A method and system for inter-frame brightness flicker control of high-definition video

Abstract

The application discloses a high-definition video interframe brightness flicker control method and system, and relates to the technical field of computers.The method comprises the following steps: decoding an AIGC high-definition video to be processed frame by frame to obtain a frame image sequence; extracting a brightness distribution feature vector of the frame image sequence, calculating a difference vector of the brightness distribution feature vector to determine an interframe brightness flicker error; performing brightness gain compensation on the frame image sequence based on the interframe brightness flicker error by using a nonlinear partition compensation strategy; determining target optical flow information of the frame image sequence to determine a local area brightness change curve, performing local brightness smoothing on the frame image sequence based on the local area brightness change curve; and performing edge area brightness cross verification on the frame image sequence to perform video flicker suppression.The application realizes local adaptive fine control, ensures that the brightness transition on both sides of the edge is reasonable, and strengthens the reliability of flicker suppression.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method and system for controlling inter-frame brightness flicker in high-definition video. Background Technology

[0002] During the generation of AIGC high-definition videos, the frame-by-frame inference mechanism lacks constraints on the consistency of inter-frame brightness, leading to irregular fluctuations in the brightness distribution of consecutive frames. This results in visually perceptible inter-frame brightness flicker, which severely disrupts visual smoothness and viewing immersion, reduces the overall video quality, and even affects subsequent video analysis and editing applications. Therefore, efficient and precise inter-frame brightness flicker control for AIGC high-definition videos is crucial for improving their image quality and usability. However, existing video flicker suppression methods often employ global brightness equalization or simple temporal low-pass filtering, which are ill-suited to the complex local brightness variations and high precision requirements of moving and edge regions in AIGC videos. These methods tend to introduce problems such as loss of motion details and edge blurring while suppressing flicker, or result in incomplete flicker suppression in areas of drastic brightness changes. They fail to achieve fine-grained, locally adaptive control and thus cannot adequately meet the inter-frame brightness flicker control requirements of AIGC high-definition videos. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a method and system for inter-frame brightness flicker control of high-definition video, which realizes fine control with local adaptive control, ensures reasonable brightness transition on both sides of the edge, and further enhances the reliability of flicker suppression.

[0004] To address the aforementioned technical problems, this invention provides a method for controlling inter-frame brightness flicker in high-definition video, the method comprising: The AIGC high-definition video content to be processed is decoded frame by frame to obtain a sequence of frame images. Extract the brightness distribution feature vector of the frame image sequence, calculate the difference vector of the brightness distribution feature vector, and determine the inter-frame brightness flicker error based on the difference vector; Based on the inter-frame brightness flicker error, a nonlinear partitioning compensation strategy is used to perform brightness gain compensation on the frame image sequence to obtain a brightness gain compensated frame image sequence. Determine the target optical flow information of the frame image sequence after brightness gain compensation, and determine the brightness change curve of the local region based on the target optical flow information. Perform local brightness smoothing on the frame image sequence after brightness gain compensation based on the local brightness change curve to obtain the frame image sequence after local brightness smoothing. Edge region brightness cross-validation is performed on the frame image sequence after local brightness smoothing to obtain the edge region brightness cross-validation result. Based on the edge region brightness cross-validation result, video flicker suppression is performed to obtain flicker-suppressed AIGC high-definition video.

[0005] Optionally, the step of decoding the AIGC high-definition video to be processed frame by frame to obtain a sequence of frame images includes: Metadata parsing is performed on the AIGC high-definition video to be processed to obtain the metadata dataset; The target decoder is determined based on the metadata, and the AIGC high-definition video to be processed is decoded frame by frame based on the target decoder to obtain a frame image sequence.

[0006] Optionally, the step of extracting the brightness distribution feature vector of the frame image sequence, calculating the difference vector of the brightness distribution feature vector, and determining the inter-frame brightness flicker error based on the difference vector includes: The frame image sequence is divided into macroblock grids to obtain several macroblock grid images. The arithmetic mean of the brightness of each macroblock grid image is analyzed, and the brightness distribution feature vector is determined based on the arithmetic mean of the brightness. Calculate the difference vector of the brightness distribution feature vector, and determine the global fluctuation intensity of the frame brightness fluctuation severity based on the difference vector; The inter-frame brightness flicker error is determined based on the global fluctuation intensity.

[0007] Optionally, determining the inter-frame brightness flicker error based on the global fluctuation intensity includes: Based on the global fluctuation intensity, a basic flicker metric analysis is performed to obtain basic flicker metric information; Based on the difference vector, an overall brightness change tendency analysis is performed to obtain the overall brightness change tendency factor; The inter-frame brightness flicker error is determined based on the aforementioned basic flicker measurement information and the overall brightness change tendency factor.

[0008] Optionally, the step of performing brightness gain compensation on the frame image sequence based on the inter-frame brightness flicker error using a nonlinear partitioning compensation strategy to obtain a brightness gain-compensated frame image sequence includes: Based on the inter-frame brightness flicker error, compensation decision information is determined using a preset perception threshold. Based on the compensation decision information, a partition marking map is determined, and based on the partition marking map, a nonlinear gain curve is determined; The pixel-level brightness gain map is determined based on the nonlinear gain curve. The frame image sequence is subjected to brightness gain compensation based on the pixel-level brightness gain map to obtain a brightness gain compensated frame image sequence.

[0009] Optionally, determining the pixel-level brightness gain map based on the nonlinear gain curve includes: Pixel-level region attribution probability analysis is performed based on the partitioned labeling map to obtain pixel-level region attribution probability information. Target weight information is determined based on the pixel-level region attribution probability information; Based on the nonlinear gain curve, pixel-level brightness gain analysis is performed using the target weight information to obtain a pixel-level brightness gain map.

[0010] Optionally, the step of performing brightness gain compensation on the frame image sequence based on the pixel-level brightness gain map to obtain a brightness gain-compensated frame image sequence includes: Based on the pixel-level brightness gain map, a constraint gain map is determined using a bright area pixel constraint table; Perform joint bilateral filtering on the constraint gain map to obtain a filtered gain map; The frame image sequence is subjected to luminance gain compensation based on the filter gain map to obtain the luminance gain compensated frame image sequence.

[0011] Optionally, determining the target optical flow information of the brightness gain-compensated frame image sequence, determining the local region brightness variation curve based on the target optical flow information, and performing local brightness smoothing on the brightness gain-compensated frame image sequence based on the local region brightness variation curve to obtain a locally brightness-smoothed frame image sequence includes: Bidirectional optical flow analysis is performed on the frame image sequence after brightness gain compensation to obtain bidirectional optical flow information, and the target optical flow information of the frame image sequence after brightness gain compensation is determined based on the bidirectional optical flow information. Projection tracking is performed based on the target optical flow information to obtain the brightness variation curve of the local area; A time-domain filter is set, and the brightness change curve of the local area is smoothed based on the time-domain filter to obtain a smooth brightness curve; Based on the smoothed brightness curve, local brightness smoothing is performed on the frame image sequence after brightness gain compensation to obtain the locally brightness smoothed frame image sequence.

[0012] Optionally, the step of performing edge region brightness cross-validation on the locally brightness-smoothed frame image sequence to obtain edge region brightness cross-validation results, and performing video flicker suppression based on the edge region brightness cross-validation results to obtain flicker-suppressed AIGC high-definition video, includes: A single-pixel-width strong edge map is determined based on the frame image sequence after local brightness smoothing, and the edge region mask is analyzed based on the single-pixel-width strong edge map; Based on the edge region mask, perform edge region brightness cross-validation on the frame image sequence after local brightness smoothing to obtain the edge region brightness cross-validation result. Based on the edge region brightness cross-validation results, flicker suppression is performed on each brightness frame to obtain each brightness frame after flicker suppression; The flicker-suppressed AIGC high-definition video is obtained by merging the luminance frames after flicker suppression.

[0013] In addition, the present invention also provides an inter-frame brightness flicker control system for high-definition video, the system comprising: Video decoding module: used to decode AIGC high-definition video content to be processed frame by frame to obtain a sequence of frame images; Flickering error determination module: used to extract the brightness distribution feature vector of the frame image sequence, calculate the difference vector of the brightness distribution feature vector, and determine the inter-frame brightness flickering error based on the difference vector; Brightness gain compensation module: used to perform brightness gain compensation on the frame image sequence based on the inter-frame brightness flicker error using a nonlinear partitioning compensation strategy, to obtain a brightness gain compensated frame image sequence; Local brightness smoothing module: used to determine the target optical flow information of the frame image sequence after brightness gain compensation, and to determine the brightness change curve of the local region based on the target optical flow information, and to perform local brightness smoothing on the frame image sequence after brightness gain compensation based on the local region brightness change curve to obtain the locally brightness smoothed frame image sequence. Video flicker suppression module: used to perform edge region brightness cross-validation on the frame image sequence after local brightness smoothing, obtain the edge region brightness cross-validation result, and perform video flicker suppression based on the edge region brightness cross-validation result to obtain flicker-suppressed AIGC high-definition video.

[0014] In this embodiment of the invention, the AIGC (Artificial Intelligence Generated Content) high-definition video to be processed is decoded frame by frame to obtain a frame image sequence. The brightness distribution feature vector of the frame image sequence is extracted, and the difference vector is calculated. Based on the difference vector, the inter-frame brightness flicker error is determined, providing a clear and reliable correction target for the compensation stage. Based on the inter-frame brightness flicker error, a nonlinear partitioning compensation strategy is used to perform brightness gain compensation on the frame image sequence, reducing the overall flicker amplitude while maintaining a natural transition in image brightness levels. The target optical flow information of the brightness gain-compensated frame image sequence is determined, and the brightness change curve of the local region is determined based on the target optical flow information. Local brightness smoothing is then performed on the brightness gain-compensated frame image sequence based on the local brightness change curve. This eliminates local residual flicker while maintaining the edges and texture details of moving objects, achieving fine-grained local adaptive control and temporal brightness consistency in moving areas. Edge region brightness cross-validation is performed on the locally brightness-smoothed frame image sequence. Based on the edge region brightness cross-validation results, video flicker suppression is performed to ensure a reasonable brightness transition on both sides of the edges, further enhancing the reliability of flicker suppression. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the inter-frame brightness flicker control method for high-definition video in an embodiment of the present invention. Figure 2 This is a flowchart illustrating a high-definition video inter-frame brightness flicker control method according to another embodiment of the present invention. Figure 3 This is a schematic diagram of the structural composition of the high-definition video inter-frame brightness flicker control system in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1 Please see Figure 1 , Figure 1This is a flowchart illustrating the inter-frame brightness flicker control method for high-definition video according to an embodiment of the present invention. The method includes: S11: Decode the AIGC high-definition video content to be processed frame by frame to obtain a sequence of frame images; In the specific implementation of this invention, metadata parsing is performed on the high-definition video of the Artificial Intelligence Generated Content (AIGC) to be processed to obtain a metadata dataset; a target decoder is determined based on the metadata dataset, and the AIGC high-definition video to be processed is decoded frame by frame based on the target decoder to obtain a frame image sequence, providing basic data for subsequent frame-level analysis and processing, ensuring that the brightness information of each frame can be accurately manipulated, and avoiding the impact of encoding and decoding interference on flicker detection.

[0019] S12: Extract the brightness distribution feature vector of the frame image sequence, calculate the difference vector of the brightness distribution feature vector, and determine the inter-frame brightness flicker error based on the difference vector; In the specific implementation of this invention, each frame of the frame image sequence is divided into macroblock grids to obtain several macroblock grid images. The arithmetic mean of the brightness of each macroblock grid image is analyzed, and a brightness distribution feature vector is determined based on the arithmetic mean of the brightness. The difference vector of the brightness distribution feature vector is calculated, and the global fluctuation intensity of the frame brightness fluctuation is determined based on the difference vector. The inter-frame brightness flicker error is determined based on the global fluctuation intensity, transforming the visual flicker phenomenon into a measurable inter-frame brightness flicker error, providing a clear and reliable correction target for the compensation process.

[0020] S13: Based on the inter-frame brightness flicker error, a nonlinear partitioning compensation strategy is used to perform brightness gain compensation on the frame image sequence to obtain a brightness gain compensated frame image sequence. In the specific implementation of this invention, compensation decision information is determined based on the inter-frame brightness flicker error using a preset perception threshold; a partition marker map is determined based on the compensation decision information, and a nonlinear gain curve is determined based on the partition marker map; a pixel-level brightness gain map is determined based on the nonlinear gain curve; and brightness gain compensation is performed on the frame image sequence based on the pixel-level brightness gain map to obtain a brightness gain compensated frame image sequence, effectively avoiding overcompensation or undercompensation caused by global uniform compensation, and maintaining a natural transition of image brightness levels while reducing the overall flicker amplitude.

[0021] S14: Determine the target optical flow information of the frame image sequence after brightness gain compensation, and determine the local region brightness change curve based on the target optical flow information. Perform local brightness smoothing on the frame image sequence after brightness gain compensation based on the local region brightness change curve to obtain the locally brightness smoothed frame image sequence. In the specific implementation of this invention, bidirectional optical flow analysis is performed on the frame image sequence after brightness gain compensation to obtain bidirectional optical flow information, and the target optical flow information of the frame image sequence after brightness gain compensation is determined based on the bidirectional optical flow information; projection tracking is performed based on the target optical flow information to obtain the brightness change curve of the local area; a temporal filter is set, and the brightness change curve of the local area is smoothed based on the temporal filter to obtain a smooth brightness curve; local brightness smoothing is performed on the frame image sequence after brightness gain compensation based on the smooth brightness curve to obtain a frame image sequence with local brightness smoothing. This can eliminate local residual flicker while keeping the edges and texture details of moving objects from being blurred, thus achieving temporal brightness consistency in the moving area.

[0022] S15: Perform edge region brightness cross-validation on the frame image sequence after local brightness smoothing to obtain the edge region brightness cross-validation result, and perform video flicker suppression based on the edge region brightness cross-validation result to obtain flicker-suppressed AIGC high-definition video.

[0023] In the specific implementation of this invention, a single-pixel width strong edge map is determined based on the frame image sequence after local brightness smoothing, and an edge region mask is analyzed based on the single-pixel width strong edge map; edge region brightness cross-validation is performed on the frame image sequence after local brightness smoothing based on the edge region mask to obtain the edge region brightness cross-validation result; flicker suppression is performed on each brightness frame based on the edge region brightness cross-validation result to obtain each brightness frame after flicker suppression; the brightness frames after flicker suppression are merged to obtain the flicker-suppressed AIGC high-definition video, ensuring that the brightness transition on both sides of the edge is reasonable, further enhancing the reliability of flicker suppression, and finally obtaining a clear AIGC high-definition video with no visible flicker.

[0024] In this embodiment of the invention, the AIGC (Artificial Intelligence Generated Content) high-definition video to be processed is decoded frame by frame to obtain a frame image sequence. The brightness distribution feature vector of the frame image sequence is extracted, and the difference vector is calculated. Based on the difference vector, the inter-frame brightness flicker error is determined, providing a clear and reliable correction target for the compensation stage. Based on the inter-frame brightness flicker error, a nonlinear partitioning compensation strategy is used to perform brightness gain compensation on the frame image sequence, reducing the overall flicker amplitude while maintaining a natural transition in image brightness levels. The target optical flow information of the brightness gain-compensated frame image sequence is determined, and the brightness change curve of the local region is determined based on the target optical flow information. Local brightness smoothing is then performed on the brightness gain-compensated frame image sequence based on the local brightness change curve. This eliminates local residual flicker while maintaining the edges and texture details of moving objects, achieving fine-grained local adaptive control and temporal brightness consistency in moving areas. Edge region brightness cross-validation is performed on the locally brightness-smoothed frame image sequence. Based on the edge region brightness cross-validation results, video flicker suppression is performed to ensure a reasonable brightness transition on both sides of the edges, further enhancing the reliability of flicker suppression.

[0025] Example 2 Please see Figure 2 , Figure 2 This is a flowchart illustrating a method for controlling inter-frame brightness flicker in high-definition video according to another embodiment of the present invention, the method comprising: S201: Decode the AIGC high-definition video content to be processed frame by frame to obtain a sequence of frame images; In the specific implementation of this invention, the step of decoding the AIGC high-definition video to be processed frame by frame to obtain a frame image sequence includes: parsing the metadata of the AIGC high-definition video to be processed to obtain a metadata dataset; determining a target decoder based on the metadata dataset; and decoding the AIGC high-definition video to be processed frame by frame based on the target decoder to obtain a frame image sequence.

[0026] Specifically, the AIGC high-definition video to be processed is subjected to metadata parsing. The AIGC high-definition video is input into the demultiplexer, which parses the AIGC high-definition video and extracts the encoding format identifier of the video encoding track, the precise fractional representation of the frame rate, the image width and height, and the color format information. A rendering timestamp mapping table is established for each frame to obtain a metadata dataset that fully describes the timing and encoding attributes of the video frames.

[0027] Based on the metadata dataset, a target decoder is determined. A corresponding decoder instance is initialized according to the encoding format identifier in the metadata set. The output pixel format is forced into a color space where the luma and chroma planes are completely separated in the initialization parameters, resulting in a decoder capable of outputting independent luma and chroma components for each frame. The AIGC high-definition video to be processed is decoded frame by frame using the target decoder. The demultiplexed compressed frame data is sequentially fed into the target decoder. When the target decoder outputs each reconstructed image frame, its luma plane data is immediately copied to an independent buffer, while the chroma plane data is transferred unchanged to an independent frame sequence buffer, with only an index association established. All luma images are sorted according to the presentation timestamps in the metadata set to obtain a frame image sequence.

[0028] S202: Extract the brightness distribution feature vector of the frame image sequence, calculate the difference vector of the brightness distribution feature vector, and determine the inter-frame brightness flicker error based on the difference vector; In a specific implementation of this invention, the step of extracting the brightness distribution feature vector of the frame image sequence, calculating the difference vector of the brightness distribution feature vector, and determining the inter-frame brightness flicker error based on the difference vector includes: dividing each frame image of the frame image sequence into macroblock grids to obtain several macroblock grid images; analyzing the arithmetic mean of the brightness of each macroblock grid image; determining the brightness distribution feature vector based on the arithmetic mean of the brightness; calculating the difference vector of the brightness distribution feature vector; determining the global fluctuation intensity of the frame brightness fluctuation severity based on the difference vector; and determining the inter-frame brightness flicker error based on the global fluctuation intensity.

[0029] Specifically, the frame image sequence is divided into macroblock grids. The macroblock size is adaptively determined based on the video resolution in the metadata set. Each frame's brightness image in the frame image sequence is divided into a macroblock grid, resulting in several macroblock grid images. The arithmetic mean of brightness for each macroblock grid image is analyzed, i.e., the arithmetic mean of brightness within each macroblock is calculated after traversing each macroblock and removing abnormal pixels. Based on the arithmetic mean of brightness, a brightness distribution feature vector is determined. All brightness arithmetic means are arranged into a one-dimensional vector according to the spatial scanning order, and the brightness distribution feature vector for each frame is extracted.

[0030] The difference vector of the brightness distribution feature vector is calculated. Starting from the second frame of the time series, the brightness distribution feature vector of the current frame is a signed algebraic subtraction with the brightness distribution feature vector of the previous frame under the same macroblock index to obtain the inter-frame difference vector. Based on the difference vector, the global fluctuation intensity of the frame brightness fluctuation is determined, and a Gaussian decay spatial weight matrix with the image center in the metadata set as the origin is generated. The absolute value of each component of the difference vector is multiplied by the corresponding spatial weight to obtain the perceptually weighted brightness change amplitude distribution. The root mean square value of all amplitudes in the perceptually weighted brightness change amplitude distribution is calculated, and this root mean square value is used as the global fluctuation intensity characterizing the brightness fluctuation of the entire frame.

[0031] The inter-frame brightness flicker error is determined based on the global fluctuation intensity. Basic flicker measurement analysis can be performed through global fluctuation intensity, and overall brightness change trend analysis can be performed through difference vector. The two are combined to generate the inter-frame brightness flicker error.

[0032] For example, consider a 1080P high-definition video generated by AIGC, in which three consecutive frames exhibit slight brightness jitter in the same scene. First, these three brightness frames are divided into macroblock grids, with each frame uniformly divided into 64 macroblocks (8x8), each macroblock measuring 240x135 pixels. The arithmetic mean of the brightness of the 64 macroblocks in each frame is calculated. For instance, the macroblock numbered in row 1, column 1 of the first frame has an average brightness of 128.5, the second frame has 125.3, and the third frame has 130.1. The average values ​​of all macroblocks are then arranged in a spatial scanning order from left to right and top to bottom, forming the brightness distribution feature vector for each frame. Next, the luminance distribution feature vector of the second frame is subtracted one by one from the 64 components at the corresponding positions in the first frame to obtain the first set of difference vectors. The component values ​​are such as -3.2, +1.8, and -0.7, reflecting the luminance changes of each macroblock region between the two frames. Similarly, the second set of difference vectors between the second and third frames is calculated. Then, the root mean square value of the absolute value of each component in the first set of difference vectors is taken. Assuming the calculated global fluctuation intensity is 2.86, this value characterizes the severity of the luminance fluctuation in the entire frame. Finally, by comprehensively observing the changing trend of global fluctuation intensity between consecutive frames and the spatial proportion of polarity-flipped macroblocks, for example, it was found that the difference sign of 18 out of 64 macroblocks alternated between positive and negative in two consecutive sets of difference vectors, accounting for 0.28. After fusing the global fluctuation intensity of 2.86 with the polarity-flipped macroblock proportion of 0.28 and applying scene fading penalty and temporal smoothing, the inter-frame brightness flicker error value of the second frame relative to the first frame was finally determined to be 0.73. This error value exceeded the preset perception threshold of 0.5, which can trigger the subsequent nonlinear partition compensation process to correct the brightness of this frame.

[0033] Furthermore, determining the inter-frame brightness flicker error based on the global fluctuation intensity includes: performing a basic flicker measurement analysis based on the global fluctuation intensity to obtain basic flicker measurement information; performing an overall brightness change tendency analysis based on the difference vector to obtain an overall brightness change tendency factor; and determining the inter-frame brightness flicker error based on the basic flicker measurement information and the overall brightness change tendency factor.

[0034] Specifically, a basic flicker metric analysis is performed based on the global fluctuation intensity. A short-time state register is maintained at each macroblock location. The sign of the difference in each component of the difference vector is detected to see whether positive and negative signs alternate within a sliding window of three consecutive frames. The proportion of macroblocks with polarity reversal in the entire frame is counted to obtain the proportion of polarity-reversed macroblocks. The global fluctuation intensity is multiplied by the proportion of polarity-reversed macroblocks to obtain the basic flicker metric value reflecting the flicker intensity and spatial dispersion, i.e., the basic flicker metric information is obtained.

[0035] Based on the difference vector, an overall brightness change tendency analysis is performed. The quantitative relationship between the same-sign and opposite-sign components in the difference vector is statistically analyzed, and the overall brightness change tendency factor is calculated. If the same-sign component is absolutely dominant, the tendency factor value approaches a suppression coefficient less than one; otherwise, the value approaches one. Thus, the overall brightness change tendency factor can be obtained.

[0036] Based on the basic flicker measurement information and the overall brightness change tendency factor, the inter-frame brightness flicker error is determined. The basic flicker measurement value is multiplied by the overall brightness change tendency factor to obtain the original inter-frame flicker error that suppresses the interference of normal scene brightness gradient. An adaptive first-order temporal smoothing filter based on scene switching detection is applied to the original inter-frame flicker error. The filter state is reset when the scene switches, and a preset smoothing coefficient is used when there is no scene switching to determine the final inter-frame brightness flicker error.

[0037] S203: Based on the inter-frame brightness flicker error, determine compensation decision information using a preset perception threshold; In the specific implementation of this invention, compensation decision information is determined based on the inter-frame brightness flicker error using a preset perception threshold. The inter-frame brightness flicker error is compared with the preset perception threshold calibrated by subjective experiment. If the inter-frame brightness flicker error is less than the preset perception threshold, a decision to skip compensation is made and a placeholder gain map with all unit values ​​is prepared. If the inter-frame brightness flicker error is greater than or equal to the preset perception threshold, a decision to perform compensation is made, that is, the compensation decision information of the current frame is obtained.

[0038] S204: Determine the partition marking map based on the compensation decision information, and determine the nonlinear gain curve based on the partition marking map; In the specific implementation of this invention, a partitioning map is determined based on the compensation decision information. For frames where the compensation decision information is executed, each component of the corresponding difference vector is compared with preset positive high threshold, positive low threshold, negative high threshold, and negative low threshold. Based on the threshold range in which the component value is located, the corresponding macroblock is labeled as a severe drop region, a mild drop region, a stable region, a mild increase region, or a severe increase region, thus obtaining a macroblock-level five-category partitioning map. Based on the partitioning map, nonlinear gain curves are determined: for the severe drop region, an exponential convex function gain curve with a constrained upper gain limit is designed; for the mild drop region, a logarithmic convex function gain curve with a slower gain increase is designed; for the stable region, a straight line with constant unity gain is designed; for the mild increase region, a concave function suppression curve with slow start-up is designed; and for the severe increase region, a high-compression concave function suppression curve with high compression ratio and protected lower gain limit is designed, thus obtaining independent nonlinear gain curves corresponding to the five regions.

[0039] S205: Determine the pixel-level brightness gain map based on the nonlinear gain curve; In a specific implementation of the present invention, determining the pixel-level brightness gain map based on the nonlinear gain curve includes: performing pixel-level region attribution probability analysis based on the partitioning marker map to obtain pixel-level region attribution probability information; determining target weight information based on the pixel-level region attribution probability information; and performing pixel-level brightness gain analysis based on the nonlinear gain curve using the target weight information to obtain the pixel-level brightness gain map.

[0040] Specifically, pixel-level region attribution probability analysis is performed based on the partition marker map. The original brightness frame in the frame image sequence is used as the guide map. Joint guided filtering is performed on the partition marker map. By utilizing the object edge gradient information in the guide map, the transition of partition categories is forced to occur only at the natural edges of the image, resulting in a pixel-level region attribution probability map with smooth boundaries and conforming to the object contour. Each pixel has a corresponding probability value for five categories, thus obtaining pixel-level region attribution probability information.

[0041] Target weight information is determined based on the pixel-level region attribution probability information. For each pixel in the current frame, the attribution probability values ​​of the five categories in the pixel-level region attribution probability information are used as weights.

[0042] Pixel-level brightness gain analysis is performed based on the nonlinear gain curve and the target weight information. The calculation results of the five types of independent nonlinear gain curves at the original brightness value of the pixel are weighted and averaged using the target weight information to obtain a pixel-level brightness gain map.

[0043] S206: Perform brightness gain compensation on the frame image sequence based on the pixel-level brightness gain map to obtain a brightness gain compensated frame image sequence; In a specific implementation of the present invention, the step of performing brightness gain compensation on the frame image sequence based on the pixel-level brightness gain map to obtain a brightness gain compensated frame image sequence includes: determining a constraint gain map based on the pixel-level brightness gain map using a bright area pixel constraint table; performing joint bilateral filtering on the constraint gain map to obtain a filtered gain map; and performing brightness gain compensation on the frame image sequence based on the filtered gain map to obtain a brightness gain compensated frame image sequence.

[0044] Specifically, based on the pixel-level brightness gain map, a constrained gain map is determined using a bright area pixel constraint table. The pixel-level brightness gain map is then sent to a brightness adaptive constraint module. The constraint module has a built-in two-dimensional lookup table indexed by the original brightness value. This lookup table sets an upper limit for gain in dark area pixels to prevent noise amplification and a lower limit for gain in bright area pixels to prevent loss of highlight levels. The constrained gain map is obtained by querying the table based on the original brightness value of each pixel and clamping the gain value.

[0045] Joint bilateral filtering is applied to the constraint gain map. The original luminance frames in the frame image sequence are used as guide maps. The joint bilateral filtering is applied to the constraint gain map. The filtering radius and intensity are adaptively adjusted according to the magnitude of the inter-frame luminance flicker error. The larger the error, the stronger the filtering to eliminate macroblock compensation traces. The smaller the error, the weaker the filtering to protect texture details. Thus, the filtered gain map is obtained.

[0046] Based on the filter gain map, brightness gain compensation is performed on the frame image sequence. The filter gain map is multiplied pixel by pixel with the original brightness frame in the frame image sequence, and the product result is limited according to the brightness quantization range of the target display device to ensure that all brightness values ​​do not exceed the legal range, thus obtaining the frame image sequence after brightness gain compensation.

[0047] S207: Determine the target optical flow information of the frame image sequence after brightness gain compensation, and determine the local region brightness change curve based on the target optical flow information. Perform local brightness smoothing on the frame image sequence after brightness gain compensation based on the local region brightness change curve to obtain the locally brightness smoothed frame image sequence. In a specific implementation of this invention, determining the target optical flow information of the brightness gain-compensated frame image sequence, determining the local region brightness variation curve based on the target optical flow information, and performing local brightness smoothing on the brightness gain-compensated frame image sequence based on the local region brightness variation curve to obtain a locally brightness-smoothed frame image sequence includes: performing bidirectional optical flow analysis on the brightness gain-compensated frame image sequence to obtain bidirectional optical flow information, and determining the target optical flow information of the brightness gain-compensated frame image sequence based on the bidirectional optical flow information; performing projection tracking based on the target optical flow information to obtain the local region brightness variation curve; setting a temporal filter, smoothing the local region brightness variation curve based on the temporal filter to obtain a smoothed brightness curve; and performing local brightness smoothing on the brightness gain-compensated frame image sequence based on the smoothed brightness curve to obtain a locally brightness-smoothed frame image sequence.

[0048] Specifically, bidirectional optical flow analysis is performed on the frame image sequence after brightness gain compensation. The brightness images of two adjacent frames in the brightness gain compensated frame image sequence are input into a deep learning optical flow estimation network. The forward optical flow field from the current frame to the next frame and the backward optical flow field from the next frame to the current frame are calculated respectively to obtain bidirectional optical flow information. Based on the bidirectional optical flow information, the target optical flow information of the brightness gain compensated frame image sequence is determined. Bidirectional consistency verification is performed on each pixel of the current frame using the bidirectional optical flow information. Pixels are mapped to the next frame via forward optical flow and then back to the current frame via backward optical flow. The deviation between the reflected coordinates and the original coordinates is calculated. Optical flow vectors corresponding to pixels with deviations exceeding a sub-pixel threshold are discarded to obtain high-confidence optical flow information, i.e., the target optical flow information.

[0049] Projection tracking is performed based on the target optical flow information. Content adaptive superpixel segmentation is performed on the current frame to generate superpixel regions that fit the boundary of the object. The target optical flow information is used to project and track each superpixel to multiple adjacent frames in both forward and reverse directions. The average brightness value of the effective pixels in the successfully tracked region is collected and arranged in frame order to determine the brightness change curve of each local region, thus obtaining the brightness change curve of the local region.

[0050] A time-domain filter is configured, and signal decomposition techniques are used to decompose the local brightness variation curve into multiple modal components. The oscillating components whose energy is concentrated in the scintillation characteristic frequency band are identified. An adaptive time-domain filter with parameters determined by the scintillation energy ratio and the dominant frequency is constructed based on these oscillating components. The local brightness variation curve is then smoothed using this time-domain filter, resulting in a smoothed brightness curve.

[0051] Based on the smoothed brightness curve, local brightness smoothing is performed on the frame image sequence after brightness gain compensation. The difference between the smoothed brightness curve and the original brightness curve at the current frame time is calculated. This difference is used as the brightness correction amount required for the local area. The brightness correction amount is spread into a pixel-level continuous correction field through spatial distance weighting. Local brightness smoothing is performed on the frame image sequence after brightness gain compensation based on the pixel-level continuous correction field to obtain the locally brightness smoothed frame image sequence.

[0052] For example, consider an AIGC video after brightness gain compensation. The image shows a candle flickering slightly, and the background wall still exhibits periodic, subtle brightness fluctuations due to incomplete compensation. Taking three consecutive frames of brightness images after compensation, the second frame and the first frame are fed into a deep learning optical flow estimation network. The forward optical flow vector of the pixel at coordinates (320, 240) in the second frame is calculated to be +2.1 pixels horizontally and -0.8 pixels vertically. Similarly, the backward optical flow of this point is calculated with the third frame, yielding a backward optical flow vector of -1.9 pixels horizontally and +0.7 pixels vertically. This constructs the preliminary bidirectional optical flow information for this pixel location. Next, a bidirectional consistency check is performed on the pixel. The coordinates (320, 240) are mapped to the first frame via forward optical flow and then back to the second frame via backward optical flow. The reflected coordinates fall at (320.15, 239.92), and the Euclidean distance deviation from the original coordinates is only 0.18 pixels, which is lower than the preset sub-pixel threshold of 0.5 pixels. Therefore, the bidirectional optical flow vector of this point is preserved, and high-confidence target optical flow information of the pixel position is obtained. Then, using the superpixel region to which the pixel belongs as the tracking unit, the target optical flow information is used to trace back three frames and push forward three frames, collecting the average brightness value of the superpixel region in these seven consecutive frames. Arranged in frame order, it forms a brightness change curve with the numerical sequence 118.3, 119.1, 121.5, 122.8, 120.2, 119.6, 118.9. It can be seen that the curve has a small sawtooth oscillation superimposed during the process of slowly rising and falling. For the brightness variation curve of this local area, an adaptive time-domain filter with a cutoff frequency determined by a flicker energy ratio of 0.35 and a dominant frequency of 8 Hz was set to smooth the curve. The output smooth brightness curve values ​​after filtering are 118.5, 119.2, 120.8, 121.9, 120.4, 119.4, and 118.7, which significantly suppresses high-frequency oscillations while preserving the low-frequency gradual change trend. The difference between the smooth brightness curve value of 121.9 at the current fourth frame and the original brightness value of 122.8 was calculated, and the required brightness correction amount for this local area was found to be -0.9. This correction amount was then spatially distance-weighted and diffused to a pixel-level correction field and superimposed on the compensated brightness frame. In the fourth frame, the brightness value of this area was adjusted from 122.8 to 121.9, eliminating residual flicker fluctuations, and finally obtaining the frame image sequence with smoothed local brightness.

[0053] S208: Perform edge region brightness cross-validation on the frame image sequence after local brightness smoothing to obtain the edge region brightness cross-validation result, and perform video flicker suppression based on the edge region brightness cross-validation result to obtain flicker-suppressed AIGC high-definition video.

[0054] In a specific implementation of this invention, the step of performing edge region brightness cross-validation on the locally brightness-smoothed frame image sequence to obtain edge region brightness cross-validation results, and performing video flicker suppression based on the edge region brightness cross-validation results to obtain flicker-suppressed AIGC high-definition video, includes: determining a single-pixel width strong edge map based on the locally brightness-smoothed frame image sequence, and analyzing an edge region mask based on the single-pixel width strong edge map; performing edge region brightness cross-validation on the locally brightness-smoothed frame image sequence based on the edge region mask to obtain edge region brightness cross-validation results; performing flicker suppression on each brightness frame based on the edge region brightness cross-validation results to obtain flicker-suppressed brightness frames; and merging the flicker-suppressed brightness frames to obtain flicker-suppressed AIGC high-definition video.

[0055] Specifically, a single-pixel-width strong edge map is determined based on the frame image sequence after local brightness smoothing. Gradient calculation, non-maximum suppression, and dual threshold filtering are performed on each frame in the locally brightness smoothed frame image sequence to obtain the single-pixel-width strong edge map. An edge region mask is analyzed based on the single-pixel-width strong edge map, and a circular structuring element is used to perform morphological dilation on the single-pixel-width strong edge map to obtain the edge region mask.

[0056] Based on the edge region mask, the frame image sequence after local brightness smoothing is cross-validated for edge region brightness. For each pixel within the edge region mask, the local contrast is sampled along its edge normal direction and calculated. The target optical flow information is used to map the pixel to the corresponding position in the previous and next frames to extract the corresponding brightness value and contrast. Local block matching search is initiated for unreliable optical flow points to obtain matching brightness and contrast. The linear interpolation of the brightness of adjacent frames is used as the predicted brightness of the current frame. The deviation between the actual brightness and the predicted brightness of the current frame is calculated, and the variance of the local contrast of the pixel in the previous, middle and next frames is calculated. When the deviation exceeds the first adaptive threshold determined by the local noise level and the contrast variance exceeds the second adaptive threshold determined by the edge intensity itself, the pixel is determined as an edge flickering abnormal pixel, thereby obtaining the edge region brightness cross-validation result.

[0057] Based on the edge region brightness cross-validation results, flicker suppression is performed on each brightness frame. The coordinate position of each pixel judged as having edge flicker abnormality is located according to the edge region brightness cross-validation results. Starting from the coordinate position, the corresponding pixel is tracked along the high confidence optical flow trajectory for several frames forward and several frames backward. The brightness value of the corresponding pixel in each tracked frame is extracted to obtain the temporal matching pixel brightness set. Based on the edge region brightness cross-validation results, for each edge flickering abnormal pixel, starting from the pixel's coordinate position, a local search range is defined within the current frame along the edge direction in the edge region mask. The brightness values ​​of neighboring pixels located on the same edge that have not been identified as edge flickering abnormal pixels are collected to obtain a spatial non-flickering pixel brightness set. The temporal matching pixel brightness set and the spatial non-flickering pixel brightness set are merged to form a spatiotemporal candidate sample set for the edge flickering abnormal pixel. Each candidate sample in the spatiotemporal candidate sample set carries the brightness value of its source pixel, the spatial distance from the center pixel, the temporal distance from the current frame, and the edge direction information of the source pixel.

[0058] For each candidate sample in the spatiotemporal candidate sample set, its brightness value is compared with the brightness value of the center pixel to calculate a brightness similarity weight. The spatial distance between the candidate sample and the center pixel is used to calculate a spatial distance weight using a Gaussian kernel function. The temporal distance between the candidate sample and the current frame is also used to calculate a temporal distance weight using a Gaussian kernel function. The edge direction consistency weight is calculated by comparing the edge direction of the source pixel with the edge direction of the center pixel. The brightness similarity weight, spatial distance weight, temporal distance weight, and edge direction consistency weight are multiplied together to obtain the weighted weight of the candidate sample. The brightness values ​​of the corresponding candidate samples in the spatiotemporal candidate sample set are then weighted and averaged using the weighted weights of each candidate sample in the spatiotemporal candidate sample set to calculate the repaired brightness value. This repaired brightness value is then used to fill in the coordinates of the corresponding edge flickering abnormal pixels, resulting in a complete brightness frame after flicker suppression. This yields the individual brightness frames after flicker suppression.

[0059] The flicker-suppressed luminance frames are merged, and the flicker-suppressed luminance frames are merged with the chrominance frame sequences one by one according to the timestamp to restore the frame sequence in the standard preset color space. The frame sequence in the standard preset color space is input into the encoder with the same encoding parameters as the original video in the metadata set for compression. The original audio track and timestamp are reused and packaged for output to obtain the flicker-suppressed AIGC high-definition video.

[0060] In this embodiment of the invention, the AIGC (Artificial Intelligence Generated Content) high-definition video to be processed is decoded frame by frame to obtain a frame image sequence. The brightness distribution feature vector of the frame image sequence is extracted, and the difference vector is calculated. Based on the difference vector, the inter-frame brightness flicker error is determined, providing a clear and reliable correction target for the compensation stage. Based on the inter-frame brightness flicker error, a nonlinear partitioning compensation strategy is used to perform brightness gain compensation on the frame image sequence, reducing the overall flicker amplitude while maintaining a natural transition in image brightness levels. The target optical flow information of the brightness gain-compensated frame image sequence is determined, and the brightness change curve of the local region is determined based on the target optical flow information. Local brightness smoothing is then performed on the brightness gain-compensated frame image sequence based on the local brightness change curve. This eliminates local residual flicker while maintaining the edges and texture details of moving objects, achieving fine-grained local adaptive control and temporal brightness consistency in moving areas. Edge region brightness cross-validation is performed on the locally brightness-smoothed frame image sequence. Based on the edge region brightness cross-validation results, video flicker suppression is performed to ensure a reasonable brightness transition on both sides of the edges, further enhancing the reliability of flicker suppression.

[0061] Example 3 Please see Figure 3 , Figure 3 This is a schematic diagram of the structural composition of the high-definition video inter-frame brightness flicker control system according to an embodiment of the present invention. The system includes: Video decoding module 31: used to decode the AIGC high-definition video to be processed frame by frame to obtain a sequence of frame images; Flickering error determination module 32: used to extract the brightness distribution feature vector of the frame image sequence, calculate the difference vector of the brightness distribution feature vector, and determine the inter-frame brightness flickering error based on the difference vector; Brightness gain compensation module 33: used to perform brightness gain compensation on the frame image sequence based on the inter-frame brightness flicker error using a nonlinear partitioning compensation strategy, to obtain a brightness gain compensated frame image sequence; Local brightness smoothing module 34: used to determine the target optical flow information of the frame image sequence after brightness gain compensation, and to determine the brightness change curve of the local region based on the target optical flow information, and to perform local brightness smoothing on the frame image sequence after brightness gain compensation based on the local region brightness change curve to obtain the frame image sequence after local brightness smoothing. Video flicker suppression module 35: used to perform edge region brightness cross-validation on the frame image sequence after local brightness smoothing, obtain the edge region brightness cross-validation result, and perform video flicker suppression based on the edge region brightness cross-validation result to obtain flicker-suppressed AIGC high-definition video.

[0062] In the specific implementation of this invention, the specific implementation methods of the system items can be referred to the implementation methods of the above-mentioned method items, and will not be repeated here.

[0063] In this embodiment of the invention, the AIGC (Artificial Intelligence Generated Content) high-definition video to be processed is decoded frame by frame to obtain a frame image sequence. The brightness distribution feature vector of the frame image sequence is extracted, and the difference vector is calculated. Based on the difference vector, the inter-frame brightness flicker error is determined, providing a clear and reliable correction target for the compensation stage. Based on the inter-frame brightness flicker error, a nonlinear partitioning compensation strategy is used to perform brightness gain compensation on the frame image sequence, reducing the overall flicker amplitude while maintaining a natural transition in image brightness levels. The target optical flow information of the brightness gain-compensated frame image sequence is determined, and the brightness change curve of the local region is determined based on the target optical flow information. Local brightness smoothing is then performed on the brightness gain-compensated frame image sequence based on the local brightness change curve. This eliminates local residual flicker while maintaining the edges and texture details of moving objects, achieving fine-grained local adaptive control and temporal brightness consistency in moving areas. Edge region brightness cross-validation is performed on the locally brightness-smoothed frame image sequence. Based on the edge region brightness cross-validation results, video flicker suppression is performed to ensure a reasonable brightness transition on both sides of the edges, further enhancing the reliability of flicker suppression.

[0064] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0065] Furthermore, the above provides a detailed description of the inter-frame brightness flicker control method and system for high-definition video provided by the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for controlling inter-frame brightness flicker in high-definition video, characterized in that, The method includes: The AIGC high-definition video content to be processed is decoded frame by frame to obtain a sequence of frame images. Extract the brightness distribution feature vector of the frame image sequence, calculate the difference vector of the brightness distribution feature vector, and determine the inter-frame brightness flicker error based on the difference vector; Based on the inter-frame brightness flicker error, a nonlinear partitioning compensation strategy is used to perform brightness gain compensation on the frame image sequence to obtain a brightness gain compensated frame image sequence, including: determining a partitioning marker map based on the inter-frame brightness flicker error, determining a nonlinear gain curve based on the partitioning marker map, and performing brightness gain compensation on the frame image sequence based on the nonlinear gain curve. Determine the target optical flow information of the frame image sequence after brightness gain compensation, and determine the brightness change curve of the local region based on the target optical flow information. Perform local brightness smoothing on the frame image sequence after brightness gain compensation based on the local brightness change curve to obtain the frame image sequence after local brightness smoothing. The process involves performing edge region brightness cross-validation on a locally brightness-smoothed frame image sequence to obtain the edge region brightness cross-validation result, and then performing video flicker suppression based on the edge region brightness cross-validation result to obtain flicker-suppressed AIGC high-definition video. This includes: determining a single-pixel-width strong edge map based on the locally brightness-smoothed frame image sequence; performing edge region brightness cross-validation on the locally brightness-smoothed frame image sequence based on the single-pixel-width strong edge map; and performing flicker suppression and merging of each brightness frame based on the edge region brightness cross-validation result.

2. The method for controlling inter-frame brightness flicker in high-definition video according to claim 1, characterized in that, The AIGC high-definition video to be processed is decoded frame by frame to obtain a sequence of frame images, including: Metadata parsing is performed on the AIGC high-definition video to be processed to obtain the metadata dataset; The target decoder is determined based on the metadata, and the AIGC high-definition video to be processed is decoded frame by frame based on the target decoder to obtain a frame image sequence.

3. The method for controlling inter-frame brightness flicker in high-definition video according to claim 1, characterized in that, The step of extracting the brightness distribution feature vector of the frame image sequence, calculating the difference vector of the brightness distribution feature vector, and determining the inter-frame brightness flicker error based on the difference vector includes: The frame image sequence is divided into macroblock grids to obtain several macroblock grid images. The arithmetic mean of the brightness of each macroblock grid image is analyzed, and the brightness distribution feature vector is determined based on the arithmetic mean of the brightness. Calculate the difference vector of the brightness distribution feature vector, and determine the global fluctuation intensity of the frame brightness fluctuation severity based on the difference vector; The inter-frame brightness flicker error is determined based on the global fluctuation intensity.

4. The method for controlling inter-frame brightness flicker in high-definition video according to claim 3, characterized in that, The determination of inter-frame brightness flicker error based on the global fluctuation intensity includes: Based on the global fluctuation intensity, a basic flicker metric analysis is performed to obtain basic flicker metric information; Based on the difference vector, an overall brightness change tendency analysis is performed to obtain the overall brightness change tendency factor; The inter-frame brightness flicker error is determined based on the aforementioned basic flicker measurement information and the overall brightness change tendency factor.

5. The method for controlling inter-frame brightness flicker in high-definition video according to claim 1, characterized in that, The step of performing brightness gain compensation on the frame image sequence based on the inter-frame brightness flicker error using a nonlinear partitioning compensation strategy to obtain a brightness gain compensated frame image sequence includes: Based on the inter-frame brightness flicker error, compensation decision information is determined using a preset perception threshold. Based on the compensation decision information, a partition marking map is determined, and based on the partition marking map, a nonlinear gain curve is determined; The pixel-level brightness gain map is determined based on the nonlinear gain curve. The frame image sequence is subjected to brightness gain compensation based on the pixel-level brightness gain map to obtain a brightness gain compensated frame image sequence.

6. The method for controlling inter-frame brightness flicker in high-definition video according to claim 5, characterized in that, The process of determining the pixel-level brightness gain map based on the nonlinear gain curve includes: Pixel-level region attribution probability analysis is performed based on the partitioned labeling map to obtain pixel-level region attribution probability information. Target weight information is determined based on the pixel-level region attribution probability information; Based on the nonlinear gain curve, pixel-level brightness gain analysis is performed using the target weight information to obtain a pixel-level brightness gain map.

7. The method for controlling inter-frame brightness flicker in high-definition video according to claim 5, characterized in that, The step of performing brightness gain compensation on the frame image sequence based on the pixel-level brightness gain map to obtain a brightness gain compensated frame image sequence includes: Based on the pixel-level brightness gain map, a constraint gain map is determined using a bright area pixel constraint table; Perform joint bilateral filtering on the constraint gain map to obtain a filtered gain map; The frame image sequence is subjected to luminance gain compensation based on the filter gain map to obtain the luminance gain compensated frame image sequence.

8. The method for controlling inter-frame brightness flicker in high-definition video according to claim 1, characterized in that, The process of determining the target optical flow information of the frame image sequence after brightness gain compensation, determining the brightness variation curve of a local region based on the target optical flow information, and performing local brightness smoothing on the frame image sequence after brightness gain compensation based on the local brightness variation curve to obtain a locally brightness smoothed frame image sequence includes: Bidirectional optical flow analysis is performed on the frame image sequence after brightness gain compensation to obtain bidirectional optical flow information, and the target optical flow information of the frame image sequence after brightness gain compensation is determined based on the bidirectional optical flow information. Projection tracking is performed based on the target optical flow information to obtain the brightness variation curve of the local area; A time-domain filter is set, and the brightness change curve of the local area is smoothed based on the time-domain filter to obtain a smooth brightness curve; Based on the smoothed brightness curve, local brightness smoothing is performed on the frame image sequence after brightness gain compensation to obtain the locally brightness smoothed frame image sequence.

9. The method for controlling inter-frame brightness flicker in high-definition video according to claim 1, characterized in that, The process involves performing edge region brightness cross-validation on the locally smoothed frame image sequence to obtain edge region brightness cross-validation results, and then performing video flicker suppression based on these results to obtain flicker-suppressed AIGC high-definition video, including: A single-pixel-width strong edge map is determined based on the frame image sequence after local brightness smoothing, and the edge region mask is analyzed based on the single-pixel-width strong edge map; Based on the edge region mask, perform edge region brightness cross-validation on the frame image sequence after local brightness smoothing to obtain the edge region brightness cross-validation result. Based on the edge region brightness cross-validation results, flicker suppression is performed on each brightness frame to obtain each brightness frame after flicker suppression; The flicker-suppressed AIGC high-definition video is obtained by merging the luminance frames after flicker suppression.

10. A high-definition video inter-frame brightness flicker control system, characterized in that, The system includes: Video decoding module: used to decode AIGC high-definition video content to be processed frame by frame to obtain a sequence of frame images; Flickering error determination module: used to extract the brightness distribution feature vector of the frame image sequence, calculate the difference vector of the brightness distribution feature vector, and determine the inter-frame brightness flickering error based on the difference vector; Brightness gain compensation module: used to perform brightness gain compensation on the frame image sequence based on the inter-frame brightness flicker error using a nonlinear partitioning compensation strategy, to obtain a brightness gain compensated frame image sequence, including: determining a partitioning marker map based on the inter-frame brightness flicker error, determining a nonlinear gain curve based on the partitioning marker map; and performing brightness gain compensation on the frame image sequence based on the nonlinear gain curve; Local brightness smoothing module: used to determine the target optical flow information of the frame image sequence after brightness gain compensation, and to determine the brightness change curve of the local region based on the target optical flow information, and to perform local brightness smoothing on the frame image sequence after brightness gain compensation based on the local region brightness change curve to obtain the locally brightness smoothed frame image sequence. The video flicker suppression module is used to perform edge region brightness cross-validation on the locally brightness-smoothed frame image sequence to obtain the edge region brightness cross-validation result, and to perform video flicker suppression based on the edge region brightness cross-validation result to obtain flicker-smoothed AIGC high-definition video. This includes: determining a single-pixel-width strong edge map based on the locally brightness-smoothed frame image sequence; performing edge region brightness cross-validation on the locally brightness-smoothed frame image sequence based on the single-pixel-width strong edge map; and performing flicker suppression and merging of each brightness frame based on the edge region brightness cross-validation result.

Images