A method for monitoring personnel's safety production behavior

Abstract

The present invention relates to the field of image processing technology, and more specifically, to a method for monitoring personnel safety production behavior, the method comprising: acquiring a construction site image and a thermal imaging image; constructing a fixed area of ​​the pixel point with the pixel point in the construction site image as the center, obtaining an initial demand coefficient of the pixel point through the grayscale difference in the fixed area of ​​the pixel point, and the initial demand coefficient is negatively correlated with the grayscale difference in the fixed area; calculating the demand coefficient of the pixel point; adjusting a preset adjustment parameter through the demand coefficient of the pixel point to obtain a target adjustment parameter of the pixel point, and the target adjustment parameter is negatively correlated with the demand coefficient; and processing the construction site image using the target adjustment parameter of the pixel point in a dark channel defogging algorithm to realize personnel safety production behavior monitoring, which can effectively improve the accuracy of personnel safety production behavior monitoring.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and more specifically, to a method for monitoring personnel safety production behavior. Background Art

[0002] A construction site is a place where construction activities such as buildings, structures, roads, pipelines, etc. are carried out, usually involving multiple activities, equipment and personnel. In order to ensure construction safety, improve production efficiency and meet quality standards at the construction site, it is usually necessary to conduct safety monitoring of the production behavior of construction site personnel.

[0003] In order to obtain the results of personnel behavior monitoring at the construction site, the prior art provides a variety of monitoring methods. For example, the patent application document with publication number CN117612344A discloses an active safety system for equipment in a construction area and a personnel behavior monitoring method. The application obtains dynamic video images of a mobile device in the construction area and its surroundings as collected images; uses a preset image recognition model to detect the collected images to obtain the real-time position of the target; predicts the moving area of ​​the mobile device based on the current position of the mobile device; and judges the risk level of the personnel behavior based on the preset monitoring area of ​​the mobile device, combined with the real-time position and moving area of ​​the target.

[0004] The above-mentioned existing technology can obtain the risk level of personnel behavior by obtaining the real-time position and movement area of ​​the target. However, smoke, haze and other interferences to the video image may be generated during the construction process, resulting in smoke in the collected image, low clarity, and inability to accurately obtain the position and behavior of the on-site personnel, thereby affecting the accuracy of personnel behavior safety monitoring and posing safety hazards.

[0005] Based on this, how to remove smoke interference in monitoring images so as to accurately obtain personnel safety production behavior monitoring results is a problem that needs to be solved urgently by technical personnel in this field. Summary of the invention

[0006] In order to solve the above technical problem of how to remove smoke interference in monitoring images and thus accurately obtain personnel safety production behavior monitoring results, the present invention proposes a personnel safety production behavior monitoring method, which includes the following steps:

[0007] Acquire the construction site image and thermal imaging image; construct a fixed area of ​​the pixel point with the pixel point in the construction site image as the center, and obtain the initial demand coefficient of the pixel point through the grayscale difference in the fixed area of ​​the pixel point, and the initial demand coefficient is negatively correlated with the grayscale difference in the fixed area;

[0008] ;

[0009] , Respectively The demand coefficient of each pixel, the initial demand coefficient, For the The number of pixels with different grayscale values ​​in the construction site image and the previous frame of the construction site image in the fixed pixel area. is a hyperparameter, is a fixed region size, , Respectively The first pixel in the fixed area The value of the R channel and the sum of the RGB channels of each pixel in the thermal image. is an exponential function with e as the base; the preset adjustment parameters are adjusted by the demand coefficient of the pixel point to obtain the target adjustment parameter of the pixel point, and the target adjustment parameter is negatively correlated with the demand coefficient; the target adjustment parameter of the pixel point is used in the dark channel defogging algorithm to process the construction site image to realize the monitoring of personnel safety production behavior.

[0010] The present invention takes into account that rain, fog, flying dust, etc. in the construction site image may affect the accuracy of the monitoring of personnel's safe production behavior, and therefore pre-processes the construction site image through a dark channel defogging algorithm, thereby improving the accuracy of the monitoring results of personnel's safe production behavior. In this process, the present invention takes into account that the setting of the adjustment parameters in the dark channel defogging algorithm will affect the final defogging effect; based on this, the present invention dynamically calculates the initial demand coefficient by analyzing the local characteristics of the pixel points, and adjusts the defogging adjustment parameters of each pixel point, so as to achieve accurate defogging, thereby improving the accuracy of personnel behavior monitoring. On this basis, the present invention also takes into account that the changes in the human body area may be similar to the grayscale changes of the moving machinery, which affects the accuracy of the initial demand coefficient; based on this, the present invention accurately obtains the demand coefficient of the pixel point by combining the changes in the pixel points in the continuous image frames and the performance of the R channel in the thermal imaging image, while retaining the important details of the construction site image, strengthening the defogging effect, and effectively improving the accuracy of the monitoring of personnel's safe production behavior.

[0011] According to a method for monitoring personnel safety production behavior provided by the present invention, obtaining construction site images and thermal imaging maps includes: taking construction site photos and construction site thermal imaging photos by multi-spectral imaging equipment, and obtaining construction site images and thermal imaging maps of the construction site after preprocessing.

[0012] The present invention takes into account that the originally collected construction site photos and thermal imaging photos may not meet the size, edge distortion, etc. of the model training, so the collected photos are preprocessed to improve the image quality.

[0013] According to a method for monitoring personnel safety production behavior provided by the present invention, the fixed area of ​​the pixel point is constructed with the pixel point in the construction site image as the center, including: presetting the fixed area size ; Get the pixel as the center pixel points, and obtain the fixed area of ​​the pixel point.

[0014] According to a personnel production safety behavior monitoring method provided by the present invention, the initial demand coefficient of the pixel point is obtained through the grayscale difference in the fixed area of ​​the pixel point, including: recording the absolute value of the difference between the grayscale value of each pixel point in the fixed area of ​​the pixel point and the grayscale mean as the grayscale difference in the fixed area; using the negative of the product of the grayscale difference in the fixed area and the grayscale extreme difference as the exponent of an exponential function with e as the base, to obtain the initial demand coefficient of the pixel point.

[0015] The present invention takes into account that the grayscale extreme difference value in a fixed area can characterize its contrast, and therefore obtains the grayscale value fluctuation degree in the fixed area by analyzing the grayscale extreme difference value in the fixed area. In addition, the present invention also takes into account that the noise in the fixed area will affect the accuracy of the fluctuation degree, and therefore verifies the accuracy of the fluctuation degree by obtaining the grayscale difference in the fixed area, thereby accurately obtaining the initial demand coefficient of the pixel point.

[0016] According to a method for monitoring personnel safety production behavior provided by the present invention, the preset adjustment parameter is adjusted by the demand coefficient of the pixel point to obtain the target adjustment parameter of the pixel point, including: ; For the The target adjustment parameters of pixels, To preset adjustment parameters, For the The demand factor of each pixel.

[0017] According to a method for monitoring personnel safety production behavior provided by the present invention, the target adjustment parameters of pixels are used in a dark channel defogging algorithm to process a construction site image, including: converting the construction site image into a dark channel image, acquiring the global atmospheric light in the dark channel image; obtaining the transmittance of the pixel based on the construction site image, the target adjustment parameters of the pixel and the global atmospheric light; restoring the construction site image according to the transmittance and the atmospheric light to obtain a target construction site image.

[0018] The present invention calculates the target adjustment parameters of each pixel point, so that each pixel point in the construction site image can be accurately adjusted, the clarity of the construction site image can be improved while retaining important features, so as to accurately obtain the target construction site image.

[0019] According to a method for monitoring personnel safety production behavior provided by the present invention, converting a construction site image into a dark channel image includes: taking the minimum value of a pixel point in the RGB channel of the construction site image as the target grayscale value of the pixel point; adjusting the grayscale values ​​of all pixels in the construction site image to the target grayscale value to obtain a dark channel image.

[0020] According to a personnel safety production behavior monitoring method provided by the present invention, the maximum target grayscale value in a dark channel image is used as the global atmospheric light in the dark channel image.

[0021] According to a method for monitoring personnel safety production behavior provided by the present invention, the target adjustment parameters of pixel points are used in the dark channel defogging algorithm to process the construction site image to realize personnel safety production behavior monitoring, including: inputting the target construction site image into a pre-trained safety recognition model to obtain the safety production behavior monitoring result.

[0022] According to a personnel safety production behavior monitoring method provided by the present invention, the implementation of personnel safety production behavior monitoring further includes: in response to the safety production behavior monitoring result being abnormal, issuing an abnormal alarm.

[0023] The present invention takes into account that when the monitoring result of personnel's production safety behavior is abnormal, there may be dangerous behavior, so the abnormal and dangerous behavior of personnel is warned in time, thereby regulating the behavior of the staff and improving the safety of the operation.

[0024] The present invention has the following beneficial effects:

[0025] Based on the above technical scheme, the present invention provides a method for monitoring personnel safety production behavior. When monitoring personnel safety behavior, the dark channel defogging algorithm is used to pre-process the construction site image, which can improve the accuracy of the monitoring results of personnel safety production behavior. In this process, the present invention takes into account that the setting of the adjustment parameters will affect the defogging effect; based on this, the present invention dynamically calculates the initial demand coefficient by analyzing the local characteristics of the pixel points, and adjusts the defogging adjustment parameters of each pixel point, so as to achieve accurate defogging, thereby improving the accuracy of personnel behavior monitoring. On this basis, the present invention also takes into account that the changes in the human body area may be similar to the grayscale changes of the moving machinery, which affects the accuracy of the initial demand coefficient; based on this, the present invention combines the changes in the pixel points in the continuous image frames and the performance of the R channel in the thermal imaging image to accurately obtain the demand coefficient of the pixel points, so as to accurately achieve the defogging of the pixel points in the construction site image, and effectively improve the accuracy of personnel safety production behavior monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A flowchart of a method for monitoring personnel safety production behavior provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.

[0028] In order to monitor the safety production behavior of personnel at the construction site, an embodiment of the present invention provides a method for monitoring the safety production behavior of personnel. The method performs abnormal behavior monitoring after enhancing the construction site image through a dark channel defogging algorithm, which can effectively improve the accuracy of the abnormal behavior monitoring results.

[0029] For details, please see Figure 1 As shown, Figure 1 A flow chart of a method for monitoring personnel safety production behavior provided by an embodiment of the present invention, the method comprising the following steps:

[0030] S1: Obtain pixel points in the construction site image.

[0031] For example, in an embodiment of the present invention, obtaining a construction site image includes: taking a photo of the construction site by a multi-spectral imaging device, and obtaining a construction site image of the construction site after pre-processing.

[0032] The preprocessing may include scaling, cropping, edge enhancement, grayscale processing, etc., which may be specifically configured according to actual needs, and the embodiment of the present invention does not impose too many restrictions thereon.

[0033] Specifically, the multispectral imaging device collects the construction site video stream through the camera and uploads it to the terminal. The terminal monitors the safety production behavior of personnel in each frame of the construction site image in the construction site video stream.

[0034] It should be noted that when workers are working at the construction site, rain, fog, flying dust, etc. may occur, causing interference in the collected construction site images, reducing the clarity of the construction site images, and thus reducing the accuracy of the monitoring results of personnel safety production behavior.

[0035] Based on this, the embodiment of the present invention enhances the construction site image through a dark channel defogging algorithm to reduce the impact of rain, fog, flying dust, etc. on the safety behavior monitoring results of construction site personnel, that is, executes the following steps.

[0036] S2: A fixed area of ​​the pixel in the construction site image is constructed with the pixel as the center, and the initial demand coefficient of the pixel is obtained through the grayscale difference in the fixed area of ​​the pixel.

[0037] Among them, the initial demand coefficient is negatively correlated with the grayscale difference in a fixed area.

[0038] It should be noted that when the dark channel defogging algorithm processes the construction site image, it will adjust the intensity of the defogging effect of the construction site image by setting a fixed adjustment parameter, thereby affecting the quality of the final defogging image. If the adjustment parameter is too large, the details in the construction site image can be maintained, but the intensity of the image defogging effect may be reduced; if the adjustment parameter is too small, the defogging effect can be enhanced, but the image details may be lost, which will ultimately affect the monitoring accuracy of the construction site image.

[0039] Based on this, the embodiment of the present invention analyzes the grayscale changes in the area around the pixel point to obtain the degree of demand for the defogging effect of the pixel point, and dynamically adjusts the defogging adjustment parameters of each pixel point according to the local characteristics of the image. It can enhance the defogging effect while retaining important details of the construction site image, thereby improving the accuracy of personnel behavior monitoring.

[0040] For example, in an embodiment of the present invention, a fixed area of ​​a pixel point in a construction site image is constructed with the pixel point as the center, including: presetting the fixed area size ; Get the pixel as the center pixel points, and obtain the fixed area of ​​the pixel point.

[0041] Among them, the fixed area size of the pixel point can be set to 49, the fixed area size is the number of pixels contained in the fixed area, and the final fixed area includes the current pixel point itself; the fixed area size of the pixel point can be set specifically according to actual needs, and the embodiment of the present invention does not impose too many restrictions here.

[0042] It is understandable that for some pixels at the edge position, when constructing a fixed area, the number of pixels in the surrounding area may be smaller than the size of the fixed area. For such pixels, the fixed range of the pixel can be supplemented by methods such as mirror filling. The specific setting can be made according to actual needs, and the embodiments of the present invention do not impose too many restrictions on this.

[0043] After the surrounding area of ​​the pixel is obtained based on the above steps, the demand intensity of the pixel for the defogging effect can be obtained based on the difference between the pixel and other pixels in the surrounding area.

[0044] By way of example, in an embodiment of the present invention, the initial demand coefficient of the pixel point is obtained by the grayscale difference in a fixed area of ​​the pixel point, including: recording the cumulative sum of the absolute values ​​of the differences between the grayscale values ​​of each pixel point in the fixed area of ​​the pixel point and the grayscale mean as the grayscale difference in the fixed area; using the negative of the product of the grayscale difference in the fixed area and the grayscale extreme value as the exponent of an exponential function with base e to obtain the initial demand coefficient of the pixel point.

[0045] For example, the initial demand coefficient of the pixel point is calculated, and the specific formula is as follows:

[0046] ;

[0047] For the The initial demand coefficient of pixels, For the The grayscale extreme value in a fixed area of ​​pixels, For the The absolute value of the difference between the gray value of each pixel and the gray mean in the fixed area of ​​pixels is accumulated. It is an exponential function with e as base, where e is a natural constant.

[0048] In the above formula, The grayscale extreme value in a fixed area of ​​pixels indicates the fluctuation range of the grayscale value in the fixed area. The smaller the grayscale extreme value in the fixed area, the more concentrated the grayscale values ​​of the pixels in the fixed area are, and the smaller the fluctuation range is. Therefore, the brightness difference in the fixed area is not obvious, and the contrast is low. At this time, the possibility of flying dust, rain and fog in the fixed area is higher, and the corresponding pixel defogging demand is higher, and the initial demand coefficient is larger.

[0049] It is understandable that there may be noise interference in the construction site image. In order to improve the accuracy of calculating the pixel adjustment coefficient, it is also necessary to eliminate the interference of noise data. Indicates The grayscale difference in the fixed area of ​​pixels is smaller, the smaller the value is, the The smaller the grayscale extreme difference value in the fixed area of ​​each pixel, the higher the credibility, the lower the possibility of noise interference, and the larger the initial demand coefficient of the corresponding pixel.

[0050] It should be noted that, based on the above steps, by analyzing the difference between the grayscale value of the pixel point in the current construction site image and the grayscale value of its surrounding area, the initial requirement coefficient of each pixel point for defogging can be accurately obtained.

[0051] However, the objects of abnormal behavior identification in the construction site image are workers. The bodies of the workers will undergo dynamic changes during the construction process, and the contrast of the human body area is also low. Correcting the preset adjustment parameters directly based on the initial demand coefficient obtained in the above steps will also enhance the defogging effect of the human body area, thereby reducing the detail of the human body area and affecting the accuracy of behavior monitoring.

[0052] Based on this, the embodiment of the present invention can correct the initial demand coefficient of each pixel by analyzing the image features of each pixel in continuous image frames, so as to accurately obtain the adjustment parameters of the pixel, that is, continue to execute the following steps.

[0053] S3: Obtain the thermal image of the construction site and calculate the demand coefficient of the pixel points.

[0054] It should be noted that by analyzing the dynamic image features of each pixel in continuous image frames and correcting the initial demand coefficient of the pixel, the possibility of each pixel being a human body area can be obtained. The higher the possibility, the higher the importance of the pixel. In order to retain the image details, the adjustment coefficient of the pixel needs to be reduced.

[0055] However, in construction site images, pixels in some non-human areas may also change dynamically, such as mechanical equipment that produces displacement during the construction process. If the initial demand coefficient of the pixel is corrected directly based on the dynamic image characteristics of the pixel in continuous image frames, the adjustment coefficient of the pixel in such areas may also be reduced, thereby affecting the defogging effect of such areas and reducing the accuracy of safety monitoring.

[0056] Based on this, the embodiment of the present invention can accurately obtain the pixel points of the human body area in the construction site image by combining the dynamic image features of the pixel points in continuous image frames and the thermal imaging image of the construction site, thereby improving the accuracy of the pixel adjustment coefficient.

[0057] For example, in an embodiment of the present invention, obtaining a thermal imaging image includes: taking a thermal imaging photo of the construction site by a multi-spectral imaging device, and obtaining a thermal imaging image of the construction site after preprocessing.

[0058] The preprocessing of the thermal imaging photo can refer to the preprocessing method of the construction site image mentioned above, and the embodiment of the present invention will not be described in detail here. The pixels in the construction site image and the thermal imaging image correspond one to one, and the pixels at the same position in the construction site image and the thermal imaging image are two representations of the same pixel.

[0059] It is understandable that the RGB image and thermal imaging photos of the construction site can be collected simultaneously through the multispectral imaging equipment. The pixel points of the human body area in the thermal imaging photo have a larger value in the R channel. Therefore, if the pixel point in the fixed area of ​​the current pixel point has a larger value in the R channel of the thermal imaging image, the possibility that the current pixel point is a human body area is higher, and the corresponding importance is greater.

[0060] For example, in the embodiment of the present invention, the demand coefficient of the pixel point is calculated, and the specific formula can be referred to as follows:

[0061] ;

[0062] For the The demand coefficient of pixels, For the The initial demand coefficient of pixels, For the The number of pixels with different grayscale values ​​in the construction site image and the previous frame of the construction site image in the fixed pixel area. is a hyperparameter, is a fixed region size, For the The first pixel in the fixed area The value of the R channel of the pixel in the thermal image, For the The first pixel in the fixed area The sum of the RGB channels of the pixels in the thermal image. is an exponential function with base e.

[0063] The value of the hyperparameter may be set to 0.1, and may be set specifically according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this.

[0064] In the above formula, The pixels in the fixed area of ​​pixels with different grayscale values ​​in the construction site image and the previous frame are the pixels that have changed between the current construction site image and the previous frame of the construction site image. The larger the The more pixels that change in the fixed pixel area, the The greater the possibility that a fixed area of ​​pixels belongs to the human body area, the higher the corresponding importance. In the dehazing process, the required coefficient needs to be reduced to retain the image details.

[0065] Indicates The first pixel in the fixed area The ratio of the value of the R channel to the sum of the RGB channels in the thermal image. The larger the value, the The first pixel in the fixed area The larger the value of a pixel in the R channel, the greater the possibility that it is a human body area, the higher the credibility, the higher the corresponding importance, and the smaller the demand coefficient.

[0066] After accurately obtaining the demand coefficient of each pixel point based on the above steps, the preset adjustment parameters can be adjusted based on the demand coefficient, so as to accurately achieve defogging of the construction site image, that is, continue to perform the following steps.

[0067] S4: adjusting the preset adjustment parameters by the demand coefficient of the pixel point to obtain the target adjustment parameters of the pixel point, and processing the construction site image by using the target adjustment parameters of the pixel point in the dark channel defogging algorithm to obtain the target construction site image.

[0068] Among them, the target adjustment parameter is negatively correlated with the demand coefficient.

[0069] Among them, the preset adjustment parameter can be set to 0.99, and can be set specifically according to actual needs. The embodiment of the present invention does not impose too many restrictions on this.

[0070] It should be noted that the larger the pixel demand coefficient is, the higher the pixel demand for the dehazing effect is. In order to enhance the dehazing effect, the value of the preset adjustment parameter needs to be lowered. The smaller the pixel demand coefficient is, the lower the pixel demand for the dehazing effect is, and the higher the need for detail retention is. In order to retain image details, the value of the preset adjustment parameter needs to be increased.

[0071] For example, in the embodiment of the present invention, the preset adjustment parameter is adjusted by the demand coefficient of the pixel point to obtain the target adjustment parameter of the pixel point, which can be specifically referred to in the following relational expression:

[0072] ;

[0073] For the The target adjustment parameters of pixels, To preset adjustment parameters, For the The demand coefficient of pixels, is an exponential function with base e.

[0074] After obtaining the target adjustment parameters of each pixel point based on the above steps, defogging can be performed based on the target adjustment parameters of each pixel point, thereby improving the clarity of the construction site image and preparing for the safety production behavior monitoring of the staff.

[0075] By way of example, in an embodiment of the present invention, the target adjustment parameters of the pixels are used in the dark channel dehazing algorithm to process the construction site image, including: converting the construction site image into a dark channel image, acquiring the global atmospheric light in the dark channel image; obtaining the transmittance of the pixels based on the construction site image, the target adjustment parameters of the pixels, and the global atmospheric light; restoring the construction site image based on the transmittance and the atmospheric light to obtain the target construction site image.

[0076] Among them, the specific steps of obtaining the transmittance of the pixel points based on the construction site image, the target adjustment parameters of the pixel points and the global atmospheric light can be obtained through the existing transmittance formula, and the embodiments of the present invention will not be described here; the specific steps of restoring the construction site image according to the transmittance and atmospheric light to obtain the target construction site image can be achieved by the existing technology, and the embodiments of the present invention will not be described here.

[0077] By way of example, in an embodiment of the present invention, a construction site image is converted into a dark channel image, including: taking the minimum value of a pixel point in the RGB channel of the construction site image as the target grayscale value of the pixel point; adjusting the grayscale values ​​of all pixels in the construction site image to the target grayscale value to obtain a dark channel image.

[0078] For example, in an embodiment of the present invention, the maximum target grayscale value in the dark channel image is used as the global atmospheric light in the dark channel image.

[0079] After dehazing the construction site image based on the above steps, the clarity of the construction site image can be effectively improved, so that the safety production behavior monitoring results can be accurately obtained based on this in the subsequent steps, that is, the following steps are executed.

[0080] S5: Based on the target construction site image, personnel safety production behavior monitoring is realized.

[0081] For example, in an embodiment of the present invention, the target adjustment parameters of the pixels are used in the dark channel dehazing algorithm to process the construction site image to realize the safety production behavior monitoring of personnel, including: inputting the target construction site image into a pre-trained safety recognition model to obtain the safety production behavior monitoring results.

[0082] For example, when pre-training a safety recognition model, a set of dehazed construction site images can be collected in advance, wherein the construction site image set contains standard operations and dangerous behaviors of workers; the standard operations and dangerous behaviors are used as labels, and labels are manually set for the images in the construction site image set to obtain a labeled construction site image set; the labeled construction site image set is divided into a training set and a verification set, the neural network model is trained with the training set, and the performance of the neural network model is verified with the verification set to obtain a safety recognition model.

[0083] The currently collected dehazed target construction site image is input into the safety recognition model to obtain the corresponding safety production behavior monitoring results.

[0084] Among them, the neural network model can be a residual network (Residual Network, referred to as ResNet) model, a visual geometry group (Visual Geometry Group, referred to as VGG) model, etc., which can be set according to actual needs, and the embodiment of the present invention does not impose too many restrictions here.

[0085] It is understandable that the final safety production behavior monitoring type includes standard operation and dangerous behavior. If the safety production behavior monitoring type is standard operation, it means that the safety production behavior monitoring result is normal; if the safety production behavior monitoring type is dangerous behavior, it means that the safety production behavior monitoring result is abnormal. For dangerous behavior, timely reminders are required to reduce operational risks.

[0086] For example, in an embodiment of the present invention, personnel safety production behavior monitoring is implemented, and then the method further includes: in response to the safety production behavior monitoring result being abnormal, issuing an abnormal alarm.

[0087] The abnormal alarm may be in the form of a sound alarm, which may be set according to actual needs, and the embodiment of the present invention does not impose too many limitations on this.

[0088] It can be seen that in the embodiment of the present invention, when monitoring the safety production behavior of personnel, the construction site image and the thermal image can be obtained; a fixed area of ​​the pixel point in the construction site image is constructed with the pixel point as the center, and the initial demand coefficient of the pixel point is obtained by the grayscale difference in the fixed area of ​​the pixel point, and the initial demand coefficient is negatively correlated with the grayscale difference in the fixed area;

[0089] ;

[0090] , Respectively The demand coefficient of each pixel, the initial demand coefficient, For the The number of pixels with different grayscale values ​​in the construction site image and the previous frame of the construction site image in the fixed pixel area. is a hyperparameter, is a fixed region size, , Respectively The first pixel in the fixed area The value of the R channel and the sum of the RGB channels of each pixel in the thermal image. is an exponential function with e as the base; the preset adjustment parameter is adjusted by the demand coefficient of the pixel point to obtain the target adjustment parameter of the pixel point, and the target adjustment parameter is negatively correlated with the demand coefficient; the target adjustment parameter of the pixel point is used in the dark channel defogging algorithm to process the construction site image to realize the safety production behavior monitoring of the personnel. In this way, the embodiment of the present invention can not only improve the clarity of the defogged image, but also retain the details of the human body area in the image, thereby effectively improving the accuracy of the safety production behavior monitoring of the personnel by adaptively adjusting the target adjustment parameter of the pixel point when defogging the construction site image.

[0091] 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 protection scope of the present invention.

Claims

1. A method for monitoring personnel safety production behavior, characterized in that: include: Acquire the construction site image and thermal imaging image; construct a fixed area of ​​the pixel point with the pixel point in the construction site image as the center, and obtain the initial demand coefficient of the pixel point through the grayscale difference in the fixed area of ​​the pixel point, and the initial demand coefficient is negatively correlated with the grayscale difference in the fixed area; ； , Respectively The demand coefficient of each pixel, the initial demand coefficient, For the The number of pixels with different grayscale values ​​in the construction site image and the previous frame of the construction site image in the fixed pixel area. is a hyperparameter, is a fixed region size, , Respectively The first pixel in the fixed area The value of the R channel and the sum of the RGB channels of each pixel in the thermal image. is an exponential function with base e; The preset adjustment parameter is adjusted by the demand coefficient of the pixel point to obtain the target adjustment parameter of the pixel point, and the target adjustment parameter is negatively correlated with the demand coefficient; In the dark channel dehazing algorithm, the target adjustment parameters of the pixels are used to process the construction site images to realize the monitoring of personnel safety production behavior.

2. A method for monitoring personnel safety production behavior according to claim 1, characterized in that: The obtaining of the construction site image and thermal image includes: The construction site photos and construction site thermal imaging photos are taken by multi-spectral imaging equipment, and the construction site images and thermal imaging pictures of the construction site are obtained after pre-processing.

3. A method for monitoring personnel safety production behavior according to claim 1, characterized in that: The method of constructing a fixed area of ​​a pixel point in the construction site image as the center includes: Preset fixed area size ; Get the pixel as the center pixel points, and obtain the fixed area of ​​the pixel point.

4. A method for monitoring personnel safety production behavior according to claim 1, characterized in that: The initial demand coefficient of the pixel point is obtained by using the grayscale difference in the fixed pixel point area, including: The absolute cumulative sum of the differences between the grayscale values ​​of each pixel in the fixed area and the grayscale mean is recorded as the grayscale difference in the fixed area; the negative of the product of the grayscale difference in the fixed area and the grayscale extreme value is used as the exponent of the exponential function with e as the base to obtain the initial demand coefficient of the pixel.

5. A method for monitoring personnel safety production behavior according to claim 1, characterized in that: The step of adjusting the preset adjustment parameter by the demand coefficient of the pixel point to obtain the target adjustment parameter of the pixel point includes: ； For the The target adjustment parameters of pixels, To preset adjustment parameters, For the The demand factor of each pixel.

6. A method for monitoring personnel safety production behavior according to claim 1, characterized in that: The method of using the target adjustment parameters of the pixel points in the dark channel defogging algorithm to process the construction site image includes: Convert the construction site image into a dark channel image and obtain the global atmospheric light in the dark channel image; The transmittance of the pixel is obtained based on the construction site image, the target adjustment parameters of the pixel and the global atmospheric light; The construction site image is restored according to the transmittance and atmospheric light to obtain the target construction site image.

7. A method for monitoring personnel safety production behavior according to claim 6, characterized in that: The step of converting the construction site image into a dark channel image comprises: The minimum value of the pixel in the RGB channel of the construction site image is taken as the target grayscale value of the pixel; the grayscale values ​​of all pixels in the construction site image are adjusted to the target grayscale value to obtain a dark channel image.

8. A method for monitoring personnel safety production behavior according to claim 7, characterized in that: The maximum target gray value in the dark channel image is taken as the global atmospheric light in the dark channel image.

9. A method for monitoring personnel safety production behavior according to claim 6, characterized in that: The dark channel defogging algorithm uses the target adjustment parameters of the pixels to process the construction site image to achieve personnel safety production behavior monitoring, including: The target construction site image is input into the pre-trained safety recognition model to obtain the safety production behavior monitoring results.

10. A method for monitoring personnel safety production behavior according to claim 9, characterized in that: The implementation of personnel safety production behavior monitoring further includes: In response to abnormal production safety behavior monitoring results, an abnormal alarm is issued.

Images