A camera adaptive anti-overexposure real-time processing method, system, medium and device
An adaptive overexposure prevention method based on real-time human detection and brightness adjustment solves the problem of overexposure of faces or bodies in infrared night vision surveillance, achieving efficient image quality control in complex scenes, reducing equipment costs and improving robustness.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HANGZHOU EBOYLAMP ELECTRONICS CO LTD
- Filing Date
- 2023-06-29
- Publication Date
- 2026-07-24
AI Technical Summary
Existing technologies struggle to effectively handle overexposure of faces or bodies in infrared night vision surveillance, especially in complex scenarios, leading to the loss of detailed facial or human body information. Furthermore, existing methods require high computing power from equipment or lack environmental adaptability, which can easily result in abnormal image brightness.
An adaptive real-time overexposure prevention method is adopted, which dynamically adjusts the brightness to avoid overexposure of faces or bodies by real-time human detection, pre-calculation of brightness adjustment value and judgment of ambient light intensity changes. Combined with confidence threshold and edge detection algorithm, the image quality is ensured.
It achieves adaptive overexposure protection for human motion in complex scenes, reduces equipment costs, improves image quality stability and robustness, avoids image flicker and brightness abnormalities, and is more adaptable.
Smart Images
Figure CN116996775B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer vision, and more specifically to a real-time adaptive anti-overexposure processing method for cameras. Background Technology
[0002] Automatic exposure refers to a technology in which a camera or camcorder automatically adjusts exposure parameters based on the lighting conditions of the scene. By measuring information such as the brightness and color of light, it automatically adjusts parameters such as shutter speed, aperture size, and ISO of the camera or camcorder to ensure that the brightness and contrast of the image are appropriate and to avoid overexposure or underexposure. Generally, the average brightness method is used, which adjusts the average brightness of the image to a certain level, which is called the target brightness.
[0003] When the overall brightness of the image meets the target brightness, if a face or human body appears in the foreground or other positions in the image, the reflected infrared intensity will be strong, causing the face or human body to be too bright and the image to be overexposed. However, faces or human bodies often occupy a small area in the monitoring image and their impact on the average brightness of the image is usually not significant. Therefore, they will not trigger the conventional automatic exposure algorithm to reduce the brightness value. As a result, this overexposure situation is difficult to handle under the conventional exposure algorithm.
[0004] Specifically, infrared night vision cameras typically rely on front-end infrared illuminators for illumination. When the environment is dark or open, the infrared image is generally dim, requiring an increase in illuminator power or image signal gain to improve overall image brightness. However, increased image brightness often leads to overexposure of human targets. When a human is near the camera, the overexposure of the human body or face is particularly severe due to the concentrated infrared illumination in the center area. Overexposure results in the loss of detailed human information, making it impossible to identify the target's identity, behavior, or other information. Therefore, it is necessary to suppress overexposure while maintaining adequate image brightness.
[0005] To avoid overexposure during automatic exposure, the following methods can be used: Changing the metering mode: Different metering modes are suitable for different scenes. Changing the metering mode can adjust the exposure parameters and avoid overexposure; Adjusting exposure compensation: Exposure compensation can adjust the camera or camcorder's exposure parameters to make the image closer to the user's desired brightness and contrast, thus avoiding overexposure; Using filters: Filters can reduce the intensity of light, thus avoiding overexposure; Using a flash: In low-light environments, a flash can be used to provide additional light, thus avoiding overexposure.
[0006] Overexposure prevention is a common problem, and existing technologies offer various targeted solutions:
[0007] CN114723632A disclosed a method and apparatus for correcting abnormal exposure images of parts based on texture information on April 1, 2022. The method based on texture information takes advantage of the fact that texture information in overexposed areas will be severely lost. It judges whether the observed object in the image is overexposed by the complexity of the texture information. However, this method is only suitable for industrial vision fields with relatively stable environments and is not applicable to complex scenarios such as video surveillance.
[0008] CN110889321A disclosed an intelligent face overexposure prevention algorithm on September 29, 2019. Based on face recognition, the algorithm performs real-time face detection on the image. When a face is detected in the image, the anti-overexposure module is turned on, and when no face is detected, the anti-overexposure module is turned off. The processing effect is good and can balance the brightness detection of static images and the anti-overexposure suppression effect of moving targets' faces. However, face detection requires a certain amount of computing power from the device, which will increase the cost of the device. At the same time, the recognition ability will be affected when the face is occluded.
[0009] A more common technical solution is to statistically analyze the global brightness of the image and use global pixel brightness distribution information and brightness histogram information to determine the overexposure of the image. When it is determined that there are overexposed objects in the image, the infrared lamp supplementary light power is reduced or the image signal gain is reduced to suppress the overall brightness of the image and ensure the clarity of human body and face. However, in actual development, this method needs to balance the anti-overexposure intensity and video monitoring brightness and does not have adaptive adjustment function.
[0010] Overexposure prevention methods based on global brightness statistics cannot distinguish whether overexposed objects in the image are people or other objects. When highly reflective objects such as walls or billboards appear in the image, or when there are obstructions such as branches or fallen leaves on the camera lens that cause overexposure, the global overexposure prevention module will activate, suppressing the overall brightness of the image. This will reduce the brightness of the monitored subject area to below a suitable level, while further loss of brightness and detail in dark areas. This method also places requirements on the camera's installation position and angle. If the camera's supplementary light is close to a wall, overexposure due to wall reflection is likely to occur. In this case, the overexposure prevention method based on global brightness statistics will control the overexposure of the wall, but the brightness of the monitored subject area will be far below normal. Therefore, the disadvantages of this method are insufficient environmental adaptability, easy occurrence of abnormal image brightness, and difficulty in balancing the overexposure prevention effect and the brightness information in dark areas.
[0011] Face detection-based methods use the presence or absence of a face in the image as the criterion for switching on / off the anti-overexposure module. This method places higher demands on the processor of the surveillance camera equipment, but it can overcome the aforementioned problems to some extent. However, face detection-based methods have many limitations under infrared night vision. Besides the limited recognition capability of face recognition modules when people are moving rapidly or their faces are obscured, if a person suddenly appears close to the camera already in a fully overexposed state, the face recognition algorithm will also struggle to effectively recognize the face and thus fail to activate the anti-overexposure module to suppress the overexposure. Furthermore, this method is ineffective when the target is wearing a helmet or is walking away from the surveillance camera, making it difficult to suppress human overexposure. Summary of the Invention
[0012] To address the aforementioned problems in the existing technology, this invention provides a camera adaptive anti-overexposure real-time processing method, system, medium, and device, which has stronger environmental adaptability, can handle more complex human motion scenarios, and has lower requirements for device computing power, avoiding abnormal problems such as failure or oscillation, and the processing system has high robustness.
[0013] In a first aspect, embodiments of this application provide a camera adaptive overexposure anti-real-time processing method, including the following steps:
[0014] A. Real-time human detection in camera footage;
[0015] B, When a valid human figure is detected in the camera frame, pre-calculate the brightness adjustment value required during the anti-overexposure process;
[0016] C. Perform anti-overexposure processing on the camera image based on the pre-calculated value of the brightness adjustment value;
[0017] D. Determine whether to end the anti-overexposure processing, including: calculating the brightness change between the current frame and the previous frame of the camera image. If the brightness change is greater than a preset threshold, it is considered that the human body has left, the anti-overexposure processing ends, and the process returns to step A. If the brightness change is not greater than the preset threshold, it is considered that the human body has not left, and the anti-overexposure processing of the camera image continues.
[0018] When the calculated change in ambient light intensity exceeds a preset threshold, it is assumed that the human body has rapidly left the field of view, and the anti-overexposure processing ends, thus improving the overall brightness of the image. When the calculated change in ambient light intensity does not exceed the preset threshold, the anti-overexposure processing continues to be applied to the camera image, avoiding the problem of periodic exposure flickering in the image.
[0019] In one alternative to the first aspect, before step A, there is a step of setting restriction information, the restriction information including a preset human body area detection range, and in step A, real-time human body detection is performed within the preset human body area detection range.
[0020] In another alternative to the first aspect, in step B, a standard brightness value and a preset percentage are set. When a valid human figure is detected in the camera image, if the percentage of the bright area in the camera image whose brightness exceeds the standard brightness value exceeds the preset percentage, the required brightness adjustment value is calculated. The brightness adjustment value can make the bright area in the camera image meet the preset percentage.
[0021] In another alternative to the first aspect, after a valid human figure is detected in the camera image, the cumulative confidence value of the valid human figure in the camera image is accumulated. In step B, when the cumulative confidence value is higher than a preset high threshold, the brightness adjustment value required for the anti-overexposure process is pre-calculated.
[0022] In another alternative to the first aspect, step D further includes: when the cumulative confidence value is lower than a preset low threshold, the anti-overexposure process is terminated.
[0023] In another alternative to the first aspect, in step D, the brightness value of the camera image is calculated using the ambient light, and the calculation formula is:
[0024] luma = 200 * log 10 (Y*100*100 / ISO) / shutter
[0025] Where luma represents ambient brightness, Y represents the brightness value of the camera image, ISO represents sensitivity, and shutter represents shutter speed.
[0026] In another alternative to the first aspect, step D further includes using a brightness difference signal as texture information to supplement the judgment on whether the human body has left the human-shaped region, and using an edge detection algorithm to obtain gradient information of the human-shaped region, and supplementing the judgment on whether the human body has left the human-shaped region based on the change of gradient information. The edge detection algorithm can be the Sobel operator or other edge detection algorithms with the same function.
[0027] Secondly, embodiments of this application provide a camera adaptive overexposure anti-real-time processing system, including a human detection module, a pre-calculation module, an overexposure anti-modification module, and a judgment module connected in sequence;
[0028] Human detection module: used for real-time human detection in camera footage;
[0029] Pre-calculation module: Used to pre-calculate the brightness adjustment value required during the anti-overexposure process when a valid human figure is detected in the camera image;
[0030] Anti-overexposure module: Used to process camera images to prevent overexposure using pre-calculated brightness adjustment values;
[0031] Judgment module: Used to determine whether to end the anti-overexposure processing, including: calculating the brightness change between the current frame and the previous frame of the camera image. If the brightness change is greater than a preset threshold, it is considered that the human body has left and the anti-overexposure processing ends; if the brightness change is not greater than the preset threshold, it is considered that the human body has not left and the anti-overexposure processing of the camera image continues.
[0032] Thirdly, embodiments of this application provide a computer storage medium storing a computer program, which includes program instructions. When executed by a processor, the program instructions can implement a camera adaptive anti-overexposure real-time processing method provided by the first aspect or any implementation of the first aspect of this application.
[0033] Fourthly, embodiments of this application provide an electronic device, including a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform a camera adaptive anti-overexposure real-time processing method provided by the first aspect or any implementation thereof of the embodiments of this application.
[0034] The beneficial technical effects of this invention include:
[0035] A camera adaptive overexposure anti-overexposure real-time processing method is proposed. It adopts the calculation method of the pre-computation module and has a fast response speed to handle fast-moving overexposed targets. The adaptive overexposure anti-overexposure method linked with human detection can handle scenes with people and scenes without people respectively. When there are people and the human body is overexposed, the brightness of the image is suppressed and the global contrast is improved to salvage information in dark areas. When there are no people, the overall brightness of the image is improved to improve the overall image signal-to-noise ratio. The parameters are adaptively configured for each scene to obtain the best image effect.
[0036] To determine whether a human body is moving or stationary, and to prevent the adaptive anti-overexposure module from shutting down or oscillating when the human figure is stationary, this method improves stability and practicality, enhances the robustness of the processing system, and enables motion estimation-based human detection algorithms to detect human targets that have moved but are now stationary.
[0037] The adaptive linkage anti-overexposure method has stronger environmental adaptability than methods based on global statistics, and can handle more complex human motion scenes than face recognition-based anti-overexposure methods, while requiring less computing power from the device, thus reducing equipment costs. To address the issue of false detection of human-like targets by human detection algorithms, a motion-filtering-based human detection algorithm is adopted. Furthermore, to address the problem that motion-filtering-based human detection algorithms cannot effectively detect stationary human targets in the image, a static human target detection method is proposed, avoiding the failure or vibration issues caused by the anti-overexposure module failing to detect stationary human figures. Attached Figure Description
[0038] 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.
[0039] Figure 1 This is a flowchart of a camera adaptive overexposure anti-real-time processing method according to the present invention;
[0040] Figure 2 This is a block diagram of a camera adaptive overexposure prevention real-time processing system according to the present invention. Detailed Implementation
[0041] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.
[0042] In the following description, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The following description provides multiple embodiments of this application, which can be substituted or combined with each other. Therefore, this application can also be considered to include all possible combinations of the same and / or different embodiments described. Thus, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then this application should also be considered to include embodiments containing one or more other possible combinations of A, B, C, and D, even if such embodiments are not explicitly described in the following text.
[0043] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes may be made to the function and arrangement of the described elements without departing from the scope of this application. Various processes or components may be appropriately omitted, substituted, or added to the examples. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Furthermore, features described with respect to some examples may be combined into other examples.
[0044] Example 1:
[0045] Reference Figure 1 A real-time adaptive overexposure protection method for cameras includes the following steps:
[0046] Step A: Perform real-time human detection on the camera feed.
[0047] Before step A, there is also a step: setting restriction information, which includes a preset human body area detection range. In step A, real-time human body detection is performed within the preset human body area detection range.
[0048] Step B: When a valid human figure is detected in the camera frame, the required brightness adjustment value for the anti-overexposure processing is pre-calculated. A standard brightness value and a preset percentage are set. When a valid human figure is detected in the camera frame, if the percentage of the bright area in the camera frame that exceeds the standard brightness value exceeds the preset percentage, the required pre-calculated brightness adjustment value is calculated. This pre-calculated brightness adjustment value ensures that the bright area in the camera frame meets the preset percentage.
[0049] You can directly adjust the brightness to quickly achieve the anti-overexposure effect, or you can set a certain convergence step size to gradually adjust to the preset percentage to obtain a more balanced anti-overexposure effect.
[0050] Ideally, the anti-overexposure module should activate when a valid human figure is detected in the image to suppress overexposure of the figure or face, and deactivate when the figure leaves to restore overall brightness. However, in complex monitoring scenarios, false detections or missed detections of human figures frequently occur. If the anti-overexposure module is frequently triggered to turn on and off, users will clearly perceive image flickering. To ensure the smooth switching of the anti-overexposure module, the inventors designed an upper and lower threshold accumulation method to control the anti-overexposure switch.
[0051] The system sets a confidence level, a preset high threshold, and a preset low threshold. When a valid human figure is detected in the camera frame, the confidence level increases. When the confidence level exceeds the preset high threshold, the camera frame undergoes anti-overexposure processing, and the brightness mapping coefficient in dark areas is improved through adaptive gamma linkage. When no valid human figure is detected in the camera frame, the confidence level decreases. When the confidence level falls below the preset low threshold, the anti-overexposure processing ends, and the gamma mapping curve is restored to normal levels, improving the overall brightness of the image.
[0052] Once a valid human figure is detected in the camera frame, the cumulative confidence value of the valid human figure in the camera frame is calculated. In step B, when the cumulative confidence value is higher than a preset high threshold, the brightness adjustment value required for the anti-overexposure processing is pre-calculated. Further, in step B, when the cumulative confidence value is lower than a preset low threshold, the anti-overexposure processing ends.
[0053] Step C: Perform anti-overexposure processing on the camera image based on the pre-calculated brightness adjustment value.
[0054] Human detection algorithms typically require motion detection as a prerequisite to filter out false positives and negatives for humanoid objects. Motion detection can be based on pixel grayscale gradient information, such as optical flow. When a human figure is in continuous motion, the pixel brightness in the moving area changes significantly, thus effectively detecting the human figure. However, this method can only effectively handle continuously moving human targets. Once the human figure stops moving in the frame, it cannot be detected, causing the anti-overexposure process to terminate. In particular, if the anti-overexposure process stops and the image brightness increases even though the human figure has not moved, the change in brightness gradient information in the human figure area will cause the motion filtering condition to be met, triggering the anti-overexposure process again. This can lead to periodic exposure flickering in the image. To address this, the inventors implemented the processing described in step D.
[0055] Step D, determining whether to end the anti-overexposure processing, includes: calculating the brightness change between the current frame and the previous frame of the camera image; if the brightness change is greater than a preset threshold, it is considered that the human body has left, the anti-overexposure processing ends, and the process returns to step A; if the brightness change is not greater than the preset threshold, it is considered that the human body has not left, and the anti-overexposure processing of the camera image continues.
[0056] Step D further includes: ending the anti-overexposure processing when the cumulative confidence value is lower than a preset low threshold. It also includes using a brightness difference signal as texture information to supplement the judgment of whether the human body has left the human-shaped region, and using an edge detection algorithm to obtain gradient information of the human-shaped region, supplementing the judgment of whether the human body has left the human-shaped region based on changes in the gradient information.
[0057] The brightness value of the camera image is calculated using the ambient light level. The formula is as follows:
[0058] luma = 200 * log 10 (Y*100*100 / ISO) / shutter
[0059] Where luma represents ambient brightness, Y represents the brightness value of the camera image, ISO represents sensitivity, and shutter represents shutter speed.
[0060] Actual testing showed that when a human body was detected in the image and met the valid judgment criteria, the anti-overexposure module was activated to effectively suppress the brightness of overexposed areas. After the human body left, the module was deactivated to improve overall brightness and transparency. The image brightness stabilized after the human body came to a standstill. In unoccupied scenes, the anti-overexposure module did not malfunction when the image was obscured by dead branches or fallen leaves, proving the effectiveness, feasibility, and practicality of the above method.
[0061] This method addresses the issues of overall low image brightness and loss of detail in dark areas caused by mishandling overexposed non-human objects in the image. It is more applicable and less expensive than face detection-based methods. It also overcomes the problem of the anti-overexposure module shutting down or flickering due to the difficulty of identifying stationary human figures by moving human figure detection algorithms, thus improving the robustness and practicality of the method.
[0062] Example 2:
[0063] A camera adaptive overexposure anti-overexposure real-time processing system, based on a camera adaptive overexposure anti-overexposure real-time processing method provided in Embodiment 1, includes a human detection module, a pre-calculation module, an overexposure anti-overexposure module and a judgment module connected in sequence;
[0064] Human detection module: used for real-time human detection in camera footage;
[0065] Pre-calculation module: Used to pre-calculate the brightness adjustment value required during the anti-overexposure process when a valid human figure is detected in the camera image;
[0066] Anti-overexposure module: Used to process camera images to prevent overexposure using pre-calculated brightness adjustment values;
[0067] Judgment module: Used to determine whether to end the anti-overexposure processing, including: calculating the brightness change between the current frame and the previous frame of the camera image. If the brightness change is greater than a preset threshold, it is considered that the human body has left and the anti-overexposure processing ends; if the brightness change is not greater than the preset threshold, it is considered that the human body has not left and the anti-overexposure processing of the camera image continues.
[0068] Example 3:
[0069] A computer-readable storage medium storing computer instructions for causing a computer to execute a camera adaptive overexposure anti-real-time processing method as described in Embodiment 1.
[0070] Example 4:
[0071] An electronic device includes a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform a real-time adaptive anti-overexposure processing method for cameras as proposed in Embodiment 1.
[0072] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope of the present invention.
Claims
1. A real-time adaptive overexposure protection method for cameras, characterized in that, Including the following steps: A. Real-time human detection in camera footage; B, When a valid human figure is detected in the camera frame, pre-calculate the brightness adjustment value required during the anti-overexposure process; C. Perform anti-overexposure processing on the camera image based on the pre-calculated value of the brightness adjustment value; D. Determine whether to end the anti-overexposure processing, including: when the human detection module cannot detect a human figure, calculate the amount of information change between the current frame and the previous frame of the camera image. The amount of information change includes brightness change. If the amount of information change is greater than a preset threshold, it is considered that the human figure has left, the anti-overexposure processing ends, and the process returns to step A. If the amount of information change is not greater than the preset threshold, it is considered that the human figure has not left, and the anti-overexposure processing of the camera image continues. Once a valid human figure is detected in the camera frame, the cumulative confidence value of the valid human figure in the camera frame is calculated. In step B, when the cumulative confidence value is higher than the preset high threshold, the brightness adjustment value required for the anti-overexposure process is pre-calculated. Step D also includes: when the cumulative confidence value is lower than a preset low threshold, the anti-overexposure process is terminated; Step D also includes using a brightness difference signal as texture information to supplement the judgment on whether the human body has left the human-shaped region, and using an edge detection algorithm to obtain gradient information of the human-shaped region, and supplementing the judgment on whether the human body has left the human-shaped region based on the changes in gradient information.
2. The camera adaptive overexposure anti-real-time processing method according to claim 1, characterized in that, Before step A, there is also a step: setting restriction information, which includes a preset human body area detection range. In step A, real-time human body detection is performed within the preset human body area detection range.
3. The camera adaptive overexposure anti-real-time processing method according to claim 1, characterized in that, In step B, a standard brightness value and a preset percentage are set. When a valid human figure is detected in the camera image, if the percentage of the bright area in the camera image that exceeds the standard brightness value exceeds the preset percentage, the required brightness adjustment value is calculated. The brightness adjustment value can make the bright area in the camera image meet the preset percentage.
4. The camera adaptive overexposure anti-real-time processing method according to claim 1, characterized in that, In step D, the information change includes changes in screen brightness and changes in ambient light intensity. Ambient light intensity is calculated using screen brightness to supplement the judgment of whether the human figure has left the human-shaped area. The calculation formula is: luma=200*log10(Y*100*100 / ISO) / shutter Where luma represents ambient brightness, Y represents the brightness value of the camera image, ISO represents sensitivity, and shutter represents shutter speed.
5. A camera adaptive overexposure anti-real-time processing system, based on the camera adaptive overexposure anti-real-time processing method according to any one of claims 1-4, characterized in that, It includes a human figure detection module, a pre-calculation module, an overexposure prevention module, and a judgment module connected in sequence; Human detection module: used for real-time human detection in camera footage; Pre-calculation module: Used to pre-calculate the brightness adjustment value required during the anti-overexposure process when a valid human figure is detected in the camera image; Anti-overexposure module: Used to process camera images to prevent overexposure using pre-calculated brightness adjustment values; Judgment module: used to determine whether to end the anti-overexposure processing, including: when the human detection module cannot detect a human figure, calculating the amount of information change between the current frame and the previous frame of the camera image, the amount of information change includes brightness change; if the amount of information change is greater than a preset threshold, it is considered that the human figure has left and the anti-overexposure processing ends. If the change in information is not greater than a preset threshold, it is assumed that the human body has not left, and the camera image is continuously subjected to anti-overexposure processing.
6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform a camera adaptive overexposure anti-real-time processing method as described in any one of claims 1-4.
7. An electronic device, characterized in that, The system includes a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform a camera adaptive overexposure anti-processing method as described in any one of claims 1-4.