A device and method for automatically cleaning a monitoring enclosure within a crude oil storage tank

Abstract

The application discloses a device and method for automatically removing dirt from an inner monitoring shell of a crude oil storage tank. The method comprises the following steps: obtaining an infrared image of the inner monitoring shell of the crude oil storage tank; performing image enhancement on the infrared image; identifying oil stains and calculating the area of the oil stains on the enhanced infrared image; and if the area of the oil stains is greater than a preset threshold, sending a cleaning instruction for starting a cleaning machine at a corresponding cleaning orientation. The application can accurately determine the position of oil stains inside the crude oil storage tank, automatically clean the oil stains on the monitoring shell, reduce repeated cleaning, save cleaning resources, and improve cleaning efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of oil tank cleaning and visual monitoring, and particularly relates to a device and method for automatically removing dirt from a monitoring shell in a crude oil storage tank. BACKGROUND

[0002] During long-term storage of crude oil in an oil storage tank, various chemicals inside the tank will decompose and recombine into a new substance, forming sludge and depositing on the tank bottom and tank wall. Therefore, cleaning and maintenance of the crude oil storage tank has become a focus of the country and the industry. Currently, a mechanical tank cleaning method is commonly used to clean oil stains. The basic principle is to dissolve and recover oil stains by spraying a certain temperature and pressure of the same type of crude oil or water through a cleaning machine nozzle after positioning the oil stains in the tank. In this process, the monitoring equipment is long-term located inside the oil tank, resulting in oil stains deposited on the monitoring shell.

[0003] Traditional oil tank cleaning uses a monitoring system to assist in cleaning, often neglecting to clean the monitoring shell, resulting in misjudgment of the oil stain position and reducing the oil tank cleaning efficiency. In addition, the oil tank is a low-illumination environment, and the image brightness collected by the infrared camera is insufficient and the contrast is not high, resulting in an inability to accurately identify and judge the oil stain position. SUMMARY

[0004] The present application provides a device and method for automatically removing dirt from a monitoring shell in a crude oil storage tank, which can enhance the collection of infrared images of the monitoring shell in the crude oil storage tank, more easily identify the oil stain position, solve the problem of reduced oil tank cleaning efficiency caused by the monitoring shell being contaminated by oil stains, and achieve the purpose of automatically removing dirt from the monitoring shell in the crude oil storage tank.

[0005] To solve the above technical problems, the present application provides a method for automatically removing dirt from a monitoring shell in a crude oil storage tank, comprising the following steps: obtaining an infrared image of a monitoring device in the crude oil storage tank using an infrared camera; image enhancement is performed on the infrared image; oil stain identification and oil stain area calculation are performed on the enhanced infrared image; if the calculated oil stain area is greater than a preset threshold, a cleaning instruction for starting a cleaning machine corresponding to the cleaning orientation is sent.

[0006] Preferably, the infrared image is image enhanced by first performing global gray scale adjustment to improve image brightness, then decomposing the gray scale adjusted image through wavelet transform to enhance image details, and then performing linear weighted fusion on the brightness improved image and the detail enhanced image to enhance the infrared image.

[0007] Preferably, the enhanced infrared image is subjected to oil stain identification and oil stain area calculation, the enhanced infrared image is first segmented into N*N parts using a dynamic threshold algorithm, local threshold values of the parts are then calculated, a stain seed is selected in the image, and the oil stain area is expanded and judged using a region growing algorithm, so as to finally determine the oil stain position and the oil stain area size.

[0008] Preferably, after sending a cleaning instruction for starting a cleaning machine corresponding to the cleaning orientation, the oil stain area is recalculated after waiting for the cleaning machine to finish cleaning or reaching a preset cleaning time, and it is judged whether the oil stain area is smaller than the preset threshold value.

[0009] Preferably, the cleaning orientation is obtained by rotating the infrared camera.

[0010] The application also provides a device for automatically removing oil stains from a crude oil storage tank monitoring shell, comprising: an infrared image acquisition module for acquiring an infrared image of the monitoring shell; an infrared image enhancement module for enhancing the infrared image; an oil stain identification and oil stain area calculation module for identifying oil stains and calculating the oil stain area of the enhanced infrared image; and an oil stain cleaning judgment module for automatically cleaning oil stains from the monitoring shell.

[0011] Preferably, the infrared image enhancement module performs global gray scale adjustment on the infrared image to improve the image brightness, decomposes the gray scale adjusted image through wavelet transform to enhance the image details, and performs linear weighted fusion on the brightness improved image and the detail enhanced image to enhance the infrared image of the monitoring shell.

[0012] Preferably, the oil stain identification and area calculation module further uses a dynamic threshold algorithm to segment the enhanced infrared image into N*N parts, calculates local threshold values of the parts, selects a stain seed in the image, expands and judges the oil stain area using a region growing algorithm, and finally determines the oil stain position and the oil stain area size.

[0013] Preferably, the oil stain cleaning judgment module judges whether the oil stain area is larger than a preset threshold value, and if yes, sends a cleaning instruction for starting a cleaning machine corresponding to the cleaning orientation.

[0014] Preferably, the oil stain cleaning judgment module is further used for waiting for the cleaning machine to finish cleaning, recalculating the oil stain area, and judging whether the oil stain area is smaller than the preset threshold value.

[0015] Preferably, the cleaning orientation is obtained by rotating the infrared camera.

[0016] The application enhances the image of the monitoring shell collected by the infrared camera, improves the image brightness and contrast, and makes it easier to identify the oil stain position.

[0017] In addition, the present application automatically identifies and cleans the oil stains on the monitoring device shell by comparing the size of the oil stains with a preset threshold, thereby improving the oil tank cleaning efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 Flow chart of the method for automatically removing the stains from the monitoring shell in the crude oil storage tank;

[0019] Figure 2 Flow chart of the infrared image enhancement;

[0020] Figure 3 Flow chart of the oil stain identification algorithm;

[0021] Figure 4 Flow chart of the automatic oil stain cleaning;

[0022] Figure 5 Block diagram of the system for automatically removing the stains from the monitoring shell in the crude oil storage tank. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0024] Embodiment 1

[0025] The present embodiment provides a method for automatically removing the stains from the monitoring shell in the crude oil storage tank, as shown in the following steps: Figure 1

[0026] Obtaining the infrared image of the monitoring shell in the crude oil tank;

[0027] Performing image enhancement on the infrared image;

[0028] Identifying the oil stains and calculating the oil stain area based on the enhanced infrared image;

[0029] If the oil stain area is greater than a preset threshold, a cleaning instruction for starting the cleaning machine at the corresponding cleaning orientation is sent.

[0030] In the present embodiment, the step of obtaining the infrared image of the monitoring shell in the crude oil storage tank is specifically as follows: an explosion-proof infrared camera is installed inside the oil tank, a data line is connected to a hard disk recorder in the monitoring room through an explosion-proof flexible tube, and a PC terminal is arranged in the monitoring room to control the rotation of the camera and to record and read the monitoring data in real time, so as to obtain the infrared image of the monitoring shell.

[0031] ​In this embodiment, as Figure 2 As shown, the specific steps for enhancing the infrared image are as follows: first, global grayscale adjustment is performed to increase the image brightness; then, the grayscale adjusted image is decomposed by wavelet transform to enhance image details; and finally, the brightness-enhanced image and the detail-enhanced image are linearly weighted and fused to enhance the infrared image.

[0032] Image enhancement of the infrared image mainly consists of three stages: global grayscale adjustment, wavelet decomposition to enhance high-frequency details and improve contrast, and linear weighted fusion of the brightness-enhanced image and the detail-enhanced image. First, a nonlinear mapping function is constructed by combining the properties of Gamma correction and linear grayscale transformation. Adaptive coefficients for mean and mean square error are introduced to adjust the transformation function. The nonlinear transformation function processes the low, medium, and high grayscale regions of the image separately, appropriately reducing the high grayscale levels and enhancing the low grayscale levels to obtain a brightness-enhanced image. Then, wavelet transform is used to decompose the brightness image, obtaining a low-frequency background contour map and a high-frequency edge detail map. Adaptive histogram equalization is applied to the low-frequency background contour map, and threshold transformation is applied to the high-frequency edge detail map to amplify the high-frequency wavelet coefficients and improve detail contrast. An inverse wavelet transform yields the detail-enhanced image. Finally, the detail-enhanced image and the brightness-enhanced image are linearly weighted and fused.

[0033] In this embodiment, the brightness in the infrared image is normalized. The brightness normalization formula is as follows:

[0034]

[0035] In the formula I n (x,y) represents the normalized value; I(x,y) represents the gray value of the pixel; min is the minimum value among the non-zero gray levels in the histogram; max is the maximum value among the non-zero gray levels in the histogram.

[0036] The above normalization operation extends the image's grayscale range to the entire grayscale space, but this can lead to excessive weakening or enhancement of some pixel grayscale values. Therefore, this paper proposes an adaptive nonlinear function based on Gamma correction, which can be expressed as:

[0037]

[0038] In the formula, I n The value represents the normalized value, λ is the stretching coefficient, which controls the degree of gray-level expansion in dark areas, k is the preservation coefficient, and z represents the adaptive coefficient, which is determined by the characteristics and properties of the image itself. The calculation formula is:

[0039]

[0040] wherein M i = mean - ε * var, M a = mean + ε * var, mean and var represent the mean value and the mean square deviation of the image respectively, and ε is a weight value of 0.8.

[0041] In this embodiment, as shown in Figure 3 the step of identifying the oil stains and calculating the oil stain area of the enhanced infrared image is specifically: the enhanced infrared image is segmented into N x N parts by using a dynamic threshold algorithm, the local threshold of the parts is calculated, a stain seed is selected in the image, and the oil stain area is expanded and judged by using a region growing algorithm; the position of the oil stain and the size of the oil stain area are determined.

[0042] As a preferred embodiment, the specific implementation steps of the oil stain identification and oil stain area calculation are as follows: the target of oil stain identification is to mark the oil stains from the image, in order to avoid large deviation of the result caused by a single threshold, the image is first segmented into N x N subgraphs, the local threshold of each subgraph is calculated, and the calculation formula is:

[0043]

[0044] wherein Q(x, y) is the local threshold, that is, the mean value of the field M x N.

[0045]

[0046] wherein L'(x, y) is the pixel value after threshold transformation, L(x, y) is the original pixel value, and D is the changed stain pixel value. Then a stain seed is selected, generally the darkest point in the image or the point located in the center of the point group, which can select D in formula (5). After the stain seed is determined, a non-seed 8-connected domain pixel discrimination method is used, that is, the stain seed is taken as the center to grow in the four directions and the diagonal direction, and the growth rule is: if there is no seed in the current pixel field, the seed number increases, and if there is a seed, two seeds are added to the stain set. Then the region growing algorithm is used for expansion judgment, and then the size of the oil stain area is calculated.

[0047] Since the oil stain on the shell of the monitoring device is closer to the camera, the oil stain area will be larger than the oil stain on the tank wall, so a threshold value is preset, if the oil stain area is greater than the preset threshold value, the monitoring shell oil stain is determined first, and the monitoring shell cleaning program is started; if the oil stain area is less than the preset threshold value, the oil stain on the tank is determined, and the tank wall cleaning program is started.

[0048] In the embodiment, after sending the cleaning instruction for starting the cleaning machine corresponding to the cleaning orientation, the oil stain area is recalculated after waiting for the cleaning machine to finish cleaning or reaching the preset cleaning time, and it is determined whether the oil stain area is less than the preset threshold. If the oil stain area becomes smaller, it proves that the cleaning shell program is correct. If the oil stain area does not change, the cleaning tank wall program is started. Similarly, after the tank wall is cleaned, the oil stain area is recalculated. If the oil stain area decreases, it proves that the cleaning tank wall is correct. If the oil stain area does not decrease, the cleaning shell program is started.

[0049] In the embodiment, the cleaning orientation is obtained through the rotation angle of the infrared camera. The cleaning work of the monitoring shell is completed by 6 cleaning machine nozzles installed around the shell. The 6 nozzles are 60 degrees apart, form a full range of encirclement to the monitoring shell, and the initial position of the camera is set to 0 degrees. According to the rotation angle of the infrared camera, the corresponding cleaning nozzle orientation is determined.

[0050] The infrared enhancement adopts three stages of image global gray scale adjustment, wavelet decomposition to enhance high frequency details to improve contrast, and linear weighted fusion of brightness enhancement image and detail strengthening to realize, which can enhance image brightness, improve image contrast, and more easily identify oil stain position.

[0051] The oil stain identification and oil stain calculation are realized by using dynamic threshold algorithm and region growing algorithm, which can quickly identify the oil stain position and calculate the oil stain area size.

[0052] The automatic cleaning program of the monitoring shell can automatically identify and clean the oil stain of the contaminated monitoring device shell by comparing the oil stain size with the preset threshold, thereby improving the oil tank cleaning efficiency.

[0053] Embodiment 2

[0054] The embodiment provides a device for automatically removing oil stains from a monitoring shell of a crude oil storage tank, as shown in Figure 5 The device comprises an infrared image acquisition module, an infrared image enhancement module, an oil stain identification and oil stain area calculation module, and an oil stain cleaning judgment module. The infrared image acquisition module acquires an infrared image of the monitoring shell. The infrared image enhancement module enhances the infrared image. The oil stain identification and oil stain area calculation module identifies oil stains and calculates the oil stain area from the enhanced infrared image. The oil stain cleaning judgment module sends a cleaning instruction for starting a cleaning machine corresponding to a cleaning orientation if the oil stain area is greater than a preset threshold.

[0055] In the embodiment, the image enhancement module also performs global gray scale adjustment on the infrared image to improve image brightness. The gray scale adjusted image is decomposed through wavelet transform to enhance image details. The brightness improved image and the detail enhanced image are linearly weighted and fused to enhance the infrared image of the monitoring shell.

[0056] The image enhancement of the infrared image mainly includes three stages: image global gray scale adjustment, wavelet decomposition to enhance high frequency details to improve contrast, and linear weighted fusion of image brightness enhancement and detail strengthening. First, a nonlinear mapping function is constructed by combining the properties of Gamma correction and linear gray scale transformation, and the adaptive coefficients of mean and mean square deviation are introduced to adjust the transformation function. The low gray level, medium gray level and high gray level regions of the image are processed respectively by the nonlinear transformation function, so that the high gray level part is appropriately reduced and the low gray level part is appropriately increased, thereby obtaining an image with improved brightness. Then, the wavelet transform is used to decompose the brightness image to obtain a low-frequency background contour map and a high-frequency edge detail map. The low-frequency background contour is processed by adaptive histogram equalization, and the high-frequency edge detail map is processed by threshold transformation to enlarge the high-frequency wavelet coefficients and improve the detail contrast. After inverse wavelet transform, a detail enhanced image is obtained. Finally, the detail enhanced image and the brightness enhanced image are linearly weighted and fused.

[0057] In this embodiment, the brightness in the infrared image is normalized, and the brightness normalization formula is:

[0058]

[0059] In the formula, I n (x, y) represents the normalized value; I(x, y) represents the gray value of the pixel point; min is the minimum value in the non-zero gray level in the histogram; and max is the maximum value in the non-zero gray level in the histogram.

[0060] After the above normalization operation, the gray level interval of the image is extended to the entire gray space, but it may cause excessive weakening and enhancement of part of the pixel gray values. Therefore, this paper proposes an adaptive nonlinear function by referring to Gamma correction, which can be represented as:

[0061]

[0062] In the formula, I n represents the normalized value, λ is the stretching coefficient, which controls the degree of expansion of the dark area gray level, k is the retention coefficient, and z represents the adaptive coefficient, which is determined by the characteristics and properties of the image itself. The calculation formula is:

[0063]

[0064] In the formula, Mi = mean - *var, Ma = mean + *var, mean and var represent the mean and mean square deviation respectively, and ε is the weight, which is 0.8.

[0065] In the embodiment, the oil stain identification and area calculation module further utilizes a dynamic threshold algorithm to divide the enhanced infrared image into N×N parts, calculates the local threshold of the parts, selects a stain seed in the image, utilizes a region growing algorithm to expand and judge the oil stain area, and determines the oil stain position and the oil stain area size.

[0066] As a preferred embodiment, as shown in Figure 3 The specific implementation steps of the oil stain identification and oil stain area calculation are as follows:

[0067] The target of the oil stain identification is to mark the oil stain from the image. In order to avoid the large deviation of the single threshold to the result, the image is first divided into N×N sub-images, and the local threshold of each sub-image is calculated. The calculation formula is as follows:

[0068]

[0069] In the formula, Q(x, y) is the local threshold, that is, the mean value of the field M×N.

[0070]

[0071] In the formula, L'(x, y) is the pixel value after the threshold transformation, L(x, y) is the original pixel value, and D is the changed stain pixel value.

[0072] Then, the stain seed is selected. Generally, the darkest point in the image or the point located in the center of the point group is selected. D in formula (5) can be selected. After the stain seed is determined, the non-seed 8-connected domain pixel discrimination method is adopted, that is, the stain seed is taken as the center to grow in the four directions and the diagonal direction. The growth rule is that if there is no seed in the current pixel field, the seed number is increased, and if there is a seed, two seeds are added to the stain set.

[0073] Then, the region growing algorithm is utilized to expand and judge, and the area size of the oil stain area is calculated.

[0074] In the embodiment, the specific implementation steps of the oil stain cleaning judgment module are as follows: first, it is judged whether the oil stain area is larger than a preset threshold. If it is larger than the threshold, the cleaning instruction of the cleaning machine of the corresponding cleaning direction is sent. Since the oil stain on the shell of the monitoring device is close to the camera, the oil stain area will be larger than the oil stain on the tank wall. Therefore, a threshold is set. If the oil stain area is larger than the preset threshold, the monitoring shell oil stain is preferentially judged, and the monitoring shell cleaning program is started. If the oil stain area is smaller than the preset threshold, the oil tank oil stain is determined, and the tank wall cleaning program is started. After the cleaning is completed, the oil stain area is recalculated, and it is judged whether the cleaning program is correct.

[0075] In the embodiment, the cleaning direction is obtained by the rotation angle of the infrared camera, and the cleaning work of the monitoring shell is completed by 6 cleaning machine nozzles installed around the shell, the 6 nozzles are separated by 60 degrees, and the monitoring shell is fully surrounded, the initial position of the camera is set to 0 degrees, and the corresponding cleaning nozzle direction is determined according to the rotation angle of the infrared camera.

[0076] The infrared image enhancement module adopts three stages of image global gray adjustment, wavelet decomposition to enhance high-frequency details to improve contrast, and linear weighted fusion of brightness enhancement image and detail strengthening to realize, which can enhance image brightness, improve image contrast, and more easily identify oil stain position.

[0077] The oil stain identification and oil stain calculation module adopts a dynamic threshold algorithm and a region growing algorithm to realize, which can quickly identify the oil stain position and calculate the oil stain area size.

[0078] The oil stain automatic cleaning module can automatically identify and clean the contaminated monitoring device shell oil stain by comparing the oil stain size with the preset threshold, thereby improving the oil tank cleaning efficiency.

[0079] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for automatic decontamination of the outer casing of a crude oil storage tank monitoring system, characterized in that, Includes the following steps: Acquire infrared images of the monitoring casing inside crude oil storage tanks; Image enhancement of the infrared image includes the following steps: constructing a nonlinear mapping function by combining the properties of Gamma correction and linear grayscale transformation; simultaneously introducing adaptive coefficients based on the image's mean and mean square error to adjust the nonlinear mapping function to obtain an adaptive nonlinear function; and processing the low grayscale, medium grayscale, and high grayscale regions of the image separately using the adaptive nonlinear function, so that the high grayscale regions are appropriately reduced and the low grayscale regions are appropriately increased, thereby obtaining an image with enhanced brightness; the adaptive nonlinear function is expressed as: ， In the formula, This represents the normalized grayscale value of a pixel in an infrared image. The stretching factor controls the degree of grayscale expansion in dark areas. To maintain the coefficient, The adaptive coefficient is determined by the features and properties of the image itself, and its calculation formula is: ， In the formula, M i =mean- *var, M a =mean+ *var, mean, and var represent the mean and standard deviation of the image, respectively. This is the weight, with a value of 0.8; Then, wavelet transform is used to decompose the brightness image to obtain a low-frequency background contour map and a high-frequency edge detail map. Adaptive histogram equalization is performed on the low-frequency background contour map, and threshold transformation is performed on the high-frequency edge detail map to amplify the high-frequency wavelet coefficients and improve detail contrast. After inverse wavelet transform, a detail-enhanced image is obtained. Finally, the detail-enhanced image and the brightness-enhanced image are linearly weighted and fused. The process of identifying oil stains and calculating the oil stain area in the enhanced infrared image includes the following steps: dividing the enhanced infrared image into N×N parts using a dynamic thresholding algorithm; calculating the local thresholds for the parts; selecting stain seeds in the oil stain image; expanding the oil stain area using a region growing algorithm; and determining the location and size of the oil stain. If the oil stain area is greater than a preset threshold, a cleaning command is sent to start the cleaning machine in the corresponding cleaning direction.

2. The method for automatic decontamination of the monitoring shell inside a crude oil storage tank as described in claim 1, characterized in that, After sending the cleaning command to start the cleaning machine in the corresponding cleaning location, the following steps are also included: waiting for the cleaning machine to finish cleaning or reach the preset cleaning time, recalculating the oil stain area, and determining whether the oil stain area is less than the preset threshold.

3. The method for automatic decontamination of the monitoring shell inside a crude oil storage tank as described in claim 1, characterized in that, The corresponding cleaning orientation is obtained by rotating an infrared camera.

4. A device for automatic decontamination of the outer casing of a crude oil storage tank, characterized in that, include: Infrared image acquisition module, acquires infrared images of the monitoring enclosure; The infrared image enhancement module is used to: construct a nonlinear mapping function by combining the properties of Gamma correction and linear grayscale transformation; simultaneously, introduce adaptive coefficients of mean and mean square error to adjust the nonlinear mapping function to obtain an adaptive nonlinear function; and process the low grayscale, medium grayscale, and high grayscale regions of the image separately through the adaptive nonlinear function, so that the high grayscale parts are appropriately reduced and the low grayscale parts are appropriately increased, thereby obtaining an image with enhanced brightness; the adaptive nonlinear function is expressed as: ， In the formula, This represents the normalized grayscale value of a pixel in an infrared image. The stretching factor controls the degree of grayscale expansion in dark areas. To maintain the coefficient, The adaptive coefficient is determined by the features and properties of the image itself, and its calculation formula is: ， In the formula, M i =mean- *var, M a =mean+ *var, mean, and var represent the mean and standard deviation of the image, respectively. This is the weight, with a value of 0.8; Then, wavelet transform is used to decompose the brightness image to obtain a low-frequency background contour map and a high-frequency edge detail map. Adaptive histogram equalization is performed on the low-frequency background contour map, and threshold transformation is performed on the high-frequency edge detail map to amplify the high-frequency wavelet coefficients and improve the detail contrast. After inverse wavelet transform, a detail-enhanced image is obtained. Finally, the detail-enhanced image and the brightness-enhanced image are linearly weighted and fused. The oil stain identification and area calculation module is used to: segment the enhanced monitoring shell infrared image into N×N parts using a dynamic threshold algorithm; calculate the local threshold of the parts; select stain seeds in the oil stain image; expand and judge the oil stain area using a region growing algorithm; and determine the location and size of the oil stain. The oil stain cleaning judgment module sends a cleaning command to start the cleaning machine in the corresponding cleaning location if the oil stain area is greater than a preset threshold.

5. The device for automatic decontamination of the monitoring shell inside a crude oil storage tank as described in claim 4, characterized in that, The oil stain cleaning judgment module is also used to: wait for the cleaning machine to finish cleaning or reach the preset cleaning time, recalculate the oil stain area, and determine whether the oil stain area is less than the preset threshold.

6. The device for automatic decontamination of the monitoring shell inside a crude oil storage tank as described in claim 4, characterized in that, The cleaning orientation is obtained by rotating an infrared camera.

Images