sub-pixel P2 of a specific emitted light color is first selected as a global or local absolute reference, defined as the first target virtual sub-pixel PM1. In a practical full-color display system, for example, the green virtual sub-pixel P22, which contributes the most to human visual brightness perception, can be selected as the first target virtual sub-pixel PM1. The sampling grid W1 is divided using this as the geometric center or primary alignment reference. Within a single sampling grid W1, the first target virtual sub-pixel PM1 is positioned at the center of the grid. Then, specifically for this first target virtual sub-pixel PM1 located at the very center of the grid, at least two (or more) sampling points 10 are set and selected within its corresponding sampling grid range, and the original scene sampling information at these sampling points 10 is obtained.

[0084] This scheme explicitly specifies that the first target virtual sub-pixel PM1 is the reference, and it is located in the center of the grid. Under this geometric configuration, the at least two sampling points 10 set for the first target virtual sub-pixel PM1 can be naturally enclosed within the sampling grid range that expands around the center point, and are not easily affected by spatial physical boundary interference from other adjacent sampling grids or external irrelevant high-frequency signals. This helps ensure that the information from the at least two sampling points 10 has high fidelity and spatial local specificity. This helps prevent erroneous grayscale or pseudo-colors from irrelevant areas outside the grid from affecting the current first target entity sub-pixel P1 due to the uncontrolled spatial span of the sampling points 10 during anti-aliasing rendering. Thus, it helps to perfectly preserve the high dynamic range contrast and color saturation that the target color channel should have while significantly improving the edge jaggedness of the 3D image.

[0085] When the first target virtual sub-pixel PM1 is a green virtual sub-pixel P22, considering that in human visual physiology, the retina's sensitivity to different wavelengths of light varies significantly, with a much higher sensitivity to green light than to red and blue light, the green channel determines more than 70% of the macroscopic contours and contrast of the image. This scheme specifically uses the first target virtual sub-pixel PM1 (e.g., a green pixel) as a reference to establish a sampling grid W1, placing it in the most central position for multi-point oversampling. This helps ensure that the image rendering system can concentrate the most crucial and valuable sampling frequency resources on the color channel that has the greatest impact on visual quality. Due to the central position of the sampling grid for the first target virtual sub-pixel PM1, its at least two sampling points 10 can capture the edge-cutting abrupt changes of the three-dimensional geometric contour with high spatial isotropy. This allows the final generated driving data to perform delicate visual grayscale smoothing modulation on the object edges, thereby significantly improving the local brightness discontinuities and rough step-like effects caused by physical viewpoint beam splitting in 3D mode.

[0086] Figure 14 The diagram illustrates another workflow for selecting sampling point 10 within the sampling grid W0, performing sampling, and rendering the image based on the sampling results. Please refer to [the diagram]. Figure 14 In one optional embodiment of this disclosure, after step S31-3, step S31-4 is further included: based on the sampling information at sampling point 10, the driving data or compensation information of the entity sub-pixel P1 corresponding to other virtual sub-pixels P2 in sampling grid W1 is calculated, and the driving data or compensation information is provided to the entity sub-pixel P1 corresponding to other virtual sub-pixels P2.

[0087] This scheme uses the first target virtual sub-pixel PM1 (e.g., green) as the reference to divide the grid, and allows red and blue pixels to share the same sampling point 10 information. This helps ensure that within a sampling grid W1, the spatial grayscale evolution trends of the three primary colors (red, green, and blue) are constrained within the same visual geometric reference frame. This helps enhance the synchronous spatial transition of the three primary color energies at object edges, suppresses color separation, and ensures the cleanliness of color edges. Furthermore, there is no need to re-divide the sampling grid for other virtual sub-pixels P2, which helps reduce the power consumption of the image rendering system.

[0088] In one optional embodiment of this disclosure, the sampling information includes pixel grayscale values, which include pixel grayscale values ​​corresponding to different virtual sub-pixels P2; in steps S31-4 above, based on the sampling information at sampling point 10, the driving data or compensation information of the physical sub-pixels P1 corresponding to other virtual sub-pixels P2 within the sampling grid W0 is calculated, including: Obtain the pixel grayscale value corresponding to any other virtual sub-pixel P2, and calculate the driving data or compensation information of the physical sub-pixel P1 corresponding to the virtual sub-pixel P2 based on the pixel grayscale values ​​corresponding to at least two sampling points 10.

[0089] In the anti-aliasing rendering process of an image rendering system, each time a grayscale value is obtained from sampling point 10, the hardware needs to perform an addressing and data reading process. If different virtual sub-pixels P2 (red, green, blue, etc.) independently select sampling point 10 and read data, the video memory bandwidth will increase linearly, easily leading to chip overload. This solution allows other virtual sub-pixels P2 in sampling grid W1 to directly share and reuse the full-channel pixel grayscale values ​​at at least two sampling points 10 corresponding to the first target virtual sub-pixel PM1. This helps to directly and sharply reduce the total number of video memory reads in a single sampling grid W1 or region (for example, directly reducing the read overhead by 2 / 3 in an RGB three-primary-color system). This helps to alleviate the bus bandwidth pressure of the image rendering system and reduce the real-time computational load of the dynamic anti-aliasing algorithm on the automotive chip or display controller.

[0090] Figure 15The diagram shows another workflow for selecting sampling point 10 within the sampling grid W0, performing sampling, and rendering the image based on the sampling results. Figure 16 The diagram shows another schematic of dividing a virtual sub-pixel P2 in a sub-viewport V0 into a sampling grid W0. Figure 16 The example is illustrated using the case where the second color virtual sub-pixel P2 is located at the center of the sampling grid W0. Please refer to [link / reference needed]. Figure 15 and Figure 16 In one optional embodiment of this disclosure, the driving method further includes: Step S31-5: Redivide the sampling grid W1 based on the second target virtual sub-pixel PM2, wherein the second target virtual sub-pixel PM2 is located in the middle position inside the redivised sampling grid W1, and the color of the second target virtual sub-pixel PM2 is different from that of the first target virtual sub-pixel PM1. Step S31-6: Set and select at least two sampling points 10 for the second target virtual sub-pixel PM2, and calculate the second driving data or second compensation information of the entity sub-pixel P1 corresponding to the second target virtual sub-pixel PM2; Step S31-7: Provide the second driving data or the second compensation information to the physical sub-pixel P1 corresponding to the second target virtual sub-pixel PM2 within the sampling grid W0.

[0091] This embodiment no longer reuses the grayscale value of sampling point 10 corresponding to the first target virtual sub-pixel PM1. Instead, it uses the second target virtual sub-pixel PM2 (such as the red virtual sub-pixel P21) with a completely different emitted light color as a reference, and re-divides a set of sampling grids specifically for the color channel corresponding to the second target virtual sub-pixel PM2. Within the re-dividated grids, at least two sampling points 10 are set and selected separately for the second target virtual sub-pixel PM2, and its second driving / compensation information is calculated and output independently.

[0092] Considering that the geometric center of the second virtual target sub-pixel is completely different from that of the first virtual target sub-pixel, if the sampling point 10 corresponding to the first virtual target sub-pixel is reused, the sampling point 10 is not symmetrical with respect to the geometric center of the second target virtual sub-pixel PM2. In some extreme scenarios, such as when facing extremely high-frequency fine monochrome lines, some microscopic jaggedness may still remain. This solution establishes a separate sampling grid W1 for the second target virtual sub-pixel PM2 and samples it in the new sampling grid W1 with the geometric center of the second target virtual sub-pixel PM2. This ensures that both the first target virtual sub-pixel PM1 and the second target virtual sub-pixel PM2 can enjoy symmetrical and high-precision oversampling calculations with sampling point 10 in visual space. This helps the physical entity sub-pixel P1 corresponding to the first target virtual sub-pixel PM1 and the second target virtual sub-pixel PM2 to obtain more perfect edge grayscale modulation, and is more conducive to eliminating microscopic geometric jaggedness at the junction of any monochrome image and different colors, achieving smoothness and color transition texture.

[0093] Figure 17 The diagram shows another schematic of dividing a virtual sub-pixel P2 in a sub-viewport V0 into a sampling grid W0. Figure 17 The example is illustrated using the case where the third color virtual sub-pixel P2 is located at the center of the sampling grid W0. Please refer to [link / reference needed]. Figure 17 In one optional embodiment of this disclosure, the driving method further includes: dividing the sampling grid W1 again based on the third target virtual sub-pixel PM3 (such as blue virtual sub-pixel P23), wherein the third target virtual sub-pixel PM3 is located in the middle position inside the divided sampling grid W1, and the colors of the first target virtual sub-pixel PM1, the second target virtual sub-pixel PM2 and the third target virtual sub-pixel PM3 are different. For the third target virtual sub-pixel PM3, set and select at least two sampling points 10, and calculate the third driving data or third compensation information of the entity sub-pixel P1 corresponding to the third target virtual sub-pixel PM3; The third driving data or the third compensation information is provided to the physical sub-pixel P1 corresponding to the third target virtual sub-pixel PM3 within the sampling grid W0.

[0094] This embodiment no longer reuses the grayscale values ​​of sampling points 10 corresponding to the first target virtual sub-pixel PM1 or the second target virtual sub-pixel PM2. Instead, it uses the third target virtual sub-pixel PM3 (such as the blue virtual sub-pixel P23), which has a completely different emitted light color, as a benchmark to re-divide a set of sampling grids W1 dedicated to the blue channel. Within the re-dividated grid, at least two sampling points 10 are set and selected separately for the blue sub-pixel, and its third driving data / third compensation information is calculated and output independently.

[0095] Considering that the geometric center of the third target virtual sub-pixel PM3 is completely different from that of the first target virtual sub-pixel PM1 or the second target virtual sub-pixel PM2, if the sampling point 10 corresponding to the first target virtual sub-pixel PM1 or the second target virtual sub-pixel PM2 is reused, the sampling point 10 is not symmetrical with respect to the geometric center of the third target virtual sub-pixel PM3. In some extreme scenarios, such as when facing extremely high-frequency fine monochrome lines, some microscopic jagged edges may still remain. This solution ensures that the first target virtual sub-pixel PM1, the second target virtual sub-pixel PM2, and the third target virtual sub-pixel PM3 can all enjoy symmetrical and high-precision oversampling calculations with sampling point 10 in visual space by independently establishing a sampling grid W1 for the third target virtual sub-pixel PM3 and sampling in the new sampling grid W1 with the geometric center of the third virtual sub-pixel P2. This helps to ensure that the physical entity sub-pixels P1 corresponding to the first target virtual sub-pixel PM1, the second target virtual sub-pixel PM2, and the third target virtual sub-pixel PM3 can all obtain relatively perfect edge grayscale modulation, which is more conducive to eliminating micro-geometric jaggedness at the junction of any monochrome image and different colors, and achieving smoothness and color transition texture.

[0096] Please refer to Figure 2 and Figure 5 In one optional embodiment of this disclosure, the main viewing area V includes at least two sub-viewing areas V0, and the sampling grid W0 is divided within the screen display range corresponding to at least one sub-viewing area V0, including: dividing the sampling grid W0 within the screen display range corresponding to each sub-viewing area V0 respectively.

[0097] The main viewport V is composed of two or more sub-viewports V0 in space. When executing the driving method, for each sub-viewport V0 contained in the main viewport V, a set of sampling mesh W0 is independently divided within its corresponding screen display area, and a full set of oversampling rendering calculations are performed.

[0098] This approach facilitates seamless stereoscopic vision coverage across the entire spatial range, ensuring a smooth, high-quality anti-aliasing experience for the observer regardless of head movement within the main viewpoint V. In many 3D displays, the observer's head and eyes move freely in space. If the system only meshes a single sub-viewpoint V0, a ​​slight head tilt or shift of the user's eye to another sub-viewpoint V0 will instantly reveal severe aliasing and jagged edges without anti-aliasing processing. This solution, by dividing and rendering sampling meshes W0 within each sub-viewpoint V0, effectively lays a complete anti-aliasing protection layer across the entire spatial eyebox. This ensures that regardless of how the observer moves left or right or changes their viewing angle, the 3D visual image received by their left and right eyes at any spatial coordinate remains consistently smooth and alias-free, significantly reducing visual artifacts during viewpoint transitions.

[0099] Figure 18 The diagram shown represents a correspondence between the human eye and sub-viewing region V0. Please refer to it. Figure 18 In one optional embodiment of this disclosure, the main viewing area V includes a first sub-viewing area and a second sub-viewing area. The coordinates of the left eye of the human eye are located within the coordinate range of the first sub-viewing area (such as sub-viewing area V2), and the coordinates of the right eye of the human eye are located within the coordinate range of the second sub-viewing area (such as sub-viewing area V4). Divide the screen display area corresponding to at least one sub-view area V0 into sampling grids W0, including: The position of the eyes is monitored, and the coordinates of the left and right eyes are obtained; Based on the left and right eye coordinates, obtain the positions of the first sub-view area (sub-view area V2 is used as an example in the figure) and the second sub-view area (sub-view area V4 is used as an example in the figure); Sampling grids are divided within the display area corresponding to the first and second sub-view areas respectively.

[0100] The system utilizes hardware such as infrared cameras and eye-tracking sensors to capture the spatial position of the observer's face and eyes in real time and at high frequency, accurately calculating the current three-dimensional physical coordinates of the left eye (left eye coordinates) and the right eye (right eye coordinates). The image rendering system can dynamically calculate and lock the current left eye coordinates within the spatial span of a physical sub-view area V0 (marking it as the target's first sub-view area), and simultaneously lock the current right eye coordinates within the spatial span of a physical sub-view area V0 (marking it as the target's second sub-view area).

[0101] In this embodiment, within the display area corresponding to the first and second sub-view areas corresponding to the current spatial position of the human eye, a sampling grid is dynamically divided, and subsequent oversampling anti-aliasing calculations are initiated. When the human eye views a 3D display, although the hardware can form many sub-view areas V0 in space (such as 5, 8, or 16), in any extremely small instantaneous time slice, the driver's eye can only actually be located in one or two of these sub-view areas V0. The light emitted from the vast majority of the other spatial sub-view areas V0 does not actually enter the human eye but is projected into other empty backgrounds. If blind rendering of the entire view area is used, a large amount of computing power will be completely wasted on pixels that no one sees. This solution uses precise dynamic positioning based on the human eye coordinates to directly limit the division range of the anti-aliasing sampling grid W0 to the two local channels of the first and second sub-view areas that the human eye is currently viewing. This allows the system to eliminate oversampling calculations for all other non-target view areas, increasing the effective computing power utilization of the image rendering system several times over. It also significantly reduces the heat generation and energy consumption of the image rendering system, greatly extending the lifespan of the hardware.

[0102] When the driver's eyes shift significantly in space due to vehicle bumps or changes in seating posture, the sampling grid automatically and seamlessly switches and re-divides between different physical sub-viewing areas, following the changes in the driver's eye coordinates due to high-frequency triggering of eye monitoring. This helps ensure that the highest quality anti-aliasing image is tracked and locked onto the user's left and right eyes, eliminating occasional fleeting jagged edges or interruptions in stereoscopic perception caused by body movement, and providing a natural, physiologically compliant 3D comfortable visual experience.

[0103] Please continue to refer to this. Figure 18 In one optional embodiment of this disclosure, the main viewing area V further includes at least one third sub-view area (e.g., sub-view area V1, V3, or V5) in addition to the first sub-view area (e.g., sub-view area V2) and the second sub-view area (e.g., sub-view area V4); dividing the sampling grid within the screen display range corresponding to at least one sub-view area includes: dividing the sampling grid only within the screen display range corresponding to the first sub-view area and the second sub-view area.

[0104] In addition to the first and second sub-view areas corresponding to the human eye, the main view area V also includes a third sub-view area. During image rendering, only the first and second sub-view areas corresponding to the human eye are sampled and rendered, while other sub-view areas are not sampled and rendered. This allows the valuable chip hardware resources to be dedicated to serving the first and second sub-view areas that the observer's eyes can actually see. It also helps to push the anti-aliasing image quality and frame rate of the current binocular image to the possible extreme under the constraint of extremely limited total hardware computing power.

[0105] Figure 19The diagram shown illustrates a method for dividing the screen display area corresponding to at least one sub-viewport V0 into a sampling grid W0. Please refer to the flowchart. Figure 2 and Figure 19 In one optional embodiment of this disclosure, the multiple sub-view areas V0 in the main view area V are pre-defined static spatial regions; Divide the display area of ​​at least one sub-view zone V0 into a sampling grid, including: Step S21: Receive view control instructions and determine at least one target sub-view area (e.g., sub-view areas V1 and V3) and at least one non-target sub-view area (e.g., at least one of sub-view areas V2, V4 and V5) from multiple static spatial regions according to the view control instructions. Step S22: Divide the sampling grid W0 only within the screen display range corresponding to the target sub-view area.

[0106] In the hardware to which this solution is applicable, the multiple sub-view areas V0 (such as V1 to V5) within the main viewing area V are not infinitely and continuously finely adjusted with the position of the human eye, but are static spatial areas scattered in different fixed positions in space, which are precisely set in advance by the hardware at the factory or during optical calibration.

[0107] The grid triggering mechanism of this solution does not rely on high-frequency continuous human eye coordinate calculation, but rather on receiving perspective control commands distributed from external sources or the system's underlying layer (these commands can be triggered once by a lightweight human eye area locator, or by user interaction buttons, or even by the current vehicle's driving mode command).

[0108] Upon receiving the view control command, the system quickly selects and identifies the target sub-view area that urgently needs to be displayed from these preset static spatial regions through a rapid lookup table mapping, and marks all remaining static areas as non-target sub-view areas. Subsequently, sampling grids are divided and image rendering is performed only within the display area corresponding to the identified static target sub-view areas, while non-target sub-view areas are not divided.

[0109] Continuous 3D eye tracking typically requires dedicated chips with high computing power and high-speed infrared cameras, which significantly increases costs. This solution cleverly utilizes pre-defined static spatial regions and view control commands for directional anti-aliasing control. This means the system only needs to execute coarse-grained, low-frequency view switching control (e.g., triggering a view control command only once after the driver makes a minor adjustment to the seat) to instantly and accurately map the anti-aliasing sampling mesh to the corresponding static target area. This allows for the achievement of localized directional anti-aliasing effects without requiring high-specification continuous tracking hardware.

[0110] The above approach improves the ease of adaptive control of the algorithm in multi-user interaction or multi-mode switching display scenarios. For example, when the vehicle switches from driver-only mode to passenger-side entertainment mode, the view control command can move the target sub-view area directly from the driver's static area to the passenger's static area with a single click. The system then automatically and uniquely divides the sampling grid within the corresponding passenger-side sub-view area. This mode, based on static areas and direct command-driven operation, greatly simplifies the software control hierarchy under multi-user interaction and improves the stability of the switching drive method.

[0111] In one optional embodiment of this disclosure, the display panel includes multiple display frames, and in each display frame, a sampling grid is divided within the screen display range corresponding to at least one sub-view area and image rendering is performed.

[0112] In the dynamic image output of a display device, a smooth video stream or dynamic interactive animation is composed of multiple display frames (such as 60, 90 or 120 frames per second) that are arranged continuously in time and rendered at a high frequency.

[0113] The driving method of the image rendering system provided in this embodiment does not involve the aforementioned steps S2 (mesh generation) and S3 (oversampling calculation and rendering) being executed statically all at once when the system starts up, nor is it sampled and executed only at intervals of several frames. Instead, it uses frame-level clock cycles as the basic driving unit, performing sampling mesh generation and image rendering calculation based on oversampling point results in each independent display frame of the video stream or dynamic image.

[0114] In in-vehicle 3D HUDs or 3D game visuals, the vast majority of displayed content is in a state of rapid, high-speed motion (such as rapidly rewinding 3D navigation street scenes, frequently changing speed figures, flashing warning signs, etc.). If the anti-aliasing sampling mesh and sampling execution are discontinuous on the timeline (e.g., updated only every few frames), then when a 3D object undergoes high-speed displacement between frames, the relative phase of its physical edges cutting across the sampling mesh will change drastically, resulting in extremely glaring edge flickering and ghosting in dynamic scenes. This solution sets mesh generation and rendering calculations to be refreshed throughout each display frame, ensuring that at every minute point in the object's movement, the macroscopic oversampling mesh in the monocular visual space can be precisely aligned with the latest 3D geometric boundaries of that frame for accurate grayscale modulation. This helps to reduce the edge jaggedness of dynamic images to an extremely low level, thus contributing to a smooth, fluid, and stable dynamic 3D stereoscopic image quality experience.

[0115] Please refer to Figure 12In one optional embodiment of this disclosure, the image rendering system further includes a prism assembly 30 disposed on the light-emitting side of the display panel. The prism assembly 30 includes a plurality of prisms 31 arranged along a third direction D3. A plurality of physical sub-pixels P1 on the display panel are arranged along a first direction D1 to form an actual sub-pixel row H0. In a direction parallel to the plane where the display panel is located, the third direction D3 and the first direction D1 form a first acute angle.

[0116] The image rendering system provided in this disclosure is applied to 3D display products. Solid sub-pixels P1 on the display panel are arranged along a first direction D1 to form multiple actual sub-pixel rows H0, which can be arranged along a second direction D2. In a prism assembly 30 disposed on the light-emitting side of the display panel, multiple strip-shaped prisms 31 are arranged along a third direction D3. The arrangement direction of the prisms 31 in the prism assembly 30 (third direction D3) forms an acute angle with the arrangement direction of the actual sub-pixels in the actual sub-pixel rows H0 of the display panel (first direction D1).

[0117] If prism 31 is placed perpendicular to the first direction D1, it can only deflect light to different viewpoints in the horizontal direction, while having no light-splitting capability in the vertical direction. This leads to an extremely unbalanced characteristic in the 3D display device; for example, the horizontal pixels are severely split, resulting in a significant reduction in single-eye resolution, while the vertical resolution is excessive, and large vertical blind slits are easily generated. This optional embodiment introduces an oblique arrangement of prisms, using the tilt angle to cleverly decompose the light-splitting capability of prism 31 into geometrical components in both the horizontal and vertical axes. This allows 3D display products to simultaneously possess multi-viewpoint and multi-angle spatial light field emission capabilities in both the horizontal and vertical directions, improving the 3D display effect.

[0118] The presence of the aforementioned acute tilt angle allows the non-emitting physical black seams between adjacent physical sub-pixels P1 to be uniformly diluted, staggered, and dispersed into different corners of space after being amplified by prism 31. This prevents the black seams from converging into large black patterns at a specific viewing angle. Combined with the frame-level, cross-pixel macro-grid oversampling anti-aliasing rendering implemented in the software layer of this disclosure, the visual rasterism and color moiré patterns caused by the physical black seams can be diluted and reduced to a barely perceptible state by the naked eye with extremely high efficiency. This greatly improves the detail and smoothness of naked-eye 3D or HUD images.

[0119] Based on the same inventive concept, this disclosure also provides an image rendering system 100. Figure 20 The diagram shown is a structural schematic of an image rendering system provided in an embodiment of this disclosure. Please refer to it. Figure 20 as well as Figures 2 to 6 The image rendering system 100 includes: The first module 101 is used to divide multiple physical sub-pixels P1 on the display panel into at least two pixel groups Z0 according to their spatial visibility in the main viewing area V. Different pixel groups Z0 correspond to different sub-viewing areas V0 of the main viewing area V for display. Among them, multiple physical sub-pixels P1 whose emitted light beams converge to the same sub-viewing area V0 are located in the same pixel group Z0. The second module 102 is used to divide the screen display area corresponding to at least one sub-view area V0 into a sampling grid W0; The third module 103 is used to select sampling points 10 within the sampling grid W0 for sampling and to perform image rendering based on the sampling results.

[0120] It should be noted that the image rendering system provided in this disclosure has the same technical effect as the driving method of the aforementioned image rendering system, and the similarities will not be repeated here.

[0121] In practical applications, the three modules in the image rendering system 100 can cooperate to process data from different display frames. For example, while the first module 101 processes the pixel space reorganization of the (K+2)th frame (as described in step S1 above), the second module 102 can calculate the dynamic mesh boundary of the (K+1)th frame in parallel (as described in step S2 above), and the third module 103 is sampling and rendering the image at the sampling point 10 of the Kth frame (as described in step S3 above). This modular design helps to reduce the overall latency of the system's screen output, eliminates the risk of high jitter, stuttering, and frame drops inherent in pure software algorithm operation, and ensures extremely high operational stability for in-vehicle or consumer-grade 3D systems.

[0122] In this embodiment, the first module 101, the second module 102, and the third module 103 work together to construct an image rendering strategy within the monocular visual perception space of the human eye, without altering the semiconductor process of the display panel, through a purely optimized integrated hardware and software driving device. This helps to improve and suppress a series of serious core image quality defects in traditional 3D display and 3D HUD technologies, such as jagged geometric edges, uneven spatial color mixing, color edge burrs, and moiré patterns, which are naturally caused by prism 31 beam splitting and optical path breaks. This helps to make the dynamic 3D AR virtual image or the three-dimensional stereoscopic scene presented in front of the driver more delicate, soft, smooth, and with clear and sharp edges. This not only greatly enhances the technological aesthetics and immersive interactive experience of stereoscopic vision, but also, for in-vehicle HUDs, helps to reduce visual fatigue and distraction caused by jagged or flickering text or lane lines, thereby significantly improving the overall safety factor of driving interaction.

[0123] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0124] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A driving method for an image rendering system, characterized in that, include: Multiple physical sub-pixels on the display panel are divided into at least two pixel groups according to their spatial visibility in the main viewing area. Different pixel groups correspond to different sub-view areas in the main viewing area for displaying images. Among them, multiple physical sub-pixels whose emitted light beams converge to the same sub-view area are located in the same pixel group. Divide the screen display area corresponding to at least one of the sub-view areas into a sampling grid; Sampling points are selected within the sampling grid for sampling, and image rendering is performed based on the sampling results.

2. The driving method according to claim 1, characterized in that, At least two virtual sub-pixels within the same sub-view area originate from at least one physical sub-pixel in a non-adjacent block on the display panel; Alternatively, at least two entity subpixels within the same pixel group are spaced apart by entity subpixels from other pixel groups.

3. The driving method according to claim 1, characterized in that, Sampling points are selected within the sampling grid for sampling, and image rendering is performed based on the sampling results, including: At least two sampling points are selected within the sampling grid, and sampling information at each sampling point is obtained, the sampling information including pixel grayscale values; The driving data or compensation information of the corresponding entity sub-pixel is calculated based on the sampling information, and the driving data or compensation information is provided to the corresponding pixel group for image rendering.

4. The driving method according to claim 3, characterized in that, Within the display range corresponding to at least one of the sub-view areas, the sampling point is located outside the center point of the corresponding virtual sub-pixel.

5. The driving method according to claim 4, characterized in that, The number of sampling points selected within the sampling grid is N, where N ≥ 3 and N is an integer. The distance between each of the N sampling points and the center point of the corresponding virtual sub-pixel is equal.

6. The driving method according to claim 5, characterized in that, Along the circumference of the sampling grid, the distance between any two adjacent sampling points is equal.

7. The driving method according to claim 3, characterized in that, The step of calculating the driving data for the corresponding entity sub-pixel based on the sampling information includes: Obtain the grayscale values ​​at at least two selected sampling points within the sampling grid, and use them as the sampling information at each sampling point; The mean or weighted average of the grayscale values ​​at each sampling point is calculated to obtain the driving data of the corresponding entity sub-pixels.

8. The driving method according to claim 3, characterized in that, Acquiring sampling information at each sampling point includes: setting up M spatially discrete virtual reference cameras within the main view area, and acquiring reference view images corresponding to each virtual reference camera, where M is greater than or equal to the number of sub-view areas contained in the main view area; Calculate the relative distance between the sampling point and each of the virtual reference cameras; The virtual reference camera with the smallest relative distance is determined as the target reference camera, and the pixel grayscale value at the geometrically corresponding position of the sampling point in the reference view image corresponding to the target reference camera is used as the sampling information at that sampling point.

9. The driving method according to claim 8, characterized in that, The number of virtual reference cameras is the same as the number of sub-view areas in the main view area, and the virtual reference cameras are located on the center line of the screen display range of the corresponding sub-view area, with the center line perpendicular to the display surface corresponding to the screen display range.

10. The driving method according to claim 1, characterized in that, Dividing the screen display area corresponding to at least one of the sub-view areas into a sampling grid includes: virtual sub-pixels of the same color are not located in the same sampling cell of the same sampling grid.

11. The driving method according to claim 1, characterized in that, Dividing the display area of ​​at least one of the sub-view areas into a sampling grid includes: The display area corresponding to the sub-view area is divided into multiple sampling grids arranged in an array to form the sampling grid; The sampling grid includes three virtual sub-pixels with different emitted light colors within the same sampling cell. The virtual sub-pixels correspond to the physical sub-pixels in the pixel group based on spatial visibility mapping.

12. The driving method according to claim 11, characterized in that, The image rendering system further includes a prism assembly disposed on the light-emitting side of the display panel; the display panel includes a pixel unit, the pixel unit includes three physical sub-pixels of different colors, and the physical sub-pixels in the same pixel unit are arranged along a first direction; The sampling grid is a quadrilateral region with length Px and width Py, where Px=P and Py=x / sin(β); Wherein, P is the prism period of the prism assembly, x is the width of a single pixel unit on the display panel in the first direction, and β is the tilt angle of the extension direction of the prism in the prism assembly relative to the second direction, wherein the second direction is perpendicular to the first direction and parallel to the plane of the display panel.

13. The driving method according to claim 11, characterized in that, The image rendering system further includes a prism assembly disposed on the light-emitting side of the display panel and an imaging system disposed in the light-emitting direction of the prism assembly; the display panel includes a pixel unit, the pixel unit includes three physical sub-pixels of different colors, and the physical sub-pixels in the same pixel unit are arranged along a first direction; The sampling grid is a quadrilateral region with length Px and width Py, where Px = A × P and Py = Ax / sin(β); Wherein, A is the magnification of the imaging system, P is the prism period of the prism assembly, x is the width of a single pixel unit on the display panel in the first direction, and β is the tilt angle of the extension direction of the prism in the prism assembly relative to the second direction, wherein the second direction is perpendicular to the first direction and parallel to the plane of the display panel.

14. The driving method according to claim 11, characterized in that, When a sampling point is selected within the sampling grid, the position of the sampling point is located within the geometric boundary mapped by multiple virtual sub-pixels in the corresponding sampling grid within the display area of ​​the screen.

15. The driving method according to claim 11, characterized in that, Sampling points are selected within the sampling grid for sampling, and image rendering is performed based on the sampling results, including: The sampling grid is divided based on the first target virtual sub-pixel, wherein the first target virtual sub-pixel is located in the middle position inside the sampling grid, at least two sampling points are set and selected for the first target virtual sub-pixel, and sampling information at each sampling point is obtained; Based on the sampling information, calculate the first driving data or first compensation information of the entity sub-pixel corresponding to the first target virtual sub-pixel; The first driving data or the first compensation information is provided to the entity sub-pixel corresponding to the first target virtual sub-pixel within the sampling grid.

16. The driving method according to claim 15, characterized in that, Also includes: Based on the sampling information at the sampling point, calculate the driving data or compensation information of the entity sub-pixels corresponding to other virtual sub-pixels within the sampling grid, and provide the driving data or compensation information to the entity sub-pixels corresponding to other virtual sub-pixels.

17. The driving method according to claim 15, characterized in that, The sampling information includes pixel grayscale values, and the pixel grayscale values ​​include pixel grayscale values ​​corresponding to different virtual sub-pixels; Based on the sampling information at the sampling point, calculate the driving data or compensation information of the physical sub-pixels corresponding to other virtual sub-pixels within the sampling grid, including: Obtain the pixel grayscale value corresponding to any other virtual sub-pixel, and calculate the driving data or compensation information of the physical sub-pixel corresponding to the virtual sub-pixel based on the pixel grayscale values ​​corresponding to at least two of the sampling points.

18. The driving method according to claim 15, characterized in that, Also includes: The sampling grid is re-divided based on the second target virtual sub-pixel, wherein the second target virtual sub-pixel is located in the middle position within the re-divided sampling grid, and the color of the second target virtual sub-pixel is different from that of the first target virtual sub-pixel; For the second target virtual sub-pixel, set and select at least two sampling points, and calculate the second driving data or second compensation information of the entity sub-pixel corresponding to the second target virtual sub-pixel; The second driving data or the second compensation information is provided to the physical sub-pixel corresponding to the second target virtual sub-pixel within the sampling grid.

19. The driving method according to claim 18, characterized in that, Also includes: The sampling grid is further divided based on the third target virtual sub-pixel. Within the newly divided sampling grid, the third target virtual sub-pixel is located in the middle position, and the colors of the first target virtual sub-pixel, the second target virtual sub-pixel, and the third target virtual sub-pixel are all different. For the third target virtual sub-pixel, set and select at least two sampling points, and calculate the third driving data or third compensation information of the entity sub-pixel corresponding to the third target virtual sub-pixel; The third driving data or third compensation information is provided to the entity sub-pixel corresponding to the third target virtual sub-pixel within the sampling grid.

20. The driving method according to claim 1, characterized in that, The main viewing area includes at least two sub-viewing areas, and the step of dividing the screen display area corresponding to at least one of the sub-viewing areas into a sampling grid includes: A sampling grid is divided within the display area corresponding to each sub-view area.

21. The driving method according to claim 1, characterized in that, The main visual area includes a first sub-visual area and a second sub-visual area. The coordinates of the left eye of the human eye are located within the coordinate range of the first sub-visual area, and the coordinates of the right eye of the human eye are located within the coordinate range of the second sub-visual area. The step of dividing the display area of ​​at least one of the sub-view areas into a sampling grid includes: The position of the eyes is monitored, and the coordinates of the left eye and the right eye are obtained; Based on the left eye coordinates and the right eye coordinates, the positions of the first sub-view area and the second sub-view area are obtained; Sampling grids are divided within the display area corresponding to the first sub-view area and the second sub-view area, respectively.

22. The driving method according to claim 21, characterized in that, The main viewing area also includes at least one third sub-view area in addition to the first sub-view area and the second sub-view area; Dividing the sampling grid within the display area corresponding to at least one of the sub-view areas includes: dividing the sampling grid only within the display area corresponding to the first sub-view area and the second sub-view area.

23. The driving method according to claim 1, characterized in that, The multiple sub-view areas within the main view area are pre-defined static spatial regions; The step of dividing the screen display area corresponding to at least one of the sub-view areas includes: receiving a view control command, and determining at least one target sub-view area and at least one non-target sub-view area from multiple static spatial regions according to the view control command; The sampling grid is divided only within the display area corresponding to the target sub-view area.

24. The driving method according to claim 1, characterized in that, The display panel includes multiple display frames. In each display frame, a sampling grid is divided within the screen display area corresponding to at least one of the sub-view areas, and image rendering is performed.

25. The driving method according to claim 1, characterized in that, The image rendering system further includes a prism assembly disposed on the light-emitting side of the display panel, the prism assembly including a plurality of prisms arranged along a third direction; a plurality of physical sub-pixels on the display panel are arranged along a first direction to form an actual sub-pixel row; in a direction parallel to the plane where the display panel is located, the third direction and the first direction form a first acute angle.

26. An image rendering system, characterized in that, The image rendering system is driven by any one of the driving methods described in claims 1 to 25, and includes: The first module is used to divide a plurality of physical sub-pixels on the display panel into at least two pixel groups according to their spatial visibility in the main viewing area, and different pixel groups correspond to different sub-view areas in the main viewing area for displaying the image; wherein, the plurality of physical sub-pixels whose emitted light beams converge to the same sub-view area are located in the same pixel group; The second module is used to divide the screen display area corresponding to at least one of the sub-view areas into a sampling grid; The third module is used to select sampling points within the sampling grid for sampling and to perform image rendering based on the sampling results.

27. An image rendering system, characterized in that, include: The first module is used to divide a plurality of physical sub-pixels on the display panel into at least two pixel groups according to their spatial visibility in the main viewing area, and different pixel groups correspond to different sub-view areas in the main viewing area for displaying the image; wherein, the plurality of physical sub-pixels whose emitted light beams converge to the same sub-view area are located in the same pixel group; The second module is used to divide the screen display area corresponding to at least one of the sub-view areas into a sampling grid; The third module is used to select sampling points within the sampling grid for sampling and to perform image rendering based on the sampling results.

Images