Dynamic scene context modeling method based on box plot

Abstract

The application discloses a dynamic scene background modeling and foreground detection method based on a box plot, which comprises the following steps: saving the obtained video images into a background queue according to a dynamic frame extraction speed; based on the current background queue state, removing a moving interference target by using a box plot statistical method, modeling the background after removing the interference target, and obtaining a background model; and performing difference operation on each frame of the obtained video images by using the background model to obtain a foreground; and experiments prove that the dynamic scene background modeling method based on the box plot can effectively improve the modeling efficiency and controllability, and has important significance in the actual dynamic scene foreground detection task.

Description

Technical Field

[0001] This invention belongs to the field of background modeling, specifically relating to a dynamic scene background modeling method based on box plots. Background Technology

[0002] With the development of computer vision technology, video surveillance has been widely applied in fields such as intelligent transportation, security monitoring, and industrial automation. Among these applications, background modeling methods, as a fundamental image processing technique, are of great significance for achieving target detection, tracking, and analysis in dynamic video surveillance.

[0003] The core task of background modeling methods is to separate foreground objects (such as pedestrians and vehicles) from the background scene in dynamic video sequences for subsequent processing and analysis. This process is valuable for reducing data volume, lowering computational complexity, and improving the accuracy of target recognition and tracking. By modeling the background, the differences between the target object and the background can be effectively distinguished, thereby enabling the detection and tracking of the target object.

[0004] However, in practical applications, background modeling methods face numerous challenges, especially in dynamic scenes, where the difficulty and complexity of background modeling are even more pronounced. The motion of dynamic backgrounds (such as rustling grass or pedestrians walking) in video sequences can interfere with background modeling, making it difficult to distinguish between dynamic backgrounds and real foreground targets, thus reducing the accuracy of target detection and tracking. Furthermore, traditional background modeling methods generally require maintaining a static scene during the modeling phase, failing to automatically filter out moving interfering targets appearing in the scene. Additionally, controlling the static time threshold for incorporating a new stationary target into the background is difficult when dealing with scene changes, such as the appearance of a new stationary target. Summary of the Invention

[0005] To address the problem that dynamic backgrounds (such as rustling grass or pedestrians walking) in video sequences can interfere with background modeling, making it difficult to distinguish between dynamic backgrounds and real foreground targets, thus reducing the accuracy of target detection and tracking, traditional background modeling methods generally require a static scene during the modeling phase, failing to automatically filter out moving interfering targets. Furthermore, controlling the static time threshold for incorporating a new static target into the scene as background is difficult when dealing with scene changes. To solve these problems, this invention provides a method for dynamic scene background modeling and foreground detection based on box plots, comprising the following steps:

[0006] A method for dynamic scene background modeling and foreground detection based on box plots includes the following steps:

[0007] The acquired video images are saved into the background queue according to the dynamic frame extraction speed;

[0008] Based on the current background queue status, the box plot statistical method is used to remove moving interference targets, and the background after removing interference targets is modeled to obtain the background model.

[0009] Each frame of the acquired video is subjected to a difference operation using a background model to obtain the foreground.

[0010] Furthermore, the process of using box plot statistical methods to remove and model moving interference targets based on the current background queue state to obtain the background model is as follows:

[0011] T21. Take the current background queue state and divide the background queue into blocks in the long and wide dimensions, hereinafter referred to as background queue sub-blocks;

[0012] T22. Use the box plot statistical method to count the pixel values ​​of the background queue sub-blocks one by one, find the frames in each block that capture moving interference targets, and remove the frames that capture moving interference targets from the background queue sub-blocks.

[0013] T23. Model the background queue sub-blocks after removal using statistics, and then combine them into a background model of the original image size.

[0014] Furthermore, the process of saving the acquired video images into the background queue according to the dynamic frame extraction speed is as follows:

[0015] Frames are acquired from the camera. Every n frames, a frame is added to the background queue, and the oldest background frame at the head of the queue is popped. The value of n is determined by the speed of the moving interference targets that need to be removed in the scene. To make the method compatible with cameras with different frame rates, the value of n is related to the frame rate and determined by the following formula:

[0016] n=F*t (1)

[0017] In the formula, F is the frame rate of the camera or video, t is the time difference between two frames in the background queue, and the value of n is ultimately determined by specifying t.

[0018] Furthermore, the background queue is divided into blocks in both the long and wide dimensions, and the granularity 'a' of the blocks is determined based on the size of the moving target appearing in the camera frame.

[0019] Furthermore, the process of sequentially using box plot statistics to count the pixel values ​​of the background queue sub-blocks, identifying the frames in each block that captured moving interference targets, and then removing the frames that captured moving interference targets from the background queue sub-blocks is as follows:

[0020] Calculate the lower quartile Q1, upper quartile Q3, and interquartile range iqr of the background queue sub-block in the queue dimension, and calculate the upper and lower whiskers according to the following formulas as the criteria for defining normal and outlier values:

[0021] lower_bound=Q1 - 1.5 * iqr (2)

[0022] upper_bound = Q3 + 1.5 * iqr (3)

[0023] Where: lower_bound is the lower whisker line and upper_bound is the upper whisker line;

[0024] Values ​​outside the upper and lower whisker lines are considered outliers, meaning that the frame sub-block has captured a moving interference target. All frames containing outliers in the background queue sub-block are removed, and the background queue after removal consists entirely of clean background frames without foreground.

[0025] Furthermore, the statistical background queue direction maximum and minimum values.

[0026] Furthermore, for each frame of the acquired video, a difference operation is performed using a background model to obtain the foreground process as follows:

[0027] Get the next frame image;

[0028] Based on the background model, the background is updated, and the current background model is used to determine whether each point in the image is within the range of the background model. Pixels that are not within the background range are considered as foreground.

[0029] Furthermore, the lower limit of the permissible speed of the moving interference target in the background model is:

[0030]

[0031] In the formula: W is the distance the moving interference target moves within a sub-block in its direction of motion; t is the time difference between two frames in the background queue, in seconds; m is the total length of the background queue; Ω is the allowable anomaly rate of the box plot, which determines the maximum percentage of outliers that the box plot can tolerate when it can normally remove outliers from the queue.

[0032] Furthermore, the lower limit of the static duration for which a stationary target is absorbed into the background model is:

[0033] T>m*t*Ω (5)

[0034] Static targets whose static duration exceeds this lower limit will be absorbed as background.

[0035] Furthermore, the background model includes upper and lower limits for the background value of each pixel position. Each pixel in the frame to be detected is compared with its corresponding upper and lower limits. If a pixel is outside the background value range, it is considered a foreground position. The comparison is based on the following formula:

[0036]

[0037] In the formula: X is the image to be detected; b0 is the 0th dimension of the background model, that is, the maximum value of each position in the background queue after removal; b1 is the 1st dimension of the background model, that is, the minimum value of each position in the background queue after removal; λ is the noise reduction factor, which needs to be determined according to the noise fluctuation range captured by the video lens. Devices prone to noise need to have this value increased appropriately.

[0038] This invention provides a dynamic scene background modeling and foreground detection method based on box plots, which has the following advantages: This method eliminates the adverse effects of moving interference targets on background modeling during the modeling stage, and makes the lower limits of the moving speed of targets to be removed and the lower limits of the static time of stationary objects to be absorbed controllable; after using box plots to remove abnormal frames from the background queue, statistical modeling is performed using the remaining clean background queue. Experiments demonstrate that this box plot dynamic scene background modeling method can effectively improve modeling efficiency and controllability, and has significant implications for practical dynamic scene foreground detection tasks; this method is an efficient, stable, and controllable background modeling method suitable for dynamic scenes, and has significant research significance and practical application value. Attached Figure Description

[0039] 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart of the dynamic scene background modeling method based on box plots of the present invention;

[0041] Figure 2 (a) is the background queue collected by thread A, (b) is the background upper limit map after filtering out interfering targets, and (c) is the background upper limit map after filtering out interfering targets.

[0042] Figure 3 (a) An image containing a moving target, (b) Foreground detection results for thread C. Detailed Implementation

[0043] It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0044] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0046] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of the invention. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following figures denote similar items; therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0047] In the description of this invention, it should be understood that the orientation or positional relationship indicated by directional terms such as "front, back, up, down, left, right", "horizontal, vertical, horizontal" and "top, bottom" is generally based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing this invention and simplifying the description. Unless otherwise stated, these directional terms do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this invention. The directional terms "inner" and "outer" refer to the inner and outer contours relative to the outline of each component itself.

[0048] For ease of description, spatial relative terms such as "above," "over," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation besides the orientation of the device as described in the figures. For example, if the device in the figures is inverted, a device described as "above" or "above" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatial relative descriptions used herein will be interpreted accordingly.

[0049] Furthermore, it should be noted that the use of terms such as "first" and "second" to define components is merely for the purpose of distinguishing the corresponding components. Unless otherwise stated, the above terms have no special meaning and therefore should not be construed as limiting the scope of protection of this invention.

[0050] Figure 1 This is a flowchart of the dynamic scene background modeling method based on box plots of the present invention;

[0051] This invention provides a dynamic scene background modeling and foreground detection method based on box plots, specifically including:

[0052] T1, Thread A saves the acquired video images into the background queue according to the dynamic frame extraction rate; the video images are captured by the camera.

[0053] T2 and thread B, based on the current background queue state, use box plot statistical methods to remove moving interference targets and model the background after removing interference targets to obtain a background model.

[0054] T3. Perform a difference operation on each frame of the acquired video using the background model to obtain the foreground.

[0055] The threads T1 / T2 / T3 are executed sequentially, with T2 starting after T1 updates at least the same length as the background queue, and T3 starting after T2 completes at least one modeling operation.

[0056] Thread B copies the current state of the background queue saved by thread A as a local variable for background modeling, while thread A continues to update the global background queue with new images from the camera so that subsequent modeling can use the latest background queue.

[0057] The process by which thread A captures images from the camera and saves them into the background queue according to the dynamic frame extraction rate is as follows:

[0058] Every n frames, the image from the current camera is added to a global background queue for use by other threads, and the oldest image at the head of the queue is popped. The value of n is determined by the speed of the moving interference targets to be removed in the scene. To match the method with cameras of different frame rates, the value of n is correlated with the frame rate; that is, n is appropriately increased for high frame rate cameras and decreased for low frame rate cameras. The essence of controlling n is to control the interval between each frame in the background queue, ensuring that the moving interference targets in the image can move sufficiently to expose the background behind them, providing sufficient background information for subsequent removal operations.

[0059] Dynamic frame skipping speed also means making the value of n dynamic. During the method's operation, n is not unique, but can be changed according to the current state of the frame. For example, when only dynamic interference targets are detected in the frame, the background queue is updated normally every n frames. However, when a stationary target appears in the frame and needs to be quickly absorbed into the background model, n can be temporarily reduced, and frames can be skipped quickly so that the new stationary target appears in the entire background queue as soon as possible to ensure that it can be absorbed quickly.

[0060] The value of n is determined by the speed of the moving interference targets that need to be removed in the scene. To associate the value of n with the frame rate and ultimately control the interval between two frames, the mapping is performed according to the following formula:

[0061] n=F*t (1)

[0062] In the formula, F is the frame rate of the camera or video, t is the time difference between two frames in the background queue, and t is ultimately changed indirectly by changing n.

[0063] Figure 2 (a) Background queue collected by thread A; (b) Background upper bound map after filtering out interfering targets; (c) Background upper bound map after filtering out interfering targets.

[0064] Based on the current background queue state, thread B uses box plot statistics to remove moving interference targets and models the background after removing the interference targets. The process of obtaining the background model is as follows:

[0065] T21. Take the current background queue state and divide the background queue into blocks in the long and wide dimensions, hereinafter referred to as background queue sub-blocks;

[0066] Thread A copies the current state of the background queue it saves as a local variable for background modeling. At the same time, thread A continues to update the global background queue with new images from the camera so that subsequent modeling can use the latest background queue. Figure 1 The dashed line represents this form of parameter passing. The saved background queue is divided into a*a sub-blocks in both the length and width dimensions.

[0067] First, the saved background queue is divided into a*a sub-blocks in both length and width, where 'a' is the block granularity. Blocking is used to avoid eliminating an entire frame simply because a small target appears in the image. If a target moves too slowly, eliminating it without block division might result in the removal of all frames in the background queue. Therefore, the block granularity 'a' should be determined based on the size of the possible moving target appearing in the camera's view; the smaller and more frequently the moving target appears, the smaller the block granularity needs to be. The block granularity also determines the lower limit of the speed of moving interference targets that the algorithm can eliminate. However, the block granularity cannot be changed during operation; therefore, it is only used to derive the lower limit of the speed under the current parameters. Dynamic control of the lower limit of the speed at which targets can be eliminated still needs to be achieved by modifying 'n'.

[0068] T22. Use the box plot statistical method to count the pixel values ​​of the background queue sub-blocks one by one, find the frames in each block that capture moving interference targets, and remove the frames that capture moving interference targets from the background queue sub-blocks.

[0069] Calculate the lower quartile Q1, upper quartile Q3, and interquartile range iqr of the background queue sub-block in the queue dimension, and calculate the upper and lower whiskers according to the following formulas as the criteria for defining normal and outlier values:

[0070] lower_bound=Q1 - 1.5 * iqr (2)

[0071] upper_bound = Q3 + 1.5 * iqr (3)

[0072] Where: lower_bound is the lower whisker line and upper_bound is the upper whisker line;

[0073] Since each pixel in a color image contains three color channels (RGB), if any one of the three channels is determined to be abnormal, the pixel should be considered abnormal. Therefore, after determining the channel dimension, the determination result matrix is ​​ORed according to the color channel dimension to obtain the final determination result for each pixel.

[0074] If an anomaly is detected in the judgment result matrix, it can be assumed that some moving interference targets have been captured in this frame's image sub-block, and this frame's image sub-block is removed from the sub-block background queue. If all frames in a sub-block are removed, the first frame of the background queue is retained as the default background.

[0075] The background queue sub-blocks are sequentially subjected to box plot statistical methods to calculate the lower quartile Q1, upper quartile Q3, and interquartile range iqr of the sub-block in the queue dimension. The range between Q1 and Q3 is then expanded by 1.5 times iqr as the normal value range, and values ​​outside the range are considered outliers.

[0076] The lower quartile Q1, upper quartile Q3, and other statistics calculated by applying box plot statistical methods to the matrix are actually statistical arrays. These statistical arrays are used to determine whether each pixel value in each frame of the background queue is within the range of the upper and lower quartiles corresponding to its position, and finally a Boolean matrix is ​​used to mark the determination result of each pixel.

[0077] T23. Model the removed background queue sub-blocks using statistics, and then combine them into a background model of the original image size;

[0078] After removing the blocks, each sub-block contains a background queue of clean, uninterrupted targets. The maximum and minimum values ​​are used to perform statistical modeling, resulting in a model matrix containing the maximum and minimum values ​​of each pixel position.

[0079] The background model ultimately established by thread B has the following allowable speed lower limit for the moving interference target:

[0080]

[0081] In the formula: W is the distance the moving interference target travels within a sub-block in its direction of movement, which is generally considered to be the diagonal length of a sub-block, in pixels; t is the time difference between two frames in the background queue, in seconds; m is the total length of the background queue; Ω is the allowable anomaly rate of the box plot, which determines the maximum percentage of outliers that the box plot can tolerate when it can normally remove outliers from the queue. The value of Ω is related to the degree of anomaly between normal and outliers. The greater the degree of anomaly, the greater the allowable value of Ω, but generally the value of Ω is between 15% and 25%.

[0082] From the above formula (4), we can derive different allowable lower limits for the speed of interfering targets under different parameter values. By controlling the parameters, we can control the targets that can be removed. The allowable lower limit for the speed of targets with a speed greater than the threshold will be removed from the background. The unit of V is pixels per second. The actual distance needs to be re-mapped according to the position of the camera and lens distortion, etc.

[0083] The lower limit of the static duration for which the static target is absorbed into the background in the background model ultimately built by thread B is:

[0084] T>m*t*Ω (5)

[0085] Static targets with a static duration greater than this lower limit will be absorbed as background. Note that it takes time for one background modeling cycle from the start of absorption to its completion.

[0086] The process of modeling the removed background queue using statistical measures and then merging them into a background model of the original image size is as follows:

[0087] After removing the blocks, each sub-block contains a background queue of clean, uninterrupted targets. The maximum and minimum values ​​are used to perform statistical modeling, resulting in a model matrix containing the maximum and minimum values ​​of each pixel position.

[0088] Note that the statistical model established in this process is the model used in the final foreground detection stage. Although it is similar to the statistical model used in the box plot statistical method to remove outliers, the essential difference between the two is that the statistical vector to be statistically analyzed in the box plot modeling stage may contain both outliers and normal values. If Q1 and Q3 are used to directly model the background queue, the modeling granularity is pixel-level. However, when the moving interference target in the picture is relatively uniform in color and long in size, the area it passes through, excluding the head and tail ends of the target, may have a large number of middle parts in the background queue. This causes the box plot to consider this moving target as part of the background and mix it with the real background for modeling. Ultimately, the final model is neither the interference target nor the pure background.

[0089] This method only uses box plot statistics to identify and remove abnormal sub-blocks, without undertaking any statistical modeling. After removing frames considered to contain interfering targets from each sub-block using box plots, the background queue after removal is then statistically modeled using maximum and minimum values. At this point, the vector to be statistically analyzed contains only normal values, avoiding the problem of misclassifying slowly moving objects as background and confusing them with the real background. Furthermore, because the lengths of the background queues in each sub-block after background removal may not be the same, they cannot be directly stitched together after removal. Instead, they are modeled using statistical methods to define the upper and lower limits of pixel values ​​before stitching. Moreover, if all frames in a sub-block are removed, the first frame of the background queue is always retained as the default background.

[0090] Figure 3 (a) An image containing a moving target, (b) Foreground detection results for thread C. Thread C performs difference operations on each frame transmitted from the camera using the model built by the current thread B to obtain the foreground, as follows:

[0091] Thread C copies the current background model as a local variable, while thread B continues to update the global background model so that the model can be used in the new frame of the image to be inspected as soon as it is updated.

[0092] That is: to obtain the next frame of the image;

[0093] Based on the current background model, the background is updated, and the current background model is used to determine whether each point in the image is within the range of the background model. Pixels that are not within the background range are considered as foreground.

[0094] The background model includes upper and lower limits for the background value of each pixel position. Each pixel in the frame to be detected is compared with its corresponding upper and lower limits. If a pixel is outside the background value range, it is considered a foreground position. To avoid the influence of noise from the camera, the actual comparison follows this formula:

[0095]

[0096] In the formula: X is the image to be detected; b0 is the 0th dimension of the background model, i.e., the maximum value at each position in the background queue after removal; b1 is the 1st dimension of the background model, i.e., the minimum value at each position in the background queue after removal; λ is the denoising factor, which needs to be determined according to the noise fluctuation range captured by the video lens. Devices prone to noise need to have this value increased appropriately. Finally, a binary mask matrix with the same length and width as the original image is obtained. This matrix can be directly displayed as a binary image, or it can be ANDed with the original image to segment the foreground.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for dynamic scene background modeling and foreground detection based on box plots, characterized in that, Includes the following steps: The acquired video images are saved into the background queue according to the dynamic frame extraction speed; Based on the current background queue status, the box plot statistical method is used to remove moving interference targets, and the background after removing interference targets is modeled to obtain the background model. Each frame of the acquired video is subjected to a difference operation using a background model to obtain the foreground. The process of using box plot statistical methods to remove and model moving interference targets based on the current background queue state, and obtaining the background model, is as follows: T21. Take the current background queue state and divide the background queue into blocks in the long and wide dimensions, hereinafter referred to as background queue sub-blocks; T22. Use the box plot statistical method to count the pixel values ​​of the background queue sub-blocks one by one, find the frames in each block that capture moving interference targets, and remove the frames that capture moving interference targets from the background queue sub-blocks. T23. Model the removed background queue sub-blocks using statistics, and then combine them into a background model of the original image size; The process of sequentially using box plot statistics to count the pixel values ​​of the background queue sub-blocks, identifying the frames in each block that captured moving interference targets, and then removing those frames from the background queue sub-blocks is as follows: Calculate the lower quartile Q1, upper quartile Q3, and interquartile range iqr of the background queue sub-block in the queue dimension, and calculate the upper and lower whiskers according to the following formulas as the criteria for defining normal and outlier values: (2) (3) in: For the lower line and For the upper part; Values ​​outside the upper and lower whisker lines are considered outliers, meaning that the frame sub-block has captured a moving interference target. All frames containing outliers in the background queue sub-block are removed, and the background queue after removal consists entirely of clean background frames without foreground.

2. The method for dynamic scene background modeling and foreground detection based on box plots according to claim 1, characterized in that: The process of saving the acquired video images into the background queue according to the dynamic frame extraction rate is as follows: Frames are acquired from the camera. Every n frames, a frame is added to the background queue, and the oldest background frame at the head of the queue is popped. The value of n is determined by the speed of the moving interference targets that need to be removed in the scene. To make the method compatible with cameras with different frame rates, the value of n is related to the frame rate and determined by the following formula: (1) In the formula, F is the frame rate of the camera or video, t is the time difference between two frames in the background queue, and the value of n is ultimately determined by specifying t.

3. The method for dynamic scene background modeling and foreground detection based on box plots according to claim 1, characterized in that: The background queue is divided into blocks in both the long and wide dimensions, and the granularity 'a' of the blocks is determined based on the size of the moving target appearing in the camera frame.

4. The method for dynamic scene background modeling and foreground detection based on box plots according to claim 1, characterized in that: The process of obtaining the foreground by performing a difference operation on each frame of the acquired video using a background model is as follows: Get the next frame image; Based on the current background model, the background is updated, and the current background model is used to determine whether each point in the image is within the range of the background model. Pixels that are not within the background range are considered as foreground.

5. The method for dynamic scene background modeling and foreground detection based on box plots according to claim 1, characterized in that: The lower limit of the permissible speed for the moving interference target in the background model is: (4) In the formula: W is the distance the moving interference target moves within a sub-block in its direction of motion; t is the time difference between two frames in the background queue, in seconds; m is the total length of the background queue; Ω is the allowable anomaly rate of the box plot, which determines the maximum percentage of outliers that the box plot can tolerate when it can normally remove outliers from the queue.

6. The method for dynamic scene background modeling and foreground detection based on box plots according to claim 1, characterized in that: The lower limit of the static duration for which a stationary target is absorbed into the background in the background model is: (5) Static targets with a static duration longer than this static duration will be absorbed as background.

7. The method for dynamic scene background modeling and foreground detection based on box plots according to claim 1, characterized in that: The background model includes upper and lower limits for the background value of each pixel position. Each pixel in the frame to be detected is compared with its corresponding upper and lower limits. If a pixel is outside the background value range, it is considered a foreground position. The comparison is based on the following formula: (6) In the formula: X is the image to be detected; b0 is the 0th dimension of the background model, that is, the maximum value of each position in the background queue after removal; b1 is the 1st dimension of the background model, that is, the minimum value of each position in the background queue after removal; λ is the noise reduction factor, which needs to be determined according to the noise fluctuation range captured by the video lens. Devices prone to noise need to have this value increased appropriately.

Images