Brightness adjustment method for night vision camera

Abstract

The present invention relates to the technical field of image data processing, and more specifically, to a brightness adjustment method for a night vision camera. The method comprises: obtaining a target ROI region of each image frame in a video image sequence of the night vision camera; normalizing the product of a gradient mean value and texture complexity of all pixel points in a sub-block of the target ROI region to obtain a complexity feature of the sub-block; using an optical flow vector of a center point of the sub-block as a reference motion vector of the sub-block, and calculating a cosine similarity entropy value between the optical flow vector of each pixel point and the reference motion vector; obtaining an importance feature of the sub-block according to a mean value of the optical flow vector module length of all pixel points in the sub-block and the cosine similarity entropy value; calculating a weight of the sub-block; and weighting the brightness value of each sub-block based on the weight of the sub-block in the target ROI region of the image frame to achieve brightness adjustment of the night vision camera, thereby effectively improving the accuracy of brightness adjustment of the night vision camera.

Description

Technical Field

[0001] The present invention relates to the field of image data processing technology, and more particularly to a method for adjusting the brightness of a night vision camera. Background Art

[0002] Night vision cameras are devices that provide clear surveillance images at night or in low-light environments, typically using infrared light sources. During operation, infrared night vision cameras use infrared light-emitting diodes to emit invisible infrared light, which is reflected by objects and captured by the camera. This light is converted into an electrical signal, generating the surveillance image. To ensure the quality of the images captured by the camera, the infrared light intensity must be adaptively adjusted based on the ambient light level and the brightness of the monitored image to avoid overexposure or underexposure.

[0003] There is a lot of research and application of night vision cameras in the prior art. For example, the patent application document with publication number CN119052656A discloses a method, device, electronic device and computer-readable medium for dynamically adjusting the brightness of the infrared lamp of a wireless camera. The application controls the wireless camera to turn on the infrared lamp on the wireless camera according to a first preset duty cycle threshold to shoot, thereby obtaining a scene image; determines the exposure value of the wireless camera when shooting the scene image; and adjusts the brightness of the infrared lamp based on the scene image, the exposure value and the preset exposure value range.

[0004] The above-mentioned existing technology adjusts the brightness of the infrared lamp through the scene image, exposure value and preset exposure value range. However, when adjusting the brightness of the camera's infrared light source to capture night environment monitoring images, the brightness distribution in the image is not uniform. If the adjustment is directly based on the preset exposure value range, the brightness adjustment effect of the night vision camera will be poor, resulting in overexposure or underexposure.

[0005] Based on this, how to accurately adjust the brightness of a night vision camera is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0006] In order to solve the above technical problem of how to accurately achieve brightness adjustment of a night vision camera, the present invention proposes a brightness adjustment method for a night vision camera, which includes the following steps:

[0007] The image frame in the video image sequence of a night vision camera and the ROI area of ​​the previous image frame of the image frame are taken as the target ROI area of ​​the image frame; the product of the gradient mean and texture complexity of all pixels in the sub-block of the target ROI area is normalized to obtain the complexity feature of the sub-block; the optical flow vector of the center point of the sub-block is used as the reference motion vector of the sub-block, and the cosine similarity entropy value of the optical flow vector of each pixel and the reference motion vector is calculated; the importance feature of the sub-block is obtained according to the mean of the modulus of the optical flow vector of all pixels in the sub-block and the cosine similarity entropy value; the weight of the sub-block is obtained according to the complexity feature and importance feature of the sub-block; the brightness value of each sub-block is weighted based on the weight of the sub-block in the target ROI area of ​​the image frame to achieve brightness adjustment of the night vision camera.

[0008] The present invention adjusts the brightness of a night vision camera by acquiring the brightness performance of each sub-block within the ROI region of an image frame of a night vision camera video image sequence. This process takes into account the varying brightness performance of different sub-blocks within the ROI region, and therefore their varying importance when calculating the image frame brightness. Based on this, the present invention determines the contribution of each sub-block to the image frame brightness by acquiring the complexity and importance of each sub-block within the ROI region, effectively improving the accuracy of night vision camera brightness adjustment.

[0009] According to a brightness adjustment method for a night vision camera provided by the present invention, the method of unioning an image frame in a video image sequence of the night vision camera with the ROI area of ​​the previous image frame of the image frame as the target ROI area of ​​the image frame also includes: preprocessing the video image captured by the night vision camera to obtain each image frame of the video image sequence; and dividing the target ROI area of ​​the image frame into multiple sub-blocks based on a preset sub-block size.

[0010] The present invention takes into account that the edges of the originally captured image frames may be deformed, which may affect subsequent image processing. Therefore, the quality of the image is improved through preprocessing to prepare for subsequent image processing.

[0011] According to the present invention, a brightness adjustment method for a night vision camera and a method for obtaining the ROI area of ​​an image frame include: using a Gaussian mixture model to model the background of a video image sequence, extracting a moving target in the image frame through a background subtraction method, and using the circumscribed rectangle of the moving target as the ROI area of ​​the image frame.

[0012] According to the present invention, a brightness adjustment method for a night vision camera and a pixel gradient acquisition method are provided, which include: calculating the gradient of the pixel using a Sobel operator.

[0013] According to a brightness adjustment method for a night vision camera provided by the present invention, the method for obtaining the texture complexity includes: taking the entropy value of each grayscale pair in the grayscale co-occurrence matrix of the sub-block as the texture complexity of the sub-block.

[0014] The present invention takes into account that the texture complexity in the sub-block is related to the grayscale features in the sub-block. Therefore, the complexity of the image grayscale distribution is measured by obtaining the entropy value in the grayscale co-occurrence matrix. When all values ​​in the co-occurrence matrix are equal or the pixel values ​​show the greatest randomness, the entropy value in the co-occurrence matrix reaches the maximum. Based on this, the texture complexity of the sub-block can be accurately obtained.

[0015] According to a brightness adjustment method for a night vision camera provided by the present invention, the brightness value of each sub-block is weighted based on the weight of the sub-block in the target ROI area of ​​the image frame, including: taking the product of the weight and the brightness value of the sub-block in the target ROI area as the weighted value of the sub-block, and taking the weighted average of all sub-blocks in the target ROI area of ​​the image frame as the brightness value of the image frame;

[0016] in, ; is the weight of the i-th sub-block, 、 are respectively the complexity feature and importance feature of the i-th sub-block, is an exponential function with base e.

[0017] According to a brightness adjustment method for a night vision camera provided by the present invention, the brightness value of each sub-block is weighted based on the weight of the sub-block in the target ROI area of ​​the image frame to achieve brightness adjustment of the night vision camera, including: obtaining a brightness gain parameter by the ratio of the brightness value of the image frame to a preset brightness value, and adjusting the brightness of the camera based on the brightness gain parameter.

[0018] The present invention adjusts the brightness of the camera by obtaining the brightness gain parameter of the image frame, so that the camera can automatically adjust the gain value according to the brightness change of the collected image frame, thereby ensuring that a clear monitoring picture is obtained at night.

[0019] According to a brightness adjustment method for a night vision camera provided by the present invention, the brightness adjustment of the night vision camera is achieved, and then the method further includes: performing abnormal monitoring on the video image sequence of the night vision camera.

[0020] By automatically detecting and marking abnormal events, the present invention can timely discover and respond to potential security threats and reduce potential security risks.

[0021] The present invention has the following beneficial effects:

[0022] Based on the above technical solution, the present invention can adjust the brightness of a night vision camera by obtaining the brightness performance of each sub-block in the ROI region of the image frame of the night vision camera's video image sequence. In this process, the present invention takes into account the different brightness performances of different sub-blocks in the ROI region, and therefore their different importance when calculating the image frame brightness. Based on this, the present invention obtains the complexity and importance of each sub-block in the ROI region to determine the contribution of each sub-block to the image frame brightness, effectively improving the accuracy of night vision camera brightness adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an illustrative and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0024] Figure 1 A schematic flow chart of a method for adjusting the brightness of a night vision camera provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0026] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0027] Night vision cameras are devices that provide clear surveillance images at night or in low-light environments, typically using infrared light sources. During operation, infrared night vision cameras use infrared light-emitting diodes to emit invisible infrared light, which is reflected by objects and captured by the camera. This light is converted into an electrical signal, generating the surveillance image. To ensure the quality of the images captured by the camera, the infrared light intensity must be adaptively adjusted based on the ambient light level and the brightness of the monitored image to avoid overexposure or underexposure.

[0028] However, when adjusting the brightness of the camera's infrared light source to capture nighttime environment monitoring images, the brightness distribution in the image is not uniform. If adjustments are made directly based on the preset exposure value range, the brightness adjustment effect of the night vision camera will be poor, resulting in overexposure or underexposure.

[0029] Based on this, an embodiment of the present invention discloses a brightness adjustment method for a night vision camera. This method analyzes the image complexity and importance of each sub-block in the image frame captured by the camera, obtains the weight of each sub-block when determining the brightness value of the image frame, thereby accurately obtaining the brightness value required for each image frame, and can effectively improve the accuracy of brightness adjustment of the night vision camera.

[0030] For details, please see Figure 1 As shown, Figure 1 A flowchart of a brightness adjustment method for a night vision camera provided in an embodiment of the present invention is provided. The method specifically includes the following steps.

[0031] S1: An image frame in a night vision camera video image sequence and an ROI region of a previous image frame are combined as a target ROI region of the image frame.

[0032] For example, in an embodiment of the present invention, an image frame in a video image sequence of a night vision camera and the ROI area of ​​the previous image frame of the image frame are combined as the target ROI area of ​​the image frame, and the process also includes: pre-processing the video image taken by the night vision camera to obtain each image frame of the video image sequence.

[0033] The preprocessing may be geometric transformation, image enhancement, etc., which may be specifically configured according to actual needs, and the embodiment of the present invention does not impose any excessive restrictions thereon.

[0034] It's important to note that the video image sequences captured by night vision cameras contain a lot of content. Image frames contain both a region of interest (ROI) (which requires attention) and background areas (which don't). To reduce background interference and reduce data processing, you can adjust the camera's brightness by focusing only on the ROI within the image frame.

[0035] It should be further explained that there may be fast-moving objects in the image frame. It is difficult to identify such targets only through the ROI area of ​​the current image frame, and the characteristics of pixels in different areas are different. If the ROI area is analyzed as a whole, the characteristics of pixels in some areas may be blurred.

[0036] Based on this, an embodiment of the present invention obtains the union of the ROI areas of the current image frame and its previous image frame as the target ROI area of ​​the current image frame, divides the target ROI area into multiple sub-blocks, and analyzes the pixel features of each sub-block in the target ROI area.

[0037] For example, in an embodiment of the present invention, a method for obtaining the ROI area of ​​an image frame includes: using a Gaussian mixture model to perform background modeling on a video image sequence, extracting a moving target in the image frame through a background subtraction method, and using the circumscribed rectangle of the moving target as the ROI area of ​​the image frame.

[0038] Specifically, when using a Gaussian mixture model to model the background of a video image sequence, the number of Gaussian distributions required for each pixel can be preset; for each frame and each pixel in the video image sequence, its pixel value is compared with each Gaussian distribution in the Gaussian mixture model to determine whether it matches a certain distribution; if the pixel value matches a certain Gaussian distribution, the mean, covariance matrix and weight of the distribution are updated, and the weight is increased to reflect the importance of the distribution to background modeling in the near future; if the pixel value does not match all Gaussian distributions, the least important distribution is selected for replacement according to the weight, and a smaller weight is set; finally, the Gaussian distribution with a larger weight and a smaller variance is used as the representative of the background area, and a Gaussian mixture model that can be used to identify the foreground area is obtained.

[0039] The number of Gaussian distributions required for each pixel can be set according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this.

[0040] For example, when extracting a moving target in an image frame by using a background subtraction method, a difference operation may be performed between the image frame and the background model, and the obtained difference image may be recorded as the moving target.

[0041] After the target ROI region of the image frame is obtained based on the above steps, each sub-block in the target ROI region may be analyzed.

[0042] For example, in an embodiment of the present invention, the target ROI region of the image frame may be divided into a plurality of sub-blocks based on a preset sub-block size.

[0043] The sub-block size may be preset to 10×10. The sub-block size may be specifically set according to actual needs, and the embodiment of the present invention does not impose any excessive restrictions thereon.

[0044] It should be noted that characteristics that affect image frame brightness include image complexity and importance. Low brightness results in underexposure, leading to loss of image detail. Higher brightness improves clarity but also increases the risk of overexposure. After obtaining sub-blocks of the target ROI region in each image frame based on the above steps, the following steps can be performed to analyze the image complexity and importance of each sub-block to accurately adjust the image frame brightness.

[0045] S2: Normalize the product of the gradient mean of all pixels in the sub-block of the target ROI area and the texture complexity to obtain the complexity feature of the sub-block.

[0046] It's important to note that complex textures within image frame sub-blocks can interfere with the human eye's perception of clarity, causing the image to appear blurry or difficult to discern details. The greater the texture complexity, the greater the impact on image frame clarity. Gradient information measures the rate of change in pixel values ​​within an image. Richer gradient information within an image frame sub-block indicates greater variation in pixel values ​​across different directions, and the area is likely to appear sharper. A higher mean gradient value within a sub-block indicates greater detail and edge information within the sub-block, thus significantly impacting image frame clarity.

[0047] Based on this, after the embodiment of the present invention obtains the sub-block of the target ROI area in each image frame based on the above steps, the complexity of the sub-block can be characterized by the texture complexity and gradient change degree in the sub-block. The higher the texture complexity in the sub-block and the higher the gradient mean of all pixels in the sub-block of the target ROI area, the higher the complexity of the sub-block and the higher the complexity feature.

[0048] For example, in an embodiment of the present invention, a method for obtaining a gradient of a pixel point includes: calculating the gradient of the pixel point by using a Sobel operator.

[0049] The specific steps of obtaining the gradient of a pixel point by using the Sobel operator can be implemented by existing technologies, and will not be described in detail in the embodiment of the present invention.

[0050] It can be understood that the gray level co-occurrence matrix can extract the texture features in the sub-blocks and generate a matrix that describes the texture features of the image.

[0051] For example, in an embodiment of the present invention, a method for obtaining texture complexity includes: using the entropy value of each grayscale pair in the grayscale co-occurrence matrix of the sub-block as the texture complexity of the sub-block.

[0052] When obtaining the gray-level co-occurrence matrix of a sub-block, the distance can be set to 1 pixel, the direction to 0°, and the initial gray-level co-occurrence matrix normalized to obtain the gray-level co-occurrence matrix of the sub-block. The distance and direction can be set according to actual needs. The specific steps for obtaining the gray-level co-occurrence matrix of the sub-block can be implemented using existing technologies and are not further described in detail in this embodiment of the present invention.

[0053] For example, in an embodiment of the present invention, the complexity characteristics of the sub-blocks are calculated, and specific reference may be made to the following relationship:

[0054] ;

[0055] is the complexity characteristic of the i-th sub-block, is the gradient mean of all pixels in the i-th sub-block, is the texture complexity of the i-th sub-block, is a linear normalization function.

[0056] In the above formula, the larger the gradient mean of all pixels in the i-th sub-block is, the more details and edge information the i-th sub-block may contain, and the greater the impact on the clarity of the image frame.

[0057] The greater the texture complexity of the i-th sub-block, the more complex the pixel grayscale distribution in the i-th sub-block is, and the more complex the texture is, the greater the impact on the clarity of the entire region of interest.

[0058] In summary, if the gradient mean of all pixels in the i-th sub-block is larger and the texture complexity is higher, it means that the i-th sub-block is more sensitive to brightness adjustment. When adjusting the brightness of the camera infrared light, more attention should be paid to this sub-block to maintain the integrity of the image details, and the corresponding complexity feature is also greater.

[0059] Based on the above steps, the complexity features of each sub-block in the target ROI area of ​​the image frame can be obtained. However, the texture features and gradient features in the sub-blocks are static features. Static features are more sensitive to noise in the image frame. If there is noise interference in the image frame captured by the night vision camera, adjusting the brightness only based on the static features of the image frame may identify the noise as grayscale features or texture features, reducing the accuracy of the camera brightness adjustment.

[0060] Therefore, the embodiment of the present invention further analyzes the importance features in the sub-blocks as dynamic features through the following steps, and combines the dynamic features and static features to obtain the weight of the sub-block for adjusting the brightness of the image frame.

[0061] S3: Obtain the importance features of the sub-block through the optical flow vector change features of the pixels in the sub-block.

[0062] It should be noted that in a video image frame, when a moving object changes, calculating the motion component of each pixel forms a motion vector field, also known as an optical flow field. The optical flow vector is a two-dimensional vector that describes the motion of each pixel in the motion vector field. The optical flow vector of a moving object in the target ROI region is generally highly significant. Therefore, the optical flow vector of the pixels in the sub-block can reflect the motion information of the object in the image. However, in the case of poor image quality or jitter, the optical flow vector may become unstable.

[0063] Based on this, the embodiment of the present invention can obtain the dynamic characteristics of each sub-block by analyzing the consistency and change speed of the optical flow vectors of the pixels in the sub-blocks of the target ROI area, thereby characterizing the importance characteristics of each sub-block.

[0064] For example, in an embodiment of the present invention, the optical flow vector of the center point of the sub-block is used as the reference motion vector of the sub-block, and the cosine similarity entropy value of the optical flow vector of each pixel point and the reference motion vector is calculated; the product of the mean normalized value of the optical flow vector module length of all pixels in the sub-block and the cosine similarity entropy value is used as the importance feature of the sub-block.

[0065] The optical flow vector of the pixel point can be obtained by using the LK algorithm (Lucas-Kanade algorithm). The specific steps can be implemented by existing technologies and will not be described in detail in the embodiment of the present invention.

[0066] It can be understood that the center point of the sub-block is also a pixel point.

[0067] For example, in the embodiment of the present invention, the importance feature of the sub-block is determined, and the details can be referred to the following relationship:

[0068] ;

[0069] is the importance feature of the i-th sub-block, is the cosine similarity entropy value between each pixel in the i-th sub-block and the reference motion vector, is the mean value of the modulus of the optical flow vectors of all pixels in the i-th sub-block, It is an S-shaped growth curve function.

[0070] in, Function used for normalization.

[0071] In the above formula, the cosine similarity entropy value between each pixel in the i-th sub-block and the reference motion vector represents the degree of disorder of the optical flow vector of the pixels in the i-th sub-block. The larger the value, the greater the disorder of the optical flow vector of the pixels in the i-th sub-block, and the higher the possibility of out-of-focus or blur, which leads to a decrease in the clarity of the image frame. Therefore, the i-th sub-block has a higher importance feature when adjusting the clarity of the image frame. Conversely, the smaller the value, the higher the consistency of the optical flow vector of the pixels in the i-th sub-block, the more stable the optical flow vector, and other sub-blocks with higher disorder can be given priority when adjusting the clarity of the image frame.

[0072] The mean of the optical flow vector moduli of all pixels in the i-th sub-block represents the speed of pixel motion in that sub-block. A larger value indicates faster pixel motion in the i-th sub-block, leading to a greater likelihood of motion blur and reduced image clarity. Therefore, when adjusting image clarity, this sub-block should be prioritized, and its corresponding feature importance should be higher.

[0073] After respectively obtaining the complexity feature and importance feature of each sub-block based on the above steps, the weight of each sub-block in determining the brightness of the image frame can be obtained, that is, continue to perform the following steps.

[0074] S4: Calculate the weight of each sub-block.

[0075] For example, in the embodiment of the present invention, the weight of each sub-block is determined, and the specific details can be referred to the following relationship:

[0076] ;

[0077] is the weight of the i-th sub-block, is the complexity characteristic of the i-th sub-block, is the importance feature of the i-th sub-block, is an exponential function with base e.

[0078] In the above formula, the higher the complexity feature and importance feature of the i-th sub-block, the greater the influence of the sub-block on the clarity of the image frame, the greater its importance in calculating the brightness of the image frame, and the higher the corresponding weight.

[0079] It represents the sum of the importance of all sub-blocks in the target ROI area. The larger the value, the higher the importance of the target ROI area of ​​the image frame.

[0080] After analyzing the features of each sub-block in the target ROI region based on the above steps to obtain the corresponding weight, the brightness of each sub-block can be weighted based on the weight, thereby accurately obtaining the brightness of the target ROI region.

[0081] S5: Weighting the brightness value of each sub-block based on the weight of the sub-block in the target ROI area of ​​the image frame to achieve brightness adjustment of the night vision camera.

[0082] For example, in an embodiment of the present invention, the brightness value of each sub-block is weighted based on the weight of the sub-block in the target ROI area of ​​the image frame, including: taking the product of the weight and the brightness value of the sub-block in the target ROI area as the weighted value of the sub-block, and taking the weighted average of all sub-blocks in the target ROI area of ​​the image frame as the brightness value of the image frame.

[0083] For example, in an embodiment of the present invention, the brightness value of each sub-block is weighted based on the weight of the sub-block in the target ROI area of ​​the image frame to achieve brightness adjustment of the night vision camera, including: obtaining a brightness gain parameter by the ratio of the brightness value of the image frame to a preset brightness value, and adjusting the brightness of the camera based on the brightness gain parameter.

[0084] Among them, adjusting the brightness of the camera based on the brightness gain parameter can be achieved by a proportional controller, an integral controller and a differential controller. The specific steps can be achieved by existing technologies and will not be described in detail in the embodiments of the present invention.

[0085] It should be noted that the larger the brightness gain parameter is, the better it is. Too high a gain will lead to increased image noise, and too low a gain may lead to insufficient brightness enhancement of the image frame, affecting the results of human eye observation. Therefore, when adjusting the brightness of the night vision camera, you can also set the brightness gain parameter threshold.

[0086] Among them, the brightness gain parameter threshold can be set to The brightness gain parameter threshold can be set according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this.

[0087] It is understandable that the video image sequence captured by the night vision camera contains important information. In order to ensure the safety of the area monitored by the night vision camera, the video image sequence of the night vision camera can be monitored for abnormalities.

[0088] For example, in an embodiment of the present invention, brightness adjustment of a night vision camera is implemented, and then the method further includes: performing abnormality monitoring on a video image sequence of the night vision camera.

[0089] For example, when performing abnormality monitoring on a night vision camera video image sequence, the video image sequence may be input into a pre-trained abnormality monitoring model to obtain an abnormality monitoring result of the night vision camera video image sequence.

[0090] Specifically, when pre-training an anomaly monitoring model, a video dataset containing normal and abnormal data can be collected first, and the anomalies in the video dataset can be manually labeled to obtain an annotated video dataset; the annotated video dataset is input into the neural network model for training to obtain an anomaly monitoring model that can be used for anomaly monitoring.

[0091] Among them, the neural network model can be a convolutional neural network, a recurrent neural network, etc., which can be set according to actual needs.

[0092] Among them, the specific steps of performing abnormal monitoring on the night vision camera video image sequence can be implemented by existing technologies, and the embodiments of the present invention will not be described in detail here.

[0093] It can be seen that in an embodiment of the present invention, when adjusting the brightness of a night vision camera, the image frame in the video image sequence of the night vision camera and the ROI area of ​​the previous image frame of the image frame can be taken as the target ROI area of ​​the image frame; the product of the gradient mean and texture complexity of all pixels in the sub-block of the target ROI area is normalized to obtain the complexity feature of the sub-block; the optical flow vector of the center point of the sub-block is used as the reference motion vector of the sub-block, and the cosine similarity entropy value of the optical flow vector of each pixel and the reference motion vector is calculated; the importance feature of the sub-block is obtained according to the mean of the optical flow vector module length of all pixels in the sub-block and the cosine similarity entropy value; the weight of the sub-block is obtained according to the complexity feature and importance feature of the sub-block; and the brightness value of each sub-block is weighted based on the weight of the sub-block in the target ROI area of ​​the image frame to achieve brightness adjustment of the night vision camera.

[0094] Thus, the embodiment of the present invention can achieve brightness adjustment for a night vision camera by obtaining the brightness performance of each sub-block in the ROI region of an image frame of a night vision camera video image sequence. During this process, the embodiment of the present invention takes into account the different brightness performances of different sub-blocks in the ROI region, and therefore their different levels of importance when calculating the image frame brightness. Based on this, the embodiment of the present invention obtains the complexity and importance of each sub-block in the ROI region to determine the contribution of each sub-block to the image frame brightness, effectively improving the accuracy of night vision camera brightness adjustment.

[0095] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for adjusting the brightness of a night vision camera, characterized in that: include: The ROI region of an image frame in a night vision camera video image sequence and the ROI region of a previous image frame of the image frame are combined as a target ROI region of the image frame; Normalize the product of the gradient mean of all pixels in the sub-block of the target ROI area and the texture complexity to obtain the complexity feature of the sub-block; The optical flow vector of the center point of the sub-block is used as the reference motion vector of the sub-block, and the cosine similarity entropy value between the optical flow vector of each pixel in the sub-block and the reference motion vector is calculated, which can represent the degree of confusion of the optical flow vector of the pixel in the sub-block; The importance feature of the sub-block is obtained according to the mean value of the optical flow vector modulus of all pixels in the sub-block and the cosine similarity entropy value; The mean value of the optical flow vector modulus of all pixels in a sub-block represents the motion speed of the pixels in the sub-block; The weight of the sub-block is obtained according to the complexity and importance characteristics of the sub-block, satisfying the relationship: , is the weight of the i-th sub-block, 、 are respectively the complexity feature and importance feature of the i-th sub-block, is an exponential function with base e; The sum of the importance of all sub-blocks in the target ROI area indicates the importance of the target ROI area in the image frame; The brightness value of each sub-block is weighted based on the weight of the sub-block in the target ROI area of ​​the image frame to achieve brightness adjustment of the night vision camera, including: taking the product of the weight and the brightness value of the sub-block in the target ROI area as the weighted value of the sub-block, taking the weighted average of all sub-blocks in the target ROI area of ​​the image frame as the brightness value of the image frame; obtaining a brightness gain parameter by the ratio of the brightness value of the image frame to a preset brightness value, and adjusting the brightness of the camera based on the brightness gain parameter.

2. The brightness adjustment method of a night vision camera according to claim 1, characterized in that: Before the ROI region of the image frame in the night vision camera video image sequence is combined with the ROI region of the previous image frame of the image frame as the target ROI region of the image frame, the method further includes: The video images captured by the night vision camera are preprocessed to obtain image frames of the video image sequence.

3. The brightness adjustment method of a night vision camera according to claim 2, characterized in that: The method for obtaining the ROI region of an image frame includes: The Gaussian mixture model is used to model the background of the video image sequence. The moving target in the image frame is extracted by background subtraction method, and the bounding rectangle of the moving target is used as the ROI area of ​​the image frame.

4. The brightness adjustment method of a night vision camera according to claim 1, characterized in that: The pixel gradient acquisition method includes: The gradient of the pixel is calculated using the Sobel operator.

5. The brightness adjustment method of a night vision camera according to claim 1, characterized in that: The method for obtaining the texture complexity includes: The entropy value of each gray-level pair in the gray-level co-occurrence matrix of the sub-block is used as the texture complexity of the sub-block.

6. The brightness adjustment method of a night vision camera according to claim 1, characterized in that: After the brightness adjustment of the night vision camera is implemented, the method further includes: Perform abnormality monitoring on night vision camera video image sequences.

Images