Exhibition hall dynamic content adaptive display method based on AI vision generation

Abstract

The present application relates to the technical field of multimedia exhibition hall, and more particularly to an exhibition hall dynamic content adaptive display method based on AI vision generation. Whether to switch display perspective sequence is determined based on scene attention distribution tendency; whether to perform interpolation output for multi-view display perspective sequence is determined based on adjacent frame change value; whether to correct the deconvolution parameters of the crosstalk coverage area is determined based on the base abnormal distribution value to determine the rendering stability tendency; after completing the area identification of the crosstalk coverage, whether to correct the picture of the texture dense area is determined based on the heat flow disturbance risk value. The present application adopts a lightweight visual perception and rendering strategy to reduce deployment cost; at the same time, it has the perception and compensation ability for the roughness of the projection base and the heat flow disturbance, effectively guarantees the picture stability in the complex exhibition hall environment, thereby improving the display efficiency.

Description

Technical Field

[0001] This invention relates to the field of multimedia exhibition hall technology, and in particular to an adaptive display method for dynamic exhibition hall content based on AI vision generation. Background Technology

[0002] Current exhibition hall multimedia systems mostly employ pre-programmed playback schemes, outputting content in a fixed order, with viewing angles and switching paths set during setup. Displays are linked via a central control system, with large exhibition halls using multi-projection splicing and geometric correction to expand the display area. Some systems incorporate sensors to detect visitors for simple interactions such as starting up and switching languages. Display viewing angles are preset during production and remain unchanged during playback.

[0003] Chinese Patent Publication No. CN122044346A discloses a management system for intelligent digital human-based interactive explanation in exhibition halls, based on spatial positioning. The system's startup and initialization phase includes: a dynamic knowledge engine acquiring the latest exhibit data through blockchain nodes, synchronizing academic database information under a federated learning framework, automatically generating digital human scripts, and adapting 3D actions; a full-process visitor management phase including: activation of tiered visitor services; real-time spatial positioning and interaction using multimodal positioning for real-time tracking, and adaptive interaction based on intent recognition; cross-device content execution; dynamic intervention for ethical risks, correcting erroneous information, and employing an anti-addiction mechanism; and a closed-loop optimization and maintenance phase. However, this technical solution suffers from the following problems: reliance on heavy hardware such as digital human modeling and high-precision positioning, resulting in high deployment costs; and a lack of perception and compensation for projection substrate roughness and thermal airflow disturbances, making it difficult to guarantee image stability in complex exhibition hall environments. Summary of the Invention

[0004] To address these issues, this invention provides an AI-based visual generation method for adaptive display of dynamic content in exhibition halls. This method overcomes the problems of existing technologies that rely on heavy hardware such as digital human modeling and high-precision positioning, resulting in high deployment costs. It also lacks the ability to perceive and compensate for the roughness of the projection substrate and thermal airflow disturbances, making it difficult to guarantee image stability in complex exhibition hall environments.

[0005] To achieve the above objectives, this invention provides a method for adaptive display of dynamic exhibition content based on AI vision generation, comprising: Determine whether to switch the perspective sequence based on the scene's attention distribution trend; Based on the change values ​​between adjacent frames, determine whether to perform interpolation output for multi-view perspective sequences; Based on the baseline anomaly distribution values, the rendering stability tendency is determined to determine whether to correct the deconvolution parameters in the crosstalk coverage area; After identifying the areas covered by crosstalk, determine whether to correct the image in areas with dense texture based on the thermal flow disturbance risk value.

[0006] Furthermore, the process of determining the distribution of scene attention includes: Based on two-dimensional thermal radiation maps, the connectivity areas of the population are determined; Identifying valid connected regions based on the number of pixels in connected regions of a crowd; The scene attention distribution tendency is determined based on each effective connected region, including the scene attention high distribution tendency and the scene attention low distribution tendency.

[0007] Furthermore, the process of determining whether to switch the display perspective sequence based on the scene's attention distribution tendency includes: When it is determined that the current scene has a high distribution of attention, switch to displaying the perspective sequence from multiple perspectives.

[0008] Furthermore, the process of determining the change values ​​between adjacent frames includes: Obtain the number of pixels whose pixel markers change between the current frame's 2D thermal radiation map and the next frame's 2D thermal radiation map, and get the change value between adjacent frames.

[0009] Furthermore, the process of switching between multiple perspective views to display the perspective sequence includes, The increase in the number of perspective view channels in a multi-view perspective sequence is positively correlated with the number of effective connected regions.

[0010] Furthermore, based on the inter-frame variation values, the process of determining whether to perform interpolation output for multi-view perspective sequences includes: Change the rendering resolution of the display video to a frame rate stable resolution, change the sharpening intensity of the display video to an anti-flicker sharpening value, or perform interpolation output for multi-view perspective sequences. The value of change between adjacent frames is the number of pixels whose pixel markers have changed between the current frame's two-dimensional thermal radiation map and the next frame's two-dimensional thermal radiation map.

[0011] Furthermore, the process of performing interpolation output for multi-view perspective sequences includes: While changing the rendering resolution of the demonstration video to a frame rate-stable resolution and the sharpening intensity of the demonstration video to an anti-flicker sharpening value, the anchor keyframes of subsequent videos are pre-rendered and cached using the initial rendering resolution and sharpening intensity, and inter-frame interpolation is performed when the corresponding demonstration video is output.

[0012] Furthermore, based on the anomalous distribution areas of the base layer, the process of determining rendering stability tendencies to determine whether to correct video in crosstalk coverage areas includes: Based on the three-dimensional height distribution information of the video projection area, the abnormal distribution area of ​​the base is determined in order to determine the crosstalk coverage area. Based on the number of abnormal distribution regions of the basal layer within the crosstalk coverage area, the rendering stability tendency is determined, and it is determined whether to perform deconvolution correction.

[0013] Furthermore, the process of determining the thermal flow disturbance risk value after completing the identification of the crosstalk coverage area includes: Obtain the regional temperature of each crosstalk coverage area and the regional temperature of the remaining sub-analysis areas, solve the variance of the temperature of each area, and obtain the heat flow disturbance risk value.

[0014] Furthermore, after identifying the areas affected by crosstalk, the process of determining whether to correct images in areas with dense textures based on thermal flow disturbance risk values ​​includes: When the thermal disturbance risk value is greater than or equal to the preset disturbance threshold, the low-pass filter cutoff frequency of the video corresponding to the crosstalk coverage area is corrected.

[0015] Compared with existing technologies, the advantages of this invention lie in its ability to dynamically decide whether to enable multi-view perspective sequences and perform interpolation rendering by analyzing scene attention distribution and inter-frame variation values ​​in real time. When scene attention distribution is high, the crowd is dispersed, and switching to multi-view perspective sequences achieves refined output from multiple perspectives. When the inter-frame variation value between adjacent frames is greater than or equal to the average inter-frame variation value, the crowd moves frequently. In this case, reducing resolution and sharpening intensity mitigates GPU overload caused by multi-angle output. When the inter-frame variation value between adjacent frames is less than the average inter-frame variation value, the crowd remains stable. Pre-rendered anchored keyframes and interpolated output with ordinary frames ensure smooth playback with high image quality and refined output of key content. Based on real-time perception of crowd behavior, the rendering strategy of the exhibition hall content is dynamically and adaptively adjusted, achieving synergistic optimization of video quality, operational stability, and energy efficiency under limited hardware resources, further improving the efficiency of exhibition hall displays.

[0016] Furthermore, due to variations in the spray gun's contact angle and pressure, the coating's micro-roughness near the zigzag surface deviates from normal levels. This increased micro-roughness broadens the light scattering function from a narrow-angle distribution to a wide-angle distribution, generating optical crosstalk between adjacent viewing channels. This leads to reduced contrast near the zigzag surface of the projected image, producing a blurry, foggy effect and weakening the multi-view stereoscopic effect. By using a white light interferometer to obtain the spatial distribution of roughness in the video projection area, the abnormal distribution value of the substrate is quantified, allowing for precise matching of the deconvolution correction intensity with the abnormal distribution value. This fully restores high-frequency details in the crosstalk-covered area, enabling a workflow from crowd behavior perception to viewpoint switching and rendering optimization, and finally to optical-physical layer correction. This collaboratively achieves joint adaptive optimization of dynamic content in the exhibition hall in terms of crowd behavior adaptation, rendering load balancing, and optical-physical compensation, further improving the exhibition hall's display efficiency and viewing experience.

[0017] Furthermore, the hot airflow exhausted from the projector forms a thermal plume in the optical path. This thermal plume converges at bends, and its turbulent structure causes fluctuations in the air's refractive index. As the projected light passes through, it undergoes slight deflection, resulting in slight spatial jitter in the projected image. Areas with dense texture in the image are perceived by the human eye as a result of this jitter, causing the viewing experience to flicker. Different color subframes of the projector exhibit color jagged edges due to experiencing varying turbulence states. By sensing the intensity of the thermal plume turbulence and quantifying it as a thermal disturbance risk value, the low-pass filter cutoff frequency is reduced in areas with dense texture at high thermal disturbance risk values ​​to eliminate image flicker and chromatic aberration. This constitutes a multi-layered joint adaptive optimization system encompassing crowd behavior perception, rendering load balancing, optical surface correction, and thermal disturbance suppression. This achieves synergistic optimization of dynamic content in the exhibition hall across multiple levels, including audience behavior adaptation, rendering resource scheduling, optical physical compensation, and environmental thermal disturbance suppression, further improving the display efficiency of large exhibition halls under complex crowd flow and installation conditions. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the steps of the AI ​​vision-based adaptive display method for dynamic exhibition content generated in an exhibition hall, as described in an embodiment of the present invention. Figure 2 This is a logic diagram for determining valid connected regions based on the number of pixels in a connected region of a crowd, as described in an embodiment of the present invention. Figure 3 This is a logic diagram for determining whether to perform interpolation output for multi-view perspective sequences based on the change values ​​between adjacent frames in an embodiment of the present invention. Figure 4 This is a logic diagram for determining whether to correct crosstalk coverage area videos based on rendering stability tendencies in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0020] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0021] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0022] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0023] Please see Figure 1 The diagram shown is a flowchart illustrating the steps of an AI-based visually generated adaptive display method for exhibition hall content, according to an embodiment of the present invention. The method described in this embodiment includes: S1, determine whether to switch the display perspective sequence based on the scene's attention distribution tendency; S2, based on the change values ​​between adjacent frames, determines whether to perform interpolation output for multi-view perspective sequences; S3, based on the basis anomaly distribution values, determines the rendering stability tendency to determine whether to correct the deconvolution parameters in the crosstalk coverage area; S4, after completing the identification of crosstalk coverage areas, determines whether to correct the image in areas with dense textures based on the thermal flow disturbance risk value.

[0024] Specifically, in S1, the process of determining the scene attention distribution includes: Based on two-dimensional thermal radiation maps, the connectivity areas of the population are determined; Identifying valid connected regions based on the number of pixels in connected regions of a crowd; The scene attention distribution tendency is determined based on each effective connected region, including a high scene attention distribution tendency and a low scene attention distribution tendency. The larger the distance between effective connected regions, the higher the dispersion of the crowd in the horizontal direction, which leads to a deviation in the perspective relationship of the scene viewed by the crowd and a weakening of the three-dimensionality of the scene.

[0025] Specifically, in S1, the process of determining whether to switch the display perspective sequence based on the scene attention distribution tendency includes: When it is determined that the scene has a high distribution of attention, switch to display the perspective sequence.

[0026] Please see Figure 2 As shown, this is a logic determination diagram for judging effective connected regions based on the number of pixels in the connected regions of the crowd, according to an embodiment of the present invention. In S1 of this embodiment, the process of determining whether to switch the display perspective sequence based on the scene attention distribution tendency includes: A two-dimensional thermal radiation map is acquired using a wide-angle thermal imaging camera, and human pixel points are marked. Adjacent human pixel points in the 2D thermal radiation map are defined as individual crowd connected regions. Crowd connected regions with a number of pixels greater than or equal to the average connectivity threshold are defined as effective connected regions to avoid misidentifying background pixels as human pixels due to localized heat accumulation in areas with slow airflow. Pixels within crowd connected regions with a number of pixels less than the average connectivity threshold are marked as background pixels. When the distance between effective connected regions is greater than or equal to a preset distance threshold, the current scene is considered to have a high attention distribution tendency, and a multi-view perspective sequence is switched to display. In this case, the crowd has a large horizontal coverage area, causing distortion and geometric proportions when viewed from the side, affecting the accuracy of information transmission; therefore, a multi-view perspective sequence is switched to display. If the distance between effective connected regions is less than the preset distance threshold, the current scene is considered to have a low attention distribution tendency, and a single-view perspective display is maintained. In this case, the crowd has a small horizontal coverage area, and a single-view perspective display ensures a good viewing experience; therefore, a single-view perspective display is maintained to reduce computational overhead and hardware resource consumption.

[0027] Human body pixels are those whose thermal radiation intensity value in the 2D thermal radiation map is greater than or equal to the lower limit threshold of human body temperature; background pixels are those whose thermal radiation intensity value in the 2D thermal radiation map is less than the lower limit threshold of human body temperature. The 2D thermal radiation map includes the thermal radiation intensity value of each pixel within the acquisition range of the wide-angle thermal imaging camera at the corresponding time. The effective connected region spacing is the maximum value among the distances between pixels in each defined effective connected region.

[0028] The lower limit threshold for human body temperature is the statistical average of normal human body surface temperature. The average connectivity threshold is the average number of pixels within connected regions of each population group in the current frame's 2D thermal radiation map. The preset region spacing threshold is half the distance between the top-left and bottom-right pixels in the current frame's 2D thermal radiation map.

[0029] Specifically, the process of switching between multiple perspective views to display a perspective sequence includes... The increase in the number of perspective view channels in a multi-view perspective sequence is positively correlated with the number of effective connected regions.

[0030] The larger the number of effectively connected areas, the larger the spatial distribution of the crowd. To ensure visual quality, the number of perspective viewing channels should be increased to cover all viewing directions.

[0031] The more perspective channels there are, the finer the viewing angles can be divided, reducing perspective distortion for marginalized groups.

[0032] The multi-view perspective sequence consists of perspective images evenly spaced by the number of perspective view channels. The images of each perspective view channel are projected sequentially by a projector, while the direction selection screen synchronously switches its transmission direction to ensure that each frame of the image is only scattered in the corresponding viewer direction, while other directions are blocked.

[0033] Specifically, in S2, the process of determining the inter-frame change value includes: The number of pixels whose pixel markers change between the current frame's 2D thermal radiation map and the next frame's 2D thermal radiation map is obtained to determine the inter-frame variation value. The lower the inter-frame variation value, the slower the overall movement of the audience, the smoother the change in spatial position, and the more stable the load distribution of each viewing channel.

[0034] Please see Figure 3 As shown, this is a logic decision diagram for determining whether to perform interpolation output for multi-view perspective sequences based on the inter-frame change values ​​in an embodiment of the present invention. In S2 of this embodiment, the process of determining whether to perform interpolation output for multi-view perspective sequences based on the inter-frame change values ​​includes: When the inter-frame variation value is greater than or equal to the average inter-frame variation value, the current state is determined to be high-fluidity change. The rendering resolution of the displayed video is changed to a frame rate stable resolution, and the sharpening intensity of the displayed video is changed to an anti-flicker sharpening value. In this case, the crowd moves frequently, the perspective sequence changes greatly, the frame alignment error is large, and the image stability is poor. At this time, the resolution is reduced to control the GPU load, and the sharpening intensity is limited to suppress image flicker, thus achieving a balance between rendering load and visual smoothness.

[0035] If the change value between adjacent frames is less than the average change value between frames, the current state is determined to be low-flow change, and interpolation output is performed for the perspective sequence displayed from multiple perspectives. In this case, the crowd moves slowly and is in a continuous viewing state, which requires high image quality and displays little change in the perspective sequence. Key sequences are pre-rendered and cached, and inter-frame interpolation is performed between adjacent cached key frames. This avoids performance fluctuations in real-time dual-path rendering while ensuring the continuity and integrity of the image.

[0036] The average inter-frame variation value is the average value of the pixel marker variation values ​​between adjacent frames in the two-dimensional thermal radiation map collected over a continuous historical period.

[0037] The frame rate stabilization resolution is the minimum critical resolution among those used to maintain frame rate stability under a multi-view perspective sequence, determined based on historical data. The anti-flicker sharpening value is the minimum sharpening intensity value among those used to prevent screen flickering before normal operation under a multi-view perspective sequence, determined based on historical data.

[0038] Specifically, in S2, the process of performing interpolation output for multi-view perspective sequences includes: While changing the rendering resolution of the demonstration video to a frame rate stable resolution and the sharpening intensity of the demonstration video to an anti-flicker sharpening value, the anchor keyframes of the subsequent video are pre-rendered and cached at the initial rendering resolution and sharpening intensity, so as to perform inter-frame interpolation when the corresponding demonstration video is output.

[0039] Among them, anchor keyframes are pre-calibrated core output content for the demonstration video. Anchor keyframes are output at the cached rendering resolution and sharpening intensity, and are output with regular frames through inter-frame interpolation to achieve alternating output of anchor keyframes and regular frames.

[0040] In this embodiment, by analyzing the scene attention distribution and inter-frame variation values ​​in real time, dynamic decisions are made on whether to enable multi-view perspective sequences and whether to perform interpolation rendering. When the scene attention distribution is high, the crowd is dispersed, and the multi-view perspective sequence is switched to achieve refined output from multiple perspectives. When the inter-frame variation value between adjacent frames is greater than or equal to the average inter-frame variation value, the crowd moves frequently. At this time, the resolution and sharpening intensity are reduced to mitigate GPU overload caused by multi-angle output. When the inter-frame variation value between adjacent frames is less than the average inter-frame variation value, the crowd remains stable. Pre-rendered anchored keyframes and ordinary frames are interpolated to ensure smooth playback with high image quality and refined output of key content. Based on real-time perception of crowd behavior, the rendering strategy of the exhibition hall content is dynamically and adaptively adjusted to achieve synergistic optimization of video quality, operational stability, and energy efficiency under limited hardware resources, further improving the efficiency of the exhibition hall display.

[0041] Specifically, in S3, the process of determining rendering stability tendencies based on the base anomalous distribution region, and then determining whether to correct the video in the crosstalk coverage area, includes: Based on the three-dimensional height distribution information of the video projection area, the base abnormal distribution area is determined in order to determine the crosstalk coverage area; Based on the number of abnormal distribution regions of the basal layer within the crosstalk coverage area, the rendering stability tendency is determined, and it is determined whether to perform deconvolution correction.

[0042] The more base abnormal distribution areas there are in the crosstalk coverage area, the larger the range of micro roughness abnormality, the wider the area affected by image fogging, and the lower the rendering stability tendency. Deconvolution correction is applied to compensate for high-frequency detail loss.

[0043] Please see Figure 4 The diagram shown illustrates the logic decision diagram for determining whether to correct crosstalk-covered video areas based on rendering stability tendencies in an embodiment of the present invention. In S3 of this embodiment, the process of determining rendering stability tendencies based on the abnormal distribution region of the base layer, and then determining whether to correct crosstalk-covered video areas, includes: The micro-roughness of each sub-analysis region in the video projection area is determined by using the three-dimensional height distribution information of the white light interferometer video projection area. If the micro-roughness is greater than or equal to the reference micro-roughness, the corresponding sub-analysis region is marked as a substrate anomalous distribution region. In this case, the coating surface undulation increases, the light scattering angle distribution widens, crosstalk between adjacent viewing angle channels is enhanced, and the risk of image fogging increases. Therefore, the corresponding sub-analysis region is marked as a substrate anomalous distribution region. Each adjacent substrate anomalous distribution region in the video projection area is determined as a single crosstalk coverage region.

[0044] If the anomalous distribution area of ​​the base in the crosstalk coverage region is greater than or equal to the average anomalous distribution threshold, the crosstalk coverage region is determined to have low rendering stability, and deconvolution correction is performed on the video in the corresponding crosstalk coverage region. In this case, the crosstalk coverage region is located on a tortuous surface with continuous rough bands, which causes multi-view scattering fogging effect due to roughness instability, affecting the viewing experience of the audience. Therefore, deconvolution correction is performed to restore the edges of scattering broadening and high-frequency information of the image, ensuring visual fidelity.

[0045] If the basal anomaly distribution area in the crosstalk coverage region is smaller than the average anomaly distribution threshold, it is determined to be a high rendering stability tendency, and the original video rendering output is maintained. In this case, the roughness is only high in a small local area, which has little impact on the stereoscopic effect of multi-view and the recognition of key data objects, so the original image is maintained to reduce computational overhead.

[0046] The reference micro-roughness is the average value of the micro-roughness values ​​corresponding to the abnormal distribution areas of each substrate within the acquisition range of the white light interferometer. The average abnormal distribution threshold is the average number of abnormal distribution areas of each substrate in each crosstalk coverage area. The average abnormal distribution threshold is used to identify the blurring of lines in the video display caused by the interconnected coverage of multiple roughness bands.

[0047] Specifically, in S3, the process of correcting the deconvolution parameters of the crosstalk coverage area includes: The base anomaly distribution values ​​of each base anomaly distribution region in the crosstalk coverage area are positively correlated with the correction strength of the deconvolution parameters, which include the standard deviation of the Gaussian kernel and the number of iterations.

[0048] The larger the substrate anomalous distribution value, the more serious the deviation of the coating's micro-roughness from the reference micro-roughness in the substrate anomalous distribution area, the more intense the micro-undulations on the coating surface, the more significant the broadening of the light scattering angle distribution, the stronger the optical crosstalk energy between adjacent viewing angle channels, and the more severe the image fogging effect. Therefore, the higher the degree of deconvolution correction, the more important it is to restore the high-frequency details weakened by scattering.

[0049] The greater the correction strength of the deconvolution parameters, the larger the standard deviation of the deconvolution Gaussian kernel, the more deconvolution iterations, the wider the spatial range of deconvolution, the more sufficient the high-frequency detail recovery depth, and the stronger the correction capability of the abnormal distribution area of ​​the substrate, so as to eliminate the fogging effect and restore the three-dimensionality of multiple views.

[0050] Among them, the substrate anomaly distribution value is the difference between the micro-roughness of the substrate anomaly distribution region and the reference micro-roughness.

[0051] In this embodiment, the coating near the zigzag surface experiences a deviation in micro-roughness from the normal level due to variations in the spray gun's contact angle and pressure. This increased micro-roughness broadens the light scattering function from a narrow-angle distribution to a wide-angle distribution, generating optical crosstalk between adjacent viewing channels. This leads to reduced contrast near the zigzag surface of the projected image, producing a blurry, foggy effect and weakening the multi-view stereoscopic effect. A white light interferometer is used to acquire the spatial distribution of roughness in the video projection area to quantify the substrate's abnormal distribution value, ensuring precise matching between the deconvolution correction intensity and the substrate's abnormal distribution value. This fully restores high-frequency details in the crosstalk-covered area, enabling a workflow from crowd behavior perception to viewpoint switching and rendering optimization, and then to optical-physical layer correction. This collaboratively achieves joint adaptive optimization of dynamic content in the exhibition hall in terms of crowd behavior adaptation, rendering load balancing, and optical-physical compensation, further improving the exhibition hall's display efficiency and viewing experience.

[0052] Specifically, in S4, the process of determining the thermal flow disturbance risk value after completing the identification of the crosstalk coverage area includes: Obtain the regional temperature of each crosstalk coverage area and the regional temperature of the remaining sub-analysis areas, solve the variance of the temperature of each area, and obtain the heat flow disturbance risk value. The higher the risk value of thermal disturbance, the more significant the heat flow concentration of the projector's exhaust airflow in the tortuous surface optical path, the more significant the non-uniformity of the air refractive index caused by heat flow concentration in the tortuous surface, the greater the deflection of the projected light, the stronger the micro-spatial jitter of the projected image, and the more flicker or color jagged edges are produced in the dense high-frequency texture areas of the image. Therefore, reducing the low-pass filter cutoff frequency and increasing the filter strength are necessary to suppress high-frequency components.

[0053] Specifically, in S4, the process of determining whether to correct the image in the texture-dense region based on the thermal flow disturbance risk value after completing the identification of the crosstalk coverage area includes: When the thermal flow disturbance risk value is greater than or equal to the preset disturbance threshold, the low-pass filter cutoff frequency of the video corresponding to the crosstalk coverage area is corrected. In this case, the intensity of thermal plume turbulence in the optical path causes amplitude deflection of the projected light, resulting in slight spatial jitter in the projected image. When there are densely textured areas in the image, the slight spatial jitter will cause image judder, and will also cause color jagged edges in the three-color subframes due to different turbulence states. Therefore, low-pass filter correction is applied to densely textured areas to suppress high-frequency components and eliminate flicker and dispersion artifacts.

[0054] When the thermal flow disturbance risk value is less than the preset disturbance threshold, the current video output parameters are maintained. In this case, the intensity of thermal plume turbulence in the optical path is within the range that is negligible to human eye sensitivity, the fluctuation amplitude of the light deflection angle corresponds to the image displacement amount less than the threshold value of human eye perception, and the flickering or color separation effect caused by slight image jitter can be ignored, so the original image output is maintained.

[0055] The regional temperature is determined by an array of temperature sensors positioned along the optical path.

[0056] The preset disturbance threshold can be the average value of each heat flow disturbance risk value when the image jitter amplitude reaches the critical value, as determined by several calibration experiments, so as to identify the existence of significant heat flow accumulation in the crosstalk coverage area.

[0057] Specifically, in S4, the process of correcting the image in areas with dense texture includes: The low-pass filter cutoff frequency of the video corresponding to the crosstalk coverage area is corrected based on the thermal flow disturbance risk value.

[0058] The risk value of thermal flux disturbance is negatively correlated with the cutoff frequency of the low-pass filter.

[0059] The lower the cutoff frequency of a low-pass filter, the stronger the filtering intensity, the more high-frequency components are filtered out, and the stronger the ability to suppress visible flicker and chromatic aberration artifacts caused by jitter in densely textured areas of the image.

[0060] The higher the risk value of thermal flow disturbance, the more significant the jitter in densely textured areas of the video projection image will be due to the instability of air refractive index caused by the accumulation of thermal flow in the crosstalk coverage area.

[0061] In this embodiment, the hot airflow exhausted by the projector forms a thermal plume in the optical path. This thermal plume accumulates at bends, and its turbulent structure causes fluctuations in the air's refractive index. As the projected light passes through, it undergoes a slight deflection, resulting in slight spatial jitter in the projected image. The densely textured areas in the image are captured by the human eye under this jitter, causing the viewing image to flicker. The different color subframes of the projector exhibit color jagged edges due to experiencing varying turbulence states. By sensing the intensity of the thermal plume turbulence and quantifying it as a thermal disturbance risk value, the low-pass filter cutoff frequency is reduced in densely textured areas under high thermal disturbance risk values ​​to eliminate image flicker and chromatic aberration. This constitutes a multi-layered joint adaptive optimization system encompassing crowd behavior perception, rendering load balancing, optical surface correction, and thermal disturbance suppression. This achieves synergistic optimization of dynamic content in the exhibition hall across multiple levels, including audience behavior adaptation, rendering resource scheduling, optical physical compensation, and environmental thermal disturbance suppression, further improving the display efficiency of large exhibition halls under complex crowd flow and installation conditions.

[0062] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0063] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for adaptive display of dynamic exhibition content based on AI vision generation, characterized in that, include: Determine whether to switch the perspective sequence based on the scene's attention distribution trend; Based on the change values ​​between adjacent frames, determine whether to perform interpolation output for multi-view perspective sequences; Based on the baseline anomaly distribution values, the rendering stability tendency is determined to determine whether to correct the deconvolution parameters in the crosstalk coverage area; After identifying the areas covered by crosstalk, determine whether to correct the image in areas with dense texture based on the thermal flow disturbance risk value.

2. The adaptive display method for dynamic exhibition content based on AI vision generation according to claim 1, characterized in that, The process of determining the distribution of attention in a scene includes: Based on two-dimensional thermal radiation maps, the connectivity areas of the population are determined; Identifying valid connected regions based on the number of pixels in connected regions of a crowd; The scene attention distribution tendency is determined based on each effective connected region, including the scene attention high distribution tendency and the scene attention low distribution tendency.

3. The adaptive display method for dynamic exhibition content based on AI vision generation according to claim 2, characterized in that, The process of determining whether to switch the display perspective sequence based on scene attention distribution includes: When it is determined that the current scene has a high distribution of attention, switch to displaying the perspective sequence from multiple perspectives.

4. The adaptive display method for dynamic exhibition content based on AI vision generation according to claim 1, characterized in that, The process of determining the change value between adjacent frames includes: Obtain the number of pixels whose pixel markers change between the current frame's 2D thermal radiation map and the next frame's 2D thermal radiation map, and get the change value between adjacent frames.

5. The adaptive display method for dynamic exhibition content based on AI vision generation according to claim 4, characterized in that, The process of switching between multiple perspective views to display a perspective sequence includes, The increase in the number of perspective view channels in a multi-view perspective sequence is positively correlated with the number of effective connected regions.

6. The adaptive display method for dynamic exhibition content based on AI vision generation according to claim 5, characterized in that, Based on the inter-frame variation values, the process of determining whether to perform interpolation output for multi-view perspective sequences includes: Change the rendering resolution of the display video to a frame rate stable resolution, change the sharpening intensity of the display video to an anti-flicker sharpening value, or perform interpolation output for multi-view perspective sequences. The value of change between adjacent frames is the number of pixels whose pixel markers have changed between the current frame's two-dimensional thermal radiation map and the next frame's two-dimensional thermal radiation map.

7. The adaptive display method for dynamic exhibition content based on AI vision generation according to claim 6, characterized in that, The process of performing interpolation output for multi-view perspective sequences includes: While changing the rendering resolution of the demonstration video to a frame rate-stable resolution and the sharpening intensity of the demonstration video to an anti-flicker sharpening value, the anchor keyframes of subsequent videos are pre-rendered and cached using the initial rendering resolution and sharpening intensity, and inter-frame interpolation is performed when the corresponding demonstration video is output.

8. The adaptive display method for dynamic exhibition content based on AI vision generation according to claim 1, characterized in that, Based on the anomalous distribution areas of the baseline, the process of determining rendering stability tendencies to decide whether to correct video in crosstalk-covered areas includes: Based on the three-dimensional height distribution information of the video projection area, the abnormal distribution area of ​​the base is determined in order to determine the crosstalk coverage area. Based on the number of abnormal distribution regions of the basal layer within the crosstalk coverage area, the rendering stability tendency is determined, and it is determined whether to perform deconvolution correction.

9. The adaptive display method for dynamic exhibition content based on AI vision generation according to claim 1, characterized in that, The process of determining the thermal flow disturbance risk value after completing the identification of crosstalk coverage areas includes: Obtain the regional temperature of each crosstalk coverage area and the regional temperature of the remaining sub-analysis areas, solve the variance of the temperature of each area, and obtain the heat flow disturbance risk value.

10. The adaptive display method for dynamic exhibition content based on AI vision generation according to claim 9, characterized in that, After identifying areas affected by crosstalk, the process of determining whether to correct images in texture-dense areas based on thermal flow disturbance risk values ​​includes: When the thermal disturbance risk value is greater than or equal to the preset disturbance threshold, the low-pass filter cutoff frequency of the video corresponding to the crosstalk coverage area is corrected.

Images