Method for adaptive regional brightness adjustment in stadium broadcast frames

Abstract

A computer-implemented method for processing image data to dynamically adjust brightness levels in stadium broadcast frames comprises receiving 100 image data from a video feed, analysing brightness l

Description

[0002] In recent years, advancements in broadcasting technology have greatly enhanced the quality of live sports broadcasts, allowing viewers to experience high-definition visuals and immersive sound. However, despite significant improvements in image clarity, color accuracy, and resolution, challenges persist when it comes to broadcasting live events in outdoor environments, such as stadiums, where lighting conditions can vary dramatically within a single frame. This is particularly problematic in stadiums where structural design and natural lighting create distinct zones of sunlight and shade across the playing field. As a result, some areas of the broadcast image may appear overly bright, washed out, or glaring, while others may seem too dark or lacking in detail. These inconsistencies in lighting conditions can disrupt the viewing experience, obscuring the visibility of key actions on the field and diminishing the overall quality of the broadcast.

[0003] Traditional approaches to addressing brightness and contrast discrepancies in broadcast images have generally relied on global adjustments that apply uniform brightness or contrast changes across the entire frame. Such methods are commonly implemented in video editing and broadcasting software, allowing operators to make real-time adjustments to overall brightness levels in response to changing light conditions. However, these techniques have significant limitations in scenarios where localized adjustments are needed. Uniform changes often fail to correct the problem of uneven lighting, as bright regions may still appear overexposed, and dark regions may remain underexposed. These global adjustments also risk diminishing image quality by affecting areas of the frame that were already adequately lit, thereby sacrificing visual balance and detail in the pursuit of generalized brightness control.

[0004] Other approaches have introduced dynamic brightness control based on scene recognition, often through ambient light sensors that detect overall light levels and adjust display brightness accordingly. These systems are typically employed in mobile devices and some high-end televisions to optimize screen brightness based on external lighting conditions. However, such ambient-light-based adjustments are not well-suited to complex environments like stadiums, where lighting variations occur within the same frame due to sunlight and shadow patterns created by the structure itself. These systems lack the granularity to distinguish between specific overexposed and underexposed areas within a single frame and thus cannot provide the localized adjustments required for broadcast images that contain varied lighting conditions across different regions.

[0005] Another limitation of current image processing methods in broadcasting is the lack of adaptability to evolving lighting conditions. In outdoor stadiums, lighting can change rapidly due to factors such as moving cloud cover, the position of the sun, or changing weather conditions. Existing solutions typically require manual adjustments or are slow to adapt to such changes, which can lead to inconsistent image quality throughout the broadcast. While some advanced systems allow for automatic adjustments based on pre-defined rules or thresholds, these systems are generally unable to respond dynamically to the nuanced lighting variations that occur within different areas of a stadium frame. Consequently, viewers may experience abrupt shifts in brightness and contrast that detract from the viewing experience, especially when lighting conditions vary frequently.

[0006] The combination of these factors underscores a pressing need for an innovative solution capable of addressing brightness variations in real-time, on a localized basis, and without disrupting the visual balance of the broadcast image. Ideally, such a system would be able to intelligently identify areas of overexposure and underexposure within each frame, apply targeted brightness adjustments, and do so continuously and seamlessly as lighting conditions change. This would allow broadcasters to deliver a consistent, high-quality viewing experience even in challenging lighting environments, enhancing both visibility and viewer satisfaction. The limitations of existing methods highlight the complexity of achieving localized brightness control in live broadcasts and demonstrate the potential value of a solution that could effectively handle these challenges in real-time. Summary

[0007] The present invention provides a computer-implemented method for processing image data to address brightness discrepancies across regions of a stadium broadcast frame. The method involves receiving image data from a video feed, analyzing brightness levels within this data using a neural network model to detect brightness variations, and segmenting the broadcast frame into regions based on these detected variations. Each region is classified as either overexposed or underexposed, depending on its luminance level, and brightness adjustments are applied specifically to these regions to achieve a balanced image. The adjusted image data is then output, providing modified luminance across regions of the stadium broadcast frame.

[0008] By implementing region-specific brightness adjustments, this method effectively balances bright and shaded areas in real time, improving the clarity of the broadcast without applying uniform brightness changes across the entire frame. The invention's design allows for continuous adaptation to changing lighting conditions, ensuring optimized broadcast quality and improved viewing experience.

[0009] In some embodiments, the method includes a preprocessing step for the received image data, normalizing the frame resolution and adjusting color data to optimize compatibility with the neural network model. This preprocessing enhances the accuracy of brightness analysis and ensures consistent results across different broadcast resolutions.

[0010] In further embodiments, the neural network model is a convolutional neural network trained on datasets featuring diverse stadium lighting conditions. This configuration improves the model's ability to distinguish overexposed and underexposed regions accurately, even in complex lighting environments.

[0011] In additional embodiments, the broadcast frame is segmented into regions of varying grid density, with higher density applied in areas showing complex lighting variations and lower density in uniformly lit areas. This dynamic segmentation enhances processing efficiency by focusing resources on areas requiring finer adjustments.

[0012] In yet further embodiments, classification of each region as overexposed or underexposed is based on a dynamically adjustable threshold brightness range. Adjusting threshold values according to ambient lighting or pre-determined stadium metrics allows for improved responsiveness to the specific lighting characteristics of each broadcast frame.

[0013] In some embodiments, brightness modifications for overexposed regions involve applying gamma correction with a value less than one. This gamma adjustment reduces luminance in bright areas while preserving detail, achieving a more natural balance across the frame.

[0014] In other embodiments, brightness modifications for underexposed regions use gamma correction with a value greater than one, allowing luminance enhancement without introducing excessive noise. This approach improves visibility in shaded areas while maintaining image quality.

[0015] In further embodiments, brightness redistribution is performed by calculating surplus brightness from overexposed regions and reallocating it to underexposed regions. This redistribution provides balanced illumination across the frame without uniformly increasing brightness, which can help preserve contrast in adequately lit areas.

[0016] In yet additional embodiments, overexposed regions are adjusted by selectively decreasing pixel intensity values to reach a target luminance level, allowing for precise control over brightness reduction.

[0017] In other embodiments, underexposed regions are adjusted by boosting pixel intensity values selectively to meet a target luminance level, ensuring improved visibility while preserving contrast.

[0018] In some embodiments, edge detection techniques are applied to identify boundaries within each region, enabling selective brightness adjustments that do not alter detected edges. This edge-preserving adjustment maintains clarity in key elements within the broadcast frame.

[0019] In further embodiments, edge detection is conducted using a Canny edge detection algorithm, which defines boundaries between sunny and shaded areas as well as distinguishing objects within the frame. This enhances the accuracy of region boundaries for brightness adjustments.

[0020] In yet further embodiments, the adjusted brightness levels are applied in real time to each broadcast frame, providing continuous brightness adaptation in response to changing stadium lighting conditions. This real-time processing supports uninterrupted broadcast quality.

[0021] In additional embodiments, brightness adjustments are dynamically skipped for frames where the neural network model does not detect a significant brightness discrepancy. Outputting unmodified image data in such cases conserves processing resources and optimizes system efficiency.

[0022] In some embodiments, the neural network model outputs a confidence score for each brightness classification, and adjustments are applied only when this confidence score exceeds a set threshold. This feature ensures reliable adjustment actions based on high-confidence classifications.

[0023] In other embodiments, an edge-preserving feature extraction technique is used to exclude important elements, such as players and game equipment, from brightness adjustments. This preserves clarity and detail for central elements within the frame.

[0024] In further embodiments, the neural network model is integrated within the camera hardware used for capturing the broadcast frame. This integration allows in-camera detection and brightness adjustment, potentially reducing downstream processing requirements.

[0025] In some embodiments, modified image data is encoded in a broadcast-compatible format, such as H.264 or H.265, for efficient transmission, ensuring compatibility with industry standards for broadcast delivery.

[0026] In additional embodiments, brightness adjustments for overexposed and underexposed regions are performed using tone mapping, improving contrast and detail in each region of the frame.

[0027] In further embodiments, the segmentation of the broadcast frame into regions is based on calculated lighting boundaries, with regions shaped to follow natural lighting divisions within the stadium. This boundary-based segmentation supports precise alignment of brightness adjustments with natural lighting contours within the frame. Brief Description of the Drawings

[0028] Various embodiments of the invention are disclosed in the following detailed description and accompanying drawings.

[0029] FIG. 1 illustrates an example flowchart of the method for dynamically adjusting brightness levels within a stadium broadcast frame.

[0030] FIG. 2 illustrates an example system architecture block diagram of the software components required to implement the brightness adjustment method.

[0031] FIG. 3A illustrates an example of an initial, unprocessed broadcast frame showing uneven lighting conditions across the stadium. FIG. 3B illustrates an example of the identified "too sunny" regions within the broadcast frame detected by the neural network. FIG. 3C illustrates an example of the thresholded view highlighting the "too shady" regions within the broadcast frame. FIG. 3D illustrates an example of the final output frame with balanced brightness adjustments applied to the "too sunny" and "too shady" regions.

[0032] Common reference numerals are used throughout the figures and the detailed description to indicate like elements. One skilled in the art will readily recognize that the above figures are examples and that other architectures, modes of operation, orders of operation, and elements / functions can be provided and implemented without departing from the characteristics and features of the invention, as set forth in the claims. Detailed Description and Preferred Embodiment

[0033] The following is a detailed description of exemplary embodiments to illustrate the principles of the invention. The embodiments are provided to illustrate aspects of the invention, but the invention is not limited to any embodiment. The scope of the invention encompasses numerous alternatives, modifications and equivalent; it is limited only by the claims.

[0034] Numerous specific details are set forth in the following description in order to provide a thorough understanding of the invention. However, the invention may be practiced according to the claims without some or all of these specific details. For the purpose of clarity, technical material that is known in the technical fields related to the invention has not been described in detail so that the invention is not unnecessarily obscured. DEFINITIONS:

[0035] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0036] As used herein, the term "and / or" includes any combinations of one or more of the associated listed items.

[0037] As used herein, the singular forms "a," "an," and "the" are intended to include the plural forms as well as the singular forms, unless the context clearly indicates otherwise.

[0038] It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.

[0039] The term "image data" refers to any digital representation of a visual scene captured from a stadium broadcast, comprising pixel values encoding luminance, color, or both. This may include, but is not limited to, frames from digital video feeds, image sequences, or still frames suitable for broadcast. In one example implementation, the image data is derived from a high-definition camera feed captured in real time, with each frame formatted according to industry standards such as H.264 or H.265, allowing efficient processing and transmission.

[0040] The term "neural network model" refers to any machine learning architecture trained to process visual input for the purpose of detecting brightness levels and categorizing regions based on luminance. This includes, but is not limited to, convolutional neural networks (CNNs), recurrent neural networks (RNNs), and hybrid neural networks designed for real-time image classification. In one example implementation, the neural network model is a CNN trained on a dataset of stadium images featuring varied lighting conditions, enabling it to detect overexposed and underexposed regions accurately.

[0041] The term "brightness discrepancy" refers to a variation in luminance levels between regions within a broadcast frame, where one or more regions exceed a predetermined brightness threshold (overexposed) and others fall below it (underexposed). This definition encompasses both subtle and stark differences in lighting caused by factors such as stadium architecture, time of day, and weather conditions. In one implementation, brightness discrepancy is determined by the neural network model based on pixel intensity values, with regions classified as overexposed if they exceed a set luminance threshold and underexposed if they fall below it.

[0042] The term "segmentation" refers to the division of the broadcast frame into discrete regions, each assessed independently for brightness adjustments. Segmentation can be achieved through various methods, including grid-based division, dynamic region shaping based on detected brightness boundaries, or adaptive partitioning based on lighting complexity. In one example, segmentation is performed by overlaying a fixed grid pattern onto the frame, dividing it into regions that can be processed independently, with finer segmentation applied to areas requiring more precise adjustments.

[0043] The term "brightness modification" refers to the adjustment of luminance levels within a region to bring its brightness within a desired range. This may include gamma correction, pixel intensity manipulation, tone mapping, or other image enhancement techniques designed to either reduce or increase luminance. In one example, overexposed regions are corrected by applying a gamma value less than one, reducing luminance without compromising detail, while underexposed regions are enhanced by applying a gamma value greater than one to improve visibility without introducing noise.

[0044] The term "outputting modified image data" refers to the generation of adjusted image frames following the brightness modification process, suitable for broadcasting or further processing. This may involve re-encoding the adjusted image data in a broadcastcompatible format, such as H.264, H.265, or MPEG-4, for efficient transmission over broadcast channels. In one example, the modified image data is streamed directly to a broadcast display pipeline, providing real-time visual enhancements that reflect the brightness adjustments made to each region of the frame.

[0045] The term "edge detection" refers to any technique that identifies boundaries within the image, typically by detecting sudden changes in pixel intensity. This may include algorithms such as Canny, Sobel, or Laplacian edge detection, or similar techniques used to define the borders of key image elements. In one example, Canny edge detection is employed to distinguish the edges of regions requiring brightness modification, ensuring that critical elements, such as players or game equipment, are preserved without alteration during brightness adjustments. DESCRIPTION OF DRAWINGS

[0046] The present invention relates to a method for dynamically adjusting brightness levels in broadcast images, specifically for live sports events held in stadiums with variable lighting conditions. This method addresses the limitations of conventional brightness adjustment techniques, which typically apply uniform changes across an entire broadcast frame and fail to account for localized differences in luminance due to structural shading or sunlight exposure. By providing an adaptive, region-based approach to brightness correction, the invention ensures balanced visibility across both sunlit and shaded areas within a stadium broadcast, delivering a consistently clear image regardless of environmental lighting disparities.

[0047] Traditional image processing methods used in broadcasting, including ambient light sensors and global brightness adjustments, are limited in their ability to respond to the complex lighting dynamics of outdoor stadiums. Uniform brightness adjustments often lead to unsatisfactory results, with overexposed areas appearing washed out and underexposed areas remaining poorly visible. Additionally, these methods lack the granularity to apply selective corrections, leading to diminished image quality and viewer dissatisfaction. The invention's region-specific approach directly addresses these shortcomings by analyzing each frame to detect and classify regions with differing luminance, applying brightness modifications tailored to the needs of each specific area.

[0048] Using a neural network model, the invention identifies regions within the broadcast frame that deviate from desired brightness levels, categorizing them as either overexposed or underexposed. The neural network model is trained to accurately assess brightness discrepancies typical of stadium environments, where structural elements, time of day, and weather conditions contribute to varying light exposures across the field. By segmenting the broadcast frame into regions and applying targeted adjustments, the invention preserves the clarity of adequately lit areas while enhancing visibility in both overexposed and underexposed sections. This selective adjustment minimizes the impact on regions that already exhibit optimal luminance, thus preserving contrast and detail across the entire frame.

[0049] The invention is designed for real-time processing, enabling brightness adjustments on a frame-by-frame basis as lighting conditions evolve during a live broadcast. This adaptability allows the system to respond dynamically to changes, such as shifting sunlight or passing cloud cover, ensuring a stable viewing experience without the disruptions typically associated with manual adjustments. The invention further enhances efficiency by selectively processing frames with identified brightness discrepancies, conserving computational resources when adjustments are unnecessary.

[0050] In FIG. 1, the flowchart depicts the invention's method for dynamically adjusting brightness levels within a stadium broadcast frame. The process initiates by receiving image data 100, which comprises a live broadcast frame captured from a stadium camera feed. This image data represents the full stadium environment, including areas with both sunlight exposure and shading. For optimal compatibility with the neural network model, the image data may be encoded in broadcast-standard formats, such as H.264 or H.265, allowing efficient frame-by-frame processing. In some implementations, the data may undergo preprocessing 102 to further enhance compatibility. Preprocessing operations may include resizing the image frame to match the input dimensions required by the neural network model or converting the image to grayscale if color information is deemed unnecessary for brightness detection. Alternative implementations may bypass preprocessing if the received data aligns closely with the model's input specifications, conserving processing time.

[0051] Following data reception, the method proceeds to analyze image data 104 to detect brightness discrepancies within the frame. A neural network model, such as a convolutional neural network (CNN) specifically trained on varied stadium lighting conditions, performs this analysis. The neural network assesses pixel luminance levels throughout the frame, identifying significant variations that signify brightness discrepancies. By identifying clusters of pixels with extreme brightness values, the neural network model can accurately locate overexposed (bright, sunny) and underexposed (dark, shaded) regions within the frame. In instances where no brightness discrepancy is detected, the method advances directly to outputting image data without modification, thus optimizing computational efficiency.

[0052] For frames where brightness discrepancies are detected, the method segments the broadcast frame into multiple regions 106. This segmentation process typically involves using edge-detection algorithms, such as Canny edge detection to create customized regions that align with natural lighting boundaries within the frame, enhancing segmentation accuracy in scenes with highly variable lighting conditions.

[0053] After segmentation, each region is classified 108 based on luminance levels, categorizing areas as either overexposed or underexposed relative to a predetermined brightness threshold. These threshold values can be adjusted dynamically according to external lighting information, stadium-specific metrics, or feedback from prior frames, ensuring precise and context-sensitive classification of each region. In one implementation, regions exceeding a threshold brightness level are marked as overexposed, while those falling below a separate threshold are classified as underexposed. This step ensures that only the regions requiring adjustment are modified, preserving the original image quality in well-balanced regions.

[0054] Key boundaries within each region may also be identified via feature extraction and taken into account, such as those of players, the ball, or goalposts, and prevents these elements from undergoing brightness adjustments. In this manner, the system ensures that the details essential to viewer experience are maintained. In an alternative implementation, object detection techniques like YOLO (You Only Look Once) may identify and exclude specific objects from brightness adjustments, further preserving the clarity of significant broadcast elements within the stadium scene.

[0055] Once regions are classified, the method applies region-specific brightness adjustments 110. In overexposed regions, a first brightness modification is applied, potentially using gamma correction with a value of less than one, reducing luminance while retaining visual details. Conversely, underexposed regions undergo a second brightness modification, often through a gamma correction with a value greater than one to enhance visibility without introducing excess noise. Other brightness adjustment techniques, such as tone mapping or pixel intensity manipulation, may be used depending on processing resources and desired visual outcomes. Regions deemed adequately lit are left unmodified, which ensures that the process does not affect image areas that already possess balanced brightness levels, thereby preserving overall frame integrity.

[0056] The final stage, outputting modified image data 112, produces a broadcast frame with adjusted brightness levels across targeted regions. The modified data is formatted for compatibility with broadcast standards and may be encoded in H.264, H.265, or other industry-standard formats, ensuring efficient transmission. This output allows broadcasters to deliver a balanced, real-time visual experience to viewers, where both overexposed and underexposed areas of the frame are optimized for enhanced clarity, regardless of stadium lighting variations.

[0057] FIG. 2 illustrates an example system architecture block diagram of the software components configured to execute the method for dynamically adjusting brightness levels within a stadium broadcast frame as described in FIG. 1. The system architecture includes interconnected modules designed for real-time image data processing, brightness discrepancy detection, region-based segmentation, brightness adjustment, and final data output. This modular architecture enables flexible, scalable implementation and facilitates integration with existing broadcast systems.

[0058] The architecture begins with the Image Data Reception Module 200, which receives broadcast frame data from an external camera feed capturing the stadium scene. This module manages image data acquisition in real-time, often through a direct input from a broadcast feed, ensuring consistent frame updates. In one implementation, the data received here may be in high-definition (HD) or ultra-high-definition (UHD) video formats encoded in H.264 or H.265. The Image Data Reception Module 200 serves as the initial entry point for frames and directs them to the Preprocessing Module 202.

[0059] The Preprocessing Module 202 is responsible for preparing the incoming image data to be compatible with subsequent analysis components, particularly the neural network model. This module may perform operations such as resizing, color normalization, and resolution adjustments, depending on the configuration and input requirements of the neural network. For example, it might resize frames to a standard resolution (e.g., 1920x1080 pixels) to match the input layer dimensions of the neural network model, or convert frames to grayscale if color data is not required for brightness analysis. In an alternative implementation, if the incoming data is already optimized for neural network input, this module may bypass certain preprocessing steps, streamlining the process for enhanced efficiency.

[0060] Once preprocessing is complete, the prepared image data is sent to the Brightness Discrepancy Detection Module 204, which houses a Convolutional Neural Network (CNN) 206 specifically trained to identify brightness discrepancies within the frame. The CNN 206, designed for high accuracy in varied stadium lighting conditions, assesses pixel luminance throughout the frame to detect regions that are either overexposed or underexposed. This CNN model may be trained on a dataset of stadium images featuring diverse lighting conditions, enabling it to recognize patterns indicative of brightness discrepancies caused by sunlight and shadow variations. The output of the CNN 206 is a set of brightness classifications, each corresponding to a specific region of the frame. In one implementation, if the CNN 206 identifies no significant brightness discrepancies within the frame, the system can skip further processing steps, directing the frame directly to the Output Module 214, thereby conserving computational resources.

[0061] For frames requiring brightness adjustments, the system proceeds to the Segmentation Module 208, which divides the frame into multiple regions to facilitate localized brightness adjustments. The Segmentation Module 208 employs adaptive segmentation methods using edge detection, such as Canny edge detection, to define custom-shaped regions that align closely with natural lighting boundaries within the stadium frame caused by sunny and shady areas. The segmentation module may also enact processes for feature preservation, identifying significant boundaries within the frame, including players, field markings, and other relevant broadcast details, and ensuring that the brightness adjustments do not obscure essential details, enhancing clarity for viewers. Additionally, feature extraction techniques, such as object detection models (e.g., YOLO), may be employed to further identify and exclude key objects from brightness adjustments.

[0062] The segmented frame data, now divided into regions, is then forwarded to the Brightness Classification Module 210. This module, in conjunction with the CNN's initial classifications, confirms and refines the brightness level of each region, categorizing them as either overexposed, underexposed, or adequately lit. The Brightness Classification Module 210 may incorporate additional criteria, such as dynamically adjustable threshold values, based on ambient lighting conditions or stadium-specific metrics, to provide context-sensitive classification. For example, threshold levels may be adjusted in real-time according to feedback from environmental sensors monitoring external lighting levels, allowing for responsive adaptation to changing conditions in outdoor stadium settings.

[0063] After classification, each region is processed by the Brightness Adjustment Module 212, which applies specific brightness modifications based on the identified classification. For overexposed regions, the module applies a first brightness adjustment, potentially using gamma correction with a gamma value less than one, effectively reducing luminance while preserving visual detail. For underexposed regions, a second brightness adjustment is applied, possibly using a gamma correction with a gamma value greater than one to enhance visibility without introducing excessive noise. The Brightness Adjustment Module 212 may also incorporate additional techniques such as tone mapping, pixel intensity manipulation, or histogram equalization if required by specific implementation needs. This module selectively bypasses adjustments for adequately lit regions, ensuring that only areas in need of correction are modified, thereby preserving the overall integrity of the broadcast frame.

[0064] Finally, the processed frame data is passed to the Output Module 214, where modified image data is prepared for broadcast. The Output Module 214 encodes the adjusted frame data into a broadcast-compatible format, such as H.264 or H.265, ensuring compatibility with existing broadcast systems and minimal latency in real-time streaming. This encoding process allows the adjusted frames to seamlessly integrate into the broadcast pipeline, providing viewers with an optimized image that balances sunlight and shadow, maintaining visibility across the stadium regardless of lighting conditions.

[0065] FIG. 3A shows the initial image data 300 received from the broadcast camera feed, representing an unprocessed frame of a live sports broadcast. In this frame, there is a noticeable variation in lighting across the stadium field, with the upper regions receiving direct sunlight, causing overexposure, while other areas, particularly towards the bottom and right, are in shadow. This imbalance in lighting makes it difficult for viewers to clearly observe the action in both brightly lit and shaded areas, highlighting the need for targeted brightness adjustments. This original frame serves as a baseline reference before any modifications are applied by the system.

[0066] FIG. 3B illustrates the outcome of the brightness discrepancy detection step 302, where the system identifies the "too sunny" problem areas within the frame. Here, a neural network model, specifically a convolutional neural network (CNN), has analyzed pixel brightness across the frame, identifying the high-luminance areas that exceed a predefined brightness threshold. These overexposed regions 304 are highlighted in black for illustrative purposes, indicating that they are excessively bright and obscuring details within the frame. The CNN processes pixel intensities and aggregates clusters of high-luminance pixels, accurately detecting regions affected by direct sunlight and designating them for brightness reduction in subsequent steps.

[0067] FIG. 3C shows the identification of the "too shady" regions 306, presented during the classification and segmentation phase. This image has undergone a thresholding operation to isolate underexposed areas, where pixel brightness values fall below a certain threshold. These low-luminance, or "too shady," areas are displayed in black in the image, visually distinguishing them from adequately lit and overexposed areas. This thresholded view assists in identifying sections of the frame that require brightness enhancement to improve visibility. By classifying both "too sunny" and "too shady" regions, the system ensures that brightness adjustments are precisely applied to the necessary parts of the frame while preserving naturally balanced areas.

[0068] FIG. 3D depicts the final output frame 308, following the application of regionspecific brightness adjustments. Here, the system has reduced brightness in the overexposed "too sunny" regions identified in FIG. 3B and increased brightness in the underexposed "too shady" regions identified in FIG. 3C. These adjustments have been made using brightness modification techniques such as gamma correction or pixel intensity adjustment, resulting in a visually balanced frame. The previously overexposed areas now display reduced glare, enhancing detail visibility, while the shaded areas are brightened for better contrast and clarity. This modified image data, which retains natural brightness in adequately lit areas, represents the final processed frame ready for broadcast, providing viewers with a clearer and more enjoyable visual experience. CONTROLLER / PROCESSOR COMPONENTS

[0069] The software operations described herein may be carried out by any suitable processor or controller. A processor or controller as described herein may include any suitable type of computing device, such as a central processing unit (CPU), microcontroller, graphics processing unit (GPU), system on a chip (SoC), or digital signal processor (DSP). It may operate with one or more cores and may be configured to execute the functions described in this disclosure.

[0070] The processor may be operably connected to one or more memory devices, such as random access memory (RAM), read-only memory (ROM), flash storage, or solid-state drives (SSD). These memory devices store computer-readable instructions that, when executed by the processor, perform the methods described. The processor and memory communicate via data buses or other suitable communication pathways.

[0071] The computing device may also include input / output (I / O) devices, such as a touchscreen, mouse, keyboard, display, or speaker, to facilitate interaction with users or other systems. Additionally, it may include a network interface, such as a wired or wireless communication module, for connecting to networks.

[0072] Control logic or software instructions may be stored in memory and executed by the processor to implement specific functionalities. This logic may be modular, consisting of software components, processes, or functions that work together to perform the operations described herein.

[0073] The described computing operations involve the manipulation of data represented as electrical, optical, or magnetic signals stored or transferred within the system. These operations are machine-executed and do not require manual intervention, though they may interface with human operators through appropriate user interfaces.

[0074] The systems and methods described are not limited to any particular hardware configuration or programming language and may be implemented on general-purpose or specialized computing devices. CONCLUSION

[0075] Unless otherwise defined, all terms (including technical terms) used herein have the same meaning as commonly understood by one having ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and the present disclosure and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

[0076] The disclosed embodiments are illustrative, not restrictive. While specific configurations of the method and system of the invention have been described in a specific manner referring to the illustrated embodiments, it is understood that the present invention can be applied to a wide variety of solutions which fit within the scope and spirit of the claims. There are many alternative ways of implementing the invention.

[0077] It is to be understood that the embodiments of the invention herein described are merely illustrative of the application of the principles of the invention. Reference herein to details of the illustrated embodiments is not intended to limit the scope of the claims, which themselves recite those features regarded as essential to the invention.

Claims

What is claimed is:

1. A computer-implemented method for processing image data, the method comprising:receiving image data representing a broadcast frame captured from a video feed of a stadium environment;analyzing, by a neural network model, brightness levels within the received image data to identify a brightness discrepancy across regions of the broadcast frame, wherein the brightness discrepancy includes a first region having a luminance level above a predetermined threshold and a second region having a luminance level below the predetermined threshold;segmenting the broadcast frame into a plurality of regions based on the identified brightness discrepancy;classifying each of the plurality of regions as either an overexposed region or an underexposed region based on respective luminance levels within each region;adjusting, for each overexposed region, the luminance level by applying a first brightness modification to reduce the luminance level;adjusting, for each underexposed region, the luminance level by applying a second brightness modification to increase the luminance level; andoutputting modified image data based on the adjusted luminance levels of each region in the broadcast frame.

2. The method of claim 1, wherein analyzing brightness levels within the received image data further comprises preprocessing the image data to normalize the frame resolution and adjust color data to optimize compatibility with the neural network model.

3. The method of claim 1, wherein the neural network model is a convolutional neural network trained on a dataset of stadium images exhibiting varied lighting conditions to accurately distinguish overexposed and underexposed regions.

4. The method of claim 1, further comprising dynamically segmenting the broadcast frame into regions of varied grid density, wherein grid density is increased in areas of complex lighting variations and decreased in areas of uniform lighting.

5. The method of claim 1, wherein the classification of each region as either overexposed or underexposed is based on a threshold brightness range, the values of which are dynamically adjusted according to ambient lighting information or precalculated metrics specific to the stadium environment.

6. The method of claim 1, wherein the first brightness modification for each overexposed region comprises applying a gamma correction with a gamma value less than one to reduce luminance while preserving visual detail.

7. The method of claim 1, wherein the second brightness modification for each underexposed region comprises applying a gamma correction with a gamma value greater than one to increase luminance without introducing excessive noise.

8. The method of claim 1, further comprising redistributing brightness within the broadcast frame by calculating a surplus brightness from the overexposed regions and reallocating it to underexposed regions, thereby balancing luminance without uniformly increasing brightness across the entire frame.

9. The method of claim 1, wherein each overexposed region is adjusted by selectively decreasing pixel intensity values within the region to achieve a predetermined target luminance level.

10. The method of claim 1, wherein each underexposed region is adjusted by selectively boosting pixel intensity values within the region to achieve a predetermined target luminance level while preserving contrast.

11. The method of claim 1, further comprising using edge detection to identify boundaries within each region, wherein brightness adjustments are selectively applied to avoid modifying detected edges and thereby maintain visual fidelity for key elements within the broadcast frame.

12. The method of claim 11, wherein the edge detection is performed using a Canny edge detection algorithm to define boundaries between sunny and shady areas, as well as to identify objects within the frame.

13. The method of claim 1, wherein the adjusted brightness levels are applied in real time to each broadcast frame, allowing continuous adaptation of brightness adjustments in response to changing lighting conditions within the stadium environment.

14. The method of claim 1, further comprising dynamically skipping brightness adjustments for frames in which the neural network model does not detect significant brightness discrepancy, outputting unmodified image data in such cases to conserve processing resources.

15. The method of claim 1, wherein the neural network model is configured to output a confidence score for each brightness classification, and brightness adjustments are only applied when the confidence score meets a predetermined threshold.

16. The method of claim 1, further comprising an edge-preserving feature extraction technique that identifies and excludes important elements, including players and game equipment, from brightness adjustments within each region.

17. The method of claim 1, wherein the neural network model is integrated within the camera hardware used to capture the broadcast frame, enabling in-camera detection and adjustment of brightness discrepancies.

18. The method of claim 1, further comprising encoding the modified image data in a broadcast-compatible format selected from H.264, H.265, or other industry-standard video formats for transmission.

19. The method of claim 1, wherein the brightness adjustments for overexposed and underexposed regions are carried out using tone mapping to improve contrast and detail within each region.

20. The method of claim 1, wherein the segmentation of the broadcast frame into regions is performed based on calculated lighting boundaries, and regions areshaped to follow natural lighting divisions within the stadium environment rather than a fixed grid layout.s

Images