Method, System, Equipment and Storage Medium for Measuring Rust on Storage Tank Roof

Abstract

The present invention discloses a method, system, device and storage medium for measuring the rust of a storage tank roof. The method includes: obtaining inspection data of a tank farm through a drone; the inspection data includes inspection photos of the tank farm obtained by an image acquisition device provided on the drone; a remote processing center obtains the inspection data including the inspection pictures, and identifies the rust on the roof of each storage tank to be measured from the inspection pictures according to a preset prediction model. Determine the photo rust recognition result of the storage tank to be measured according to the tank roof mask label and the preliminary rust recognition result generated by the mask recognition sub-model; determine the rust area ratio of the storage tank to be measured according to the photo rust recognition result. The present invention eliminates the artificial subjective judgment link, so its measurement result is more accurate and stable.

Description

Technical Field

[0001] The present invention relates to the field of chemical equipment, and particularly to a method, system, equipment and storage medium for measuring the rust of a storage tank roof. Background Art

[0002] A storage tank is a large container used to store liquid media, which is commonly seen in refining scenarios such as refineries, oil fields, oil depots, etc. Tanks form a tank farm to store various media. Storage tanks are widely used for the storage of liquid media in oil fields, refining, stations, storage areas, etc. During operation, affected by the external environment, the corrosion of the tank roof is a common phenomenon. According to the storage tank maintenance and repair regulations, it is necessary to periodically measure the corrosion degree of the storage tank in order to make inspection, maintenance and repair decisions based on this.

[0003] With the intelligent transformation of refineries, drone patrol is widely used. Through a drone equipped with a high-definition camera and along a certain patrol route, pictures of the patrol object (storage tank) can be taken to obtain patrol photos including the storage tank.

[0004] In the prior art, after collecting patrol photos by drone patrol shooting, the determination of the rust situation on the tank roof generally needs to be carried out by manual visual comparison.

[0005] The inventor has found through research that the method for determining the rust situation on the tank roof in the prior art has at least the following defects:

[0006] Since the manual comparison method lacks objective quantitative indicators and there is no unified standard, when manually determining the rust situation on the tank roof in the photo, the stability and accuracy of the determination result are likely to be poor due to personal subjective factors. Summary of the Invention

[0007] The main purpose of the present invention is to improve the stability and accuracy in measuring the rust of a storage tank roof.

[0008] To achieve the above purpose, the technical solution adopted by the present invention is:

[0009] The present invention discloses a method for measuring the rust of a storage tank roof, including:

[0010] S11. Obtain the patrol data of the tank farm through a drone; the patrol data includes the patrol pictures of the tank farm obtained by an image acquisition device provided on the drone; the patrol pictures are orthoimages;

[0011] S12. A remote processing center obtains the patrol data including the patrol pictures, and identifies the rust marks on the tank roof of each storage tank to be measured according to the patrol pictures through a preset prediction model; the prediction model includes:

[0012] A rectangle recognition sub - model for generating a rectangle box label of the to - be - measured storage tank in the inspection picture; the rectangle recognition sub - model is generated by training with the original inspection picture combined with the rectangle box label; the rectangle box label of the tank top pattern is obtained by box - selecting the original inspection picture as historical data, and a tank top mask label and a rust label are generated; the original inspection picture is obtained by an image acquisition device installed on the drone;

[0013] A mask recognition sub - model for generating a tank top mask label of the to - be - measured storage tank in the tank top rectangle image; the mask recognition sub - model is generated by training with the tank top rectangle image combined with the tank top mask label; the tank top rectangle image is obtained according to the rectangle label;

[0014] A preliminary rust recognition sub - model for obtaining a preliminary rust recognition result according to the tank top rectangle image; the preliminary rust recognition sub - model is generated by training with the tank top rectangle image combined with the rust label;

[0015] S13. The remote processing center determines the photo rust recognition result of the to - be - measured storage tank according to the tank top mask label generated by the mask recognition sub - model and the preliminary rust recognition result;

[0016] S14. Determine the corrosion area ratio of the to - be - measured storage tank according to the photo rust recognition result.

[0017] Preferably, in the present invention, the inspection data further includes scanning data; the drone is further provided with a synthetic aperture radar for obtaining scanning data synchronized with the image acquisition device; the scanning data is synchronized and one - to - one corresponding to the inspection picture obtained by the image acquisition device; the remote processing center is further used for:

[0018] S15. Generate a SAR image consistent with the coordinate system of the inspection picture corresponding to the current scanning data according to the current scanning data;

[0019] S16. Determine a reflected rust area corresponding to the rust area in the photo rust recognition result in the SAR image;

[0020] S17. Determine the corrosion depth of the reflected rust area according to the gray value of the reflected rust area.

[0021] Preferably, in the present invention, the determining the corrosion depth of the reflected rust area according to the gray value of the reflected rust area includes:

[0022] Preset the corresponding relationship between the gray value of the SAR image and the corrosion depth;

[0023] Determine the rust depth of the reflected rust area according to the gray value and the corresponding relationship of the reflected rust area in the SAR image.

[0024] Preferably, in the present invention, it includes:

[0025] The rectangular frame label is obtained in the original inspection picture by manual framing.

[0026] Preferably, in the present invention, the generation of the rust label includes:

[0027] Select the rust seed area features by the clustering method;

[0028] Obtain the preliminary rust label by the seed area growth method;

[0029] Perform clustering again to update the seed area features;

[0030] Update the rust label again by the seed area growth method;

[0031] End when the clustering center no longer changes or the iteration termination signal is reached.

[0032] Preferably, in the present invention, the generation of the tank top mask label includes:

[0033] Select the tank top candidate area by edge detection;

[0034] Select the largest polygon object as the rough label;

[0035] Obtain the final tank top mask label after manual fine-tuning.

[0036] Preferably, in the present invention, the flight altitude of the drone is greater than 200 meters.

[0037] Preferably, in the present invention, the ratio of the image of the storage tank to be measured in the inspection picture is between 0.2 and 0.4.

[0038] Preferably, in the present invention, it further includes:

[0039] The inspection data further includes the flight parameters of the drone, and the flight parameters include the longitude, latitude, yaw angle, pitch angle, roll angle and flight altitude when taking the inspection picture;

[0040] Unify the coordinate systems of the inspection pictures according to the flight parameters corresponding to the inspection pictures by image rotation;

[0041] Combined with the actual positions of the storage tanks in the storage tank area, identify each storage tank in the inspection picture;

[0042] According to the identification results of the identifiers, each of the above-mentioned rectangular images of the tank top is identified respectively.

[0043] On another aspect of the present invention, a corrosion determination system for a storage tank top is further provided, which includes a drone and a remote processing center;

[0044] The drone is used to obtain inspection data of the tank area; the inspection data includes inspection pictures of the tank area obtained by an image acquisition device provided on the drone; the inspection pictures are ortho-images;

[0045] The remote processing center includes a prediction model, a preliminary rust recognition sub-model, a photo rust recognition unit, and an area ratio calculation unit;

[0046] The prediction model is used to identify the rust on the tank top of each storage tank to be measured according to the inspection pictures; the prediction model includes:

[0047] A rectangle recognition sub-model, which is used to generate rectangle box labels of the storage tank to be measured in the inspection pictures; the rectangle recognition sub-model is generated by training with the original inspection pictures combined with the rectangle box labels; the rectangle box labels of the tank top pattern are obtained by frame selection of the original inspection pictures as historical data, and the tank top mask labels and rust labels are generated; the original inspection pictures are obtained by an image acquisition device provided on the drone;

[0048] A mask recognition sub-model, which is used to generate the tank top mask labels of the storage tank to be measured in the rectangular image of the tank top according to the rectangular image of the tank top; the mask recognition sub-model is generated by training with the rectangular image of the tank top combined with the tank top mask labels; the rectangular image of the tank top is obtained according to the rectangular label;

[0049] The preliminary rust recognition sub-model is used to obtain the preliminary rust recognition result according to the rectangular image of the tank top; the preliminary rust recognition sub-model is generated by training with the rectangular image of the tank top combined with the rust labels;

[0050] The photo rust recognition unit is used to determine the photo rust recognition result of the storage tank to be measured according to the tank top mask labels generated by the mask recognition sub-model and the preliminary rust recognition result.

[0051] The area ratio calculation unit is used to determine the corrosion area ratio of the storage tank to be measured according to the photo rust recognition result.

[0052] Preferably, in the present invention, the drone is further provided with a synthetic aperture radar for obtaining scanning data synchronized with the image acquisition device; the scanning data is synchronized and in one-to-one correspondence with the inspection pictures obtained by the image acquisition device; the inspection data further includes the scanning data;

[0053] The remote processing center further includes:

[0054] A coordinate conversion unit, configured to generate a SAR image consistent with the inspection picture coordinate system corresponding to the current scan data according to the current scan data;

[0055] A reflection rust determination unit, configured to determine a reflection rust area corresponding to the rust area in the photo rust recognition result in the SAR image;

[0056] A rust depth determination unit, configured to determine the rust depth of the reflection rust area according to the gray value of the reflection rust area.

[0057] On the other hand of the embodiment of the present invention, there is also provided a storage tank roof rust determination device, including:

[0058] A memory, configured to store a computer program;

[0059] A processor, configured to call and execute the computer program to implement each step executed by the remote processing center in the storage tank roof rust determination method described in any one of the above.

[0060] On the other hand of the embodiment of the present invention, there is also provided a storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each step executed by the remote processing center in the storage tank roof rust determination method described in any one of the above is implemented.

[0061] Beneficial effects

[0062] In the present invention, it is necessary to jointly implement the storage tank roof rust determination through a drone and a remote processing center. Among them, the drone is used as a data acquisition device for inspection; the remote processing center is a data receiving and processing device, and among them, a processor, a storage medium, and a computer program are used to generate a determination result according to the inspection data; in the present invention, taking the original inspection picture as historical data as the modeling data, a rectangular recognition sub-model for generating a rectangular frame label of the storage tank to be determined according to the inspection picture, a mask recognition sub-model for generating a mask label of the storage tank roof of the storage tank to be determined in the tank roof rectangular image according to the tank roof rectangular image, and a preliminary rust recognition sub-model for obtaining a preliminary rust recognition result according to the tank roof rectangular image are constructed; in this way, according to the current inspection picture obtained by the inspection device (such as a drone equipped with an image acquisition device), the preliminary rust recognition result of the tank roof rectangular image can be obtained through the above-mentioned respective sub-models; then, the rust on the storage tank roof can be recognized according to the tank roof mask label.

[0063] Through the present invention, rust on the top of the storage tank can be accurately identified, so the rust area ratio of the top of the storage tank can be accurately calculated according to the identification result; thus, an accurate result of the rust determination of the top of the storage tank can be obtained based on the rust area ratio; since the present invention excludes the artificial subjective judgment link, the determination result is more accurate and stable.

[0064] Furthermore, in the present invention, it may further include scan data collected by a synthetic aperture radar synchronized with the image acquisition device, and a SAR image consistent with the coordinate system of the corresponding inspection picture is generated based on the scan data; since the scattering characteristics of the SAR image can reflect the roughness of the surface of the object being photographed, therefore, the rust area can be determined in the SAR image according to the identification result of the inspection picture; then, the corresponding rust depth is determined according to the different scattering characteristics in the rust area; since through the present invention, not only the rust area (or rust area ratio) of the top of the storage tank can be determined, but also different rust depths in the rust area can be identified, so a more accurate and comprehensive determination result of the rust degree and rust condition can be obtained.

[0065] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application and be able to implement it according to the content of the specification, and in order to make the above and other purposes, technical features and advantages of the present application more understandable, one or more preferred embodiments are listed below and described in detail with the accompanying drawings as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the present invention. For those skilled in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0067] Figure 1 It is a schematic diagram of the steps of the method for determining the rust of the storage tank described in the present invention;

[0068] Figure 2 It is a schematic diagram of the structure of the system for determining the rust of the storage tank described in the present invention;

[0069] Figure 3 It is another schematic diagram of the steps of the method for determining the rust of the storage tank described in the present invention;

[0070] Figure 4 It is a schematic diagram of the structure of the equipment for determining the rust of the storage tank described in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0071] To enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0072] Embodiment 1

[0073] To improve the stability and accuracy of the determination of the rust on the tank top of a storage tank during the determination of the rust on the tank top of the storage tank, referring to Figure 1 , the embodiment of the present invention provides a method for determining the rust on the tank top of a storage tank, including:

[0074] S11. Obtain the inspection data of the tank area through a drone; the inspection data includes the inspection pictures of the tank area obtained by an image acquisition device provided on the drone; the inspection pictures are orthophotos;

[0075] Referring to Figure 2 , to determine the rust on the tank top of a storage tank, in the embodiment of the present invention, a corresponding system for determining the rust on the tank top of a storage tank is constructed; specifically, the method for determining the rust on the tank top of a storage tank in the embodiment of the present invention requires a drone and a remote processing center to cooperate to achieve; wherein the drone is used to collect the inspection data of the tank area; the remote processing center is used to perform corresponding processing and calculations based on the inspection data to obtain the result of the determination of the rust on the tank top of the storage tank.

[0076] The inspection data in the embodiment of the present invention should at least include the inspection pictures of the tank area; the inspection pictures can be obtained through an image acquisition device (such as a camera or a photographing device) provided on the drone. Preferably, the drone provided with the image acquisition device needs to take pictures of each storage tank to be determined in the tank area at a distance of at least 200 meters from the ground. On the one hand, it can effectively avoid dangerous buildings, and on the other hand, it can take pictures with the best effect as much as possible, ensuring that the proportion of the complete tank top in the image in the image is between 0.2 and <0.4>. In addition, the weather for taking the inspection pictures is preferably under the condition of cloudy and no obvious light, and the image quality of the inspection pictures should at least ensure a resolution of 6000×4000 and a bit depth of 24.

[0077] S12. The remote processing center obtains the inspection data including the inspection pictures, and identifies the rust on the tank top of each storage tank to be determined according to the inspection pictures through a preset prediction model; the prediction model includes:

[0078] A rectangle recognition sub - model, which is used to generate a rectangle frame label of the to - be - measured storage tank in the inspection picture; the rectangle recognition sub - model is generated by training with the original inspection picture combined with the rectangle frame label; the rectangle frame label of the tank top pattern is obtained by box - selecting the original inspection picture as historical data, and a tank top mask label and a rust label are generated; the original inspection picture is obtained by an image acquisition device arranged on the unmanned aerial vehicle.

[0079] A mask recognition sub - model, which is used to generate a tank top mask label of the to - be - measured storage tank in the tank top rectangle image according to the tank top rectangle image; the mask recognition sub - model is generated by training with the tank top rectangle image combined with the tank top mask label; the tank top rectangle image is obtained according to the rectangle label.

[0080] A preliminary rust recognition sub - model, which is used to obtain a preliminary rust recognition result according to the tank top rectangle image; the preliminary rust recognition sub - model is generated by training with the tank top rectangle image combined with the rust label.

[0081] In the embodiment of the present invention, the determination of the corrosion degree of the storage tank is realized by a computer according to a preset prediction model, that is, based on the inspection picture, the determination result of the corrosion degree of the to - be - measured storage tank is generated through the prediction model.

[0082] Since environmental factors such as light intensity may change at any time, there will be differences in brightness, etc. in inspection pictures of different batches; therefore, in order to facilitate the prediction model for image recognition, preferably, in the embodiment of the present invention, pre - processing such as extracting the light field can also be performed on the obtained inspection pictures.

[0083] In addition, in order to be able to distinguish and identify each storage tank in the inspection picture, in the embodiment of the present invention, the azimuth and / or coordinate system of each inspection picture during one inspection process can also be adjusted. Specifically:

[0084] The inspection data also includes the flight parameters of the unmanned aerial vehicle, and the flight parameters include longitude, latitude, yaw angle, pitch angle, roll angle and flight height when taking the inspection picture; the azimuth and / or coordinate system of each inspection picture is unified by image rotation according to the flight parameters corresponding to the inspection picture; combined with the actual positions of each storage tank in the storage tank area, each storage tank in the inspection picture is identified and recognized; according to the identification and recognition result, each tank top rectangle image is respectively marked.

[0085] The prediction model in the embodiment of the present invention includes multiple sub - models (that is, a rectangle recognition sub - model, a mask recognition sub - model, and a preliminary rust recognition sub - model) to respectively complete different sub - functions, where the rectangle recognition sub - model is used to generate a rectangle frame label of the to - be - measured storage tank in the inspection picture according to the inspection picture.

[0086] In practical applications, the rectangular recognition sub-model can be trained and generated by combining the original inspection pictures with rectangular box labels; that is, using the original inspection pictures as historical data for modeling training, which specifically may include: First, obtain the rectangular box labels of the tank top from the original inspection pictures by manual box selection; after rectangular box selection, rust labels and tank top mask labels can be further generated by automatic and semi-automatic label methods.

[0087] In practical applications, the specific steps for generating rust labels by the automatic label generation method may include:

[0088] Select rust seed region features through the clustering method;

[0089] Obtain preliminary rust labels through the seed region growth method;

[0090] Perform clustering again to update the seed region features;

[0091] Update the rust labels again through the seed region growth method;

[0092] End when the clustering center no longer changes or the iteration termination signal is reached.

[0093] In practical applications, the specific steps for generating the tank top mask label by the semi-automatic label generation method can be as follows:

[0094] Select the tank top candidate region by edge detection;

[0095] Select the largest polygon object as the rough label;

[0096] Obtain the final tank top mask label after manual fine-tuning.

[0097] It should be noted that for the same reason, in the embodiments of the present invention, the original inspection pictures also need to be pre-processed such as extracting the illumination field.

[0098] In practical applications, the original inspection pictures can also be obtained by an image acquisition device provided on the unmanned aerial vehicle.

[0099] The mask recognition sub-model in the embodiments of the present invention is trained and generated by combining the tank top rectangular image with the tank top mask label; specifically, after obtaining the rectangular box label of the tank top according to the original inspection picture, the corresponding tank top rectangular image can be further generated; then, using the tank top rectangular image as the modeling data and combining the tank top mask label generated by automatic and semi-automatic label methods for training, so as to construct a mask recognition sub-model that can generate the corresponding tank top mask label according to the tank top rectangular image.

[0100] The preliminary rust recognition sub-model in the embodiment of the present invention is trained and generated using rectangular images of tank tops combined with rust labels. Specifically, the training is performed using the rectangular images of the tank tops as modeling data, combined with rust labels generated through automatic and semi-automatic labeling methods, thereby constructing a preliminary rust recognition sub-model capable of generating corresponding rust labels based on the rectangular images of the tank tops. It should be noted that the recognition results of the preliminary rust recognition sub-model in the embodiment of the present invention are generated based on the rectangular images of the tank tops, while the orthophoto image of the tank top is supposed to be circular. Therefore, the recognition results of the preliminary rust recognition sub-model are likely to include the recognition of the vicinity of the tank top's exterior. In other words, the recognition results of the preliminary rust recognition sub-model are not yet accurate enough.

[0101] Next, the embodiment of the present invention can also use the tank top mask tag to remove redundant parts in the recognition result of the rust recognition sub-model, thereby obtaining the final photo rust recognition result of the storage tank to be measured.

[0102] In practical applications, multiple current inspection pictures will be obtained during each inspection. After preprocessing the inspection pictures, the rectangle recognition sub-model can be used to generate rectangular frame labels for each tank to be measured based on each inspection picture.

[0103] After obtaining the rectangular frame label of the storage tank to be measured, it is also necessary to generate the tank top mask label of the storage tank to be measured in the tank top rectangular image according to the tank top rectangular image through the mask recognition sub-model.

[0104] In the embodiment of the present invention, each storage tank to be measured corresponds to a corresponding tank top rectangular image, and the rust corrosion points therein can be identified through the preliminary rust recognition sub-model.

[0105] S13, the remote processing center determines the rust recognition result of the photo of the storage tank to be measured based on the tank top mask label generated by the mask recognition sub-model and the preliminary rust recognition result;

[0106] The recognition result of the preliminary rust recognition sub-model is generated based on the rectangular image of the tank top, while the orthographic image of the tank top should be circular. Therefore, the recognition result is likely to include the recognition of the ground or other equipment near the outside of the tank; that is, the recognition result of the preliminary rust recognition sub-model is not accurate enough.

[0107] To this end, in an embodiment of the present invention, the tank top mask label generated by the mask recognition sub-model is also utilized. The tank top mask label and the preliminary rust recognition result are used to eliminate the redundant parts in the preliminary recognition result of the rust recognition sub-model, thereby obtaining the final photo rust recognition result of the storage tank to be measured.

[0108] S14. Determine the corrosion area ratio of the storage tank to be measured according to the recognition result of the photo rust.

[0109] In practical applications, the corrosion degree of the storage tank top can be measured by the corrosion area ratio; specifically, the corresponding relationship between the corrosion degree of the storage tank and the corrosion area ratio can be set according to the experience of those skilled in the art or limited experiments; in this way, after obtaining the corrosion area ratio of the storage tank to be measured, the corrosion degree of the storage tank to be measured can be determined according to this corresponding relationship.

[0110] In practical applications, the specific steps for calculating the corrosion area of the storage tank to be measured may include:

[0111] Determine the direction of the storage tank to be measured in the inspection image (rectangular image of the tank top) according to the flight parameters of the aircraft; the flight parameters of the aircraft include longitude latitude yaw angle θ i , pitch angle β i , roll angle α i and flight altitude H i . According to the yaw angle θ i , rotate the inspection image counterclockwise with the image center. Any point coordinate (p0, q0) of the original inspection image becomes p and q after rotation. The transformation formula is:

[0112]

[0113] The coordinate of the image center point after transformation is Combine the position of the detection frame after transformation to obtain the search direction line l i and direction vector

[0114]

[0115] Combine the position information of the storage tank to determine which storage tank in reality is the one detected in the image. The position information of the storage tank includes the longitude of the storage tank latitude and the diameter d of the storage tank i , satisfying that the vector formed by the storage tank position point and the UAV position point is an acute angle with the direction vector and the distance from the storage tank position point to the search direction line is the smallest, then the matching calculation can be completed.

[0116] The calculation of the scale is:

[0117] δ i = 2d i / (w i + hi )

[0118] The meaning of the scale is the actual size corresponding to each pixel.

[0119] The specific working methods of the mask recognition sub-model and the preliminary rust recognition sub-model may include: respectively sending the detected rectangular image of the tank top into the tank top Mask segmentation network and the rust recognition network; the tank top segmentation network obtains the Mask information of the tank top (i.e., the tank top mask label), and the rust recognition network obtains a rough preliminary rust recognition result;

[0120] The tank top segmentation network is sensitive to the perception of shape, and the rust recognition network is sensitive to discrete rust. To increase the learnability of the rust recognition network, this solution divides the rectangular image of the tank top into image blocks of a fixed size for recognition.

[0121] Then, fuse the tank top mask label and the preliminary rust recognition result to obtain the final accurate photo rust recognition result. Statistically calculate the pixel area of the rust and the pixel area of the tank top Mask The degree of rust τ i and the rust area are:

[0122]

[0123] In summary, the embodiment of the present invention uses the original inspection pictures as historical data as modeling data, constructs a rectangular recognition sub-model for generating a rectangular frame label of the to-be-determined storage tank according to the inspection picture, a mask recognition sub-model for generating a tank top mask label of the to-be-determined storage tank in the rectangular image of the tank top according to the rectangular image of the tank top, and a preliminary rust recognition sub-model for obtaining a preliminary rust recognition result according to the rectangular image of the tank top; thus, according to the current inspection picture obtained by the inspection device (such as a drone equipped with an image acquisition device), the preliminary rust recognition result of the rectangular image of the tank top can be obtained through the above-mentioned various sub-models; then, the rust of the tank top of the storage tank can be recognized according to the tank top mask label.

[0124] Through the present invention, the rust of the tank top of the storage tank can be accurately recognized, so the rust area ratio of the tank top of the storage tank can be accurately calculated according to the recognition result; thus, the accurate result of the rust determination of the tank top of the storage tank can be obtained according to the rust area ratio; since the present invention excludes the human subjective judgment link, the determination result is more accurate and stable.

[0125] Embodiment Two

[0126] Reference Figure 3, on the basis of Embodiment 1, the embodiments of the present invention may further include the following content:

[0127] The inspection data further includes scanning data; the unmanned aerial vehicle is further provided with a synthetic aperture radar for acquiring scanning data synchronized with the image acquisition device; the scanning data is synchronized and in one-to-one correspondence with the inspection pictures acquired by the image acquisition device;

[0128] The remote processing center can be used to perform the following steps:

[0129] S15. Generate a SAR image consistent with the coordinate system of the inspection picture corresponding to the current scanning data according to the current scanning data;

[0130] In the embodiments of the present invention, the correlation between the scattering characteristics of the SAR image of the synthetic aperture radar and the surface roughness of the scanned object is further utilized to determine the corrosion depth of the storage tank; specifically, through the SAR image of the scanned object acquired by the synthetic aperture radar, the surface roughness of the scanned object can be inferred according to the scattering characteristics of the SAR image. For the storage tank, the higher the roughness of the rust on it, the deeper the corrosion depth, and the corrosion depth is also an important index for measuring the corrosion degree of the storage tank.

[0131] If the unmanned aerial vehicle carries the synthetic aperture radar alone to scan and obtain the scanning data of the storage tank area, the scanning data obtained by the moving synthetic aperture radar will have a position accuracy deviation due to the combined solution of the spatial inertial navigation. In this way, the SAR image obtained thereby cannot be directly used; for this reason, in the embodiments of the present invention, the unmanned aerial vehicle is adopted to carry the image acquisition device and the synthetic aperture radar at the same time, and the inspection pictures and the scanning data are acquired synchronously. In this way, referring to the coordinates of the inspection pictures corresponding to the scanning data to generate the SAR image can avoid the problem of position accuracy deviation existing in the SAR image generated by the unmanned aerial vehicle carrying the synthetic aperture radar alone.

[0132] S16. Determine a reflected rust area corresponding to the rust area in the photo rust recognition result in the SAR image;

[0133] Since the coordinates of the SAR image and the coordinates of the inspection pictures are the same, in the embodiments of the present invention, the corresponding area in the SAR image can be determined according to the rust area in the photo rust recognition result as the reflected rust area.

[0134] S17. Determine the corrosion depth of the reflected rust area according to the gray value of the reflected rust area.

[0135] The scattering characteristics of the surface of the object to be scanned can be obtained through synthetic aperture radar. Some typical ones include: the roughness of the surface of the object to be scanned determines the gray value (i.e., the echo intensity) of the SAR image; the inventor found through research that the roughness of the rust on the surface of the storage tank is strongly correlated with the corrosion depth. Therefore, in the embodiments of the present invention, the scattering characteristics represented by the SAR image are used as the measurement parameter for the corrosion depth, that is, the greater the gray value of the SAR image, the deeper the corrosion depth, and vice versa.

[0136] Specifically, when determining the corrosion depth of the reflected rust area, the corresponding relationship between the gray value of the SAR image and the corrosion depth can be preset; then, according to the gray value of the reflected rust area in the SAR image and the corresponding relationship, the corrosion depth of the reflected rust area is determined.

[0137] In summary, on the basis of Embodiment 1, the embodiments of the present invention can also include the scan data collected by a synthetic aperture radar synchronized with the image acquisition device, and generate a SAR image consistent with the coordinate system of the corresponding inspection picture according to the scan data; since the scattering characteristics of the SAR image can reflect the roughness of the surface of the object being photographed, the rust area can be determined in the SAR image according to the recognition result of the inspection picture; then, according to the different scattering characteristics in the rust area, the corresponding corrosion depth is determined; since through the embodiments of the present invention, not only the rust area (or rust area ratio) of the tank top of the storage tank can be determined, but also different corrosion depths in the rust area can be identified, more accurate and comprehensive measurement results of the corrosion degree and corrosion situation can be obtained.

[0138] Embodiment 3

[0139] On the other hand of the embodiments of the present invention, a system for measuring the rust of the tank top of a storage tank is also provided. Figure 2 The structure diagram of the system for measuring the rust of the tank top of a storage tank provided by the embodiments of the present invention is shown. The system for measuring the rust of the tank top of a storage tank is Figure 1 or Figure 3 the system corresponding to the method for measuring the rust of the tank top of a storage tank described in the corresponding embodiment, that is, it is implemented in the form of a virtual device Figure 1 or Figure 3 the method for measuring the rust of the tank top of a storage tank described in the corresponding embodiment. Each virtual module constituting the system for measuring the rust of the tank top of a storage tank can be executed by an electronic device, such as a network device, a terminal device, or a server. The system for measuring the rust of the tank top of a storage tank in the embodiments of the present invention can achieve the measurement of the rust of the tank top required for industrial control. Specifically, the system for measuring the rust of the tank top of a storage tank in the embodiments of the present invention includes: a drone 01 and a remote processing center 02;

[0140] The drone 01 is used to obtain the inspection data of the tank farm 03; the inspection data includes the inspection pictures of the tank farm 03 obtained by the image acquisition device 11 provided on the drone 01; the inspection pictures are orthophotos;

[0141] The remote processing center 02 includes a prediction model, a preliminary rust recognition sub-model, a photo rust recognition unit, and an area ratio calculation unit;

[0142] The prediction model is used to identify the rust on the tank tops of each tank to be measured according to the inspection pictures; the prediction model includes:

[0143] A rectangle recognition sub-model, which is used to generate rectangle box labels of the tanks to be measured in the inspection pictures according to the inspection pictures; the rectangle recognition sub-model is trained and generated by combining the original inspection pictures with the rectangle box labels; the rectangle box labels of the tank top patterns are obtained by box selection of the original inspection pictures as historical data, and the tank top mask labels and rust labels are generated; the original inspection pictures are obtained by the image acquisition device 11 provided on the drone 01;

[0144] A mask recognition sub-model, which is used to generate the tank top mask labels of the tanks to be measured in the tank top rectangle images according to the tank top rectangle images; the mask recognition sub-model is trained and generated by combining the tank top rectangle images with the tank top mask labels; the tank top rectangle images are obtained according to the rectangle labels;

[0145] The preliminary rust recognition sub-model is used to obtain the preliminary rust recognition result according to the tank top rectangle image; the preliminary rust recognition sub-model is trained and generated by combining the tank top rectangle image with the rust label;

[0146] The photo rust recognition unit is used to determine the photo rust recognition result of the tank to be measured according to the tank top mask label generated by the mask recognition sub-model and the preliminary rust recognition result.

[0147] The area ratio calculation unit is used to determine the rust area ratio of the tank to be measured according to the photo rust recognition result.

[0148] Further, in the embodiment of the present invention, the drone 01 is also provided with a synthetic aperture radar (not shown in the figure), which is used to obtain the scanning data synchronized with the image acquisition device 11; the scanning data is synchronized and in one-to-one correspondence with the inspection pictures obtained by the image acquisition device 11; the inspection data also includes the scanning data;

[0149] The remote processing center 02 also includes:

[0150] A coordinate conversion unit, configured to generate a SAR image consistent with the inspection picture coordinate system corresponding to the current scan data according to the current scan data;

[0151] A reflection rust determination unit, configured to determine a reflection rust area corresponding to the rust area in the photo rust recognition result in the SAR image;

[0152] A rust depth determination unit, configured to determine the rust depth of the reflection rust area according to the gray value of the reflection rust area.

[0153] Since the working principle and beneficial effects of the storage tank top rust determination system in the embodiments of the present invention have been recorded and described in the Figure 1 corresponding storage tank top rust determination method, they can be referred to each other and will not be elaborated here.

[0154] Embodiment 4

[0155] Corresponding to the method embodiment, the present application further provides a storage tank top rust determination device, such as a terminal, a server, etc. Among them, the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc., but is not limited thereto.

[0156] An example diagram of the hardware structure block diagram of the storage tank top rust determination device provided by the embodiments of the present invention is as Figure 4 shown, and may include:

[0157] A processor 1, a communication interface 2, a memory 3, and a communication bus 4;

[0158] Wherein, the processor 1, the communication interface 2, and the memory 3 complete mutual communication through the communication bus 4;

[0159] Optionally, the communication interface 2 can be an interface of a communication module, such as an interface of a GSM module; the processor 1 may be a central processing unit CPU, or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention.

[0160] The memory 3 may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory.

[0161] Among them, the processor 1 is specifically configured to execute the computer program stored in the memory 3 to perform the following steps:

[0162] Identify the rust on the tank tops of each tank to be measured according to the inspection pictures through a preset prediction model; the prediction model includes:

[0163] A rectangle recognition sub-model, which is used to generate rectangle box labels of the tanks to be measured in the inspection pictures; the rectangle recognition sub-model is generated by training with the original inspection pictures combined with the rectangle box labels; the rectangle box labels of the tank top patterns are obtained by box selection of the original inspection pictures as historical data, and the tank top mask labels and rust labels are generated; the original inspection pictures are obtained by the image acquisition device provided on the unmanned aerial vehicle;

[0164] A mask recognition sub-model, which is used to generate the tank top mask labels of the tanks to be measured in the tank top rectangle image; the mask recognition sub-model is generated by training with the tank top rectangle image combined with the tank top mask labels; the tank top rectangle image is obtained according to the rectangle label;

[0165] A preliminary rust recognition sub-model, which is used to obtain a preliminary rust recognition result according to the tank top rectangle image; the preliminary rust recognition sub-model is generated by training with the tank top rectangle image combined with the rust label;

[0166] Determine the photo rust recognition result of the tank to be measured according to the tank top mask label generated by the mask recognition sub-model and the preliminary rust recognition result;

[0167] Determine the rust area ratio of the tank to be measured according to the photo rust recognition result.

[0168] Further, in the embodiment of the present invention, the inspection data further includes scan data; the unmanned aerial vehicle is further provided with a synthetic aperture radar for acquiring scan data synchronized with the image acquisition device; the scan data is synchronized and in one-to-one correspondence with the inspection pictures acquired by the image acquisition device;

[0169] The processor 1 is specifically further configured to execute the computer program stored in the memory 3 to perform the following steps:

[0170] S15. Generate a SAR image corresponding to the current scan data and consistent with the coordinate system of the inspection picture;

[0171] S16. Determine the reflected rust area corresponding to the rust area in the photo rust recognition result in the SAR image;

[0172] S17. Determine the corrosion depth of the reflected rust area according to the gray value of the reflected rust area.

[0173] When the program instructions included in the computer program product of the storage tank top corrosion determination device in the embodiment of the present invention are executed by a computer, the computer can execute each step performed by the remote processing center in the storage tank top corrosion determination method described in each of the above aspects, and achieve the same technical effects.

[0174] Embodiment Five

[0175] In the embodiment of the present invention, a storage medium is further provided. The storage medium can store a program suitable for being executed by a processor. The program is used for:

[0176] Identifying the top rust of each storage tank to be measured according to the inspection pictures through a preset prediction model. The prediction model includes:

[0177] A rectangle recognition sub-model for generating a rectangle box label of the storage tank to be measured in the inspection picture according to the inspection picture. The rectangle recognition sub-model is generated by training with the original inspection picture combined with the rectangle box label. The rectangle box label of the tank top pattern is obtained by box selection of the original inspection picture as historical data, and the tank top mask label and the rust label are generated. The original inspection picture is obtained by an image acquisition device provided on the unmanned aerial vehicle.

[0178] A mask recognition sub-model for generating a tank top mask label of the storage tank to be measured in the tank top rectangle image according to the tank top rectangle image. The mask recognition sub-model is generated by training with the tank top rectangle image combined with the tank top mask label. The tank top rectangle image is obtained according to the rectangle label.

[0179] A preliminary rust recognition sub-model for obtaining a preliminary rust recognition result according to the tank top rectangle image. The preliminary rust recognition sub-model is generated by training with the tank top rectangle image combined with the rust label.

[0180] Determine the photo rust recognition result of the storage tank to be measured according to the tank top mask label generated by the mask recognition sub-model and the preliminary rust recognition result.

[0181] Determine the corrosion area ratio of the storage tank to be measured according to the photo rust recognition result.

[0182] Further, in the embodiment of the present invention, the inspection data further includes scanning data. The unmanned aerial vehicle is further provided with a synthetic aperture radar for obtaining scanning data synchronized with the image acquisition device. The scanning data is synchronized and in one-to-one correspondence with the inspection pictures obtained by the image acquisition device. The program is further used for:

[0183] S15. Generate a SAR image that is consistent with the inspection picture coordinate system corresponding to the current scan data based on the current scan data;

[0184] S16. Determine a reflected rust area corresponding to the rust area in the photo rust recognition result in the SAR image;

[0185] S17. Determine the rust depth of the reflected rust area according to the gray value of the reflected rust area.

[0186] Optionally, the refinement function and expansion function of the program can refer to the above description.

[0187] The above product can execute the method provided by the embodiment of the present invention and has corresponding functional modules and beneficial effects for executing the method. For technical details not described in detail in this embodiment, reference can be made to the method provided by the embodiment of the present invention.

[0188] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0189] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. Another point is that the couplings or direct couplings or communication connections shown or discussed with each other can be through some interfaces, and the indirect couplings or communication connections of devices or units can be in an electrical, mechanical, or other form.

[0190] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0191] In addition, in each embodiment of the present application, the functional units can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0192] It should be understood that in the embodiments of the present application, the dependent claims, each embodiment, and the features can be combined with each other to achieve the solution of the foregoing technical problems.

[0193] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such understanding, the technical solution of this application, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0194] The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use this application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application will not be limited to the embodiments shown herein, but rather will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for measuring the corrosion of a storage tank roof, characterized in that Including the steps: S11. Obtain the inspection data of the tank farm by using a drone; the inspection data includes the inspection pictures of the tank farm obtained by an image acquisition device provided on the drone; the inspection pictures are orthophotos; S12. The remote processing center obtains the inspection data including the inspection pictures, and identifies the rust on the tank tops of each tank to be measured according to the inspection pictures by using a preset prediction model; the prediction model includes: A rectangle recognition sub-model, which is used to generate a rectangle box label of the tank to be measured in the inspection picture; the rectangle recognition sub-model is generated by training with the original inspection picture combined with the rectangle box label; the rectangle box label of the tank top pattern is obtained by frame selection of the original inspection picture as historical data, and a tank top mask label and a rust label are generated; the original inspection picture is obtained by an image acquisition device provided on the drone; A mask recognition sub-model, which is used to generate a tank top mask label of the tank to be measured in the tank top rectangle image according to the tank top rectangle image; the mask recognition sub-model is generated by training with the tank top rectangle image combined with the tank top mask label; the tank top rectangle image is obtained according to the rectangle label; A preliminary rust recognition sub-model, which is used to obtain a preliminary rust recognition result according to the tank top rectangle image; the preliminary rust recognition sub-model is generated by training with the tank top rectangle image combined with the rust label; S13. The remote processing center determines the photo rust recognition result of the tank to be measured according to the tank top mask label generated by the mask recognition sub-model and the preliminary rust recognition result; S14. Determine the rust area ratio of the tank to be measured according to the photo rust recognition result; The inspection data further includes scanning data; the drone is further provided with a synthetic aperture radar for obtaining scanning data synchronized with the image acquisition device; the scanning data is synchronized and in one-to-one correspondence with the inspection pictures obtained by the image acquisition device; the remote processing center is further used for: S15. Generate a synthetic aperture radar (SAR) image corresponding to the current scanning data and consistent with the coordinate system of the inspection picture; S16. Determine a reflected rust area corresponding to the rust area in the photo rust recognition result in the SAR image; S17. Determine the rust depth of the reflected rust area according to the gray value of the reflected rust area; The determining the rust depth of the reflected rust area according to the gray value of the reflected rust area includes: Presetting the correspondence between the gray value of the SAR image and the rust depth; Determining the rust depth of the reflected rust area in the SAR image according to the gray value of the reflected rust area in the SAR image and the correspondence; The generating the tank top mask label includes: Selecting a tank top candidate area through edge detection; Selecting the largest polygon object as a rough label; Obtaining the final tank top mask label after manual fine-tuning.

2. The method for measuring the corrosion of the storage tank roof according to claim 1, characterized in that, Including: The rectangle box label is obtained in the original inspection picture by manual frame selection.

3. The method for measuring the corrosion of the storage tank roof according to claim 1, characterized in that, Generating the rust label includes: Selecting the rust seed area features by using a clustering method; Obtain preliminary rust labels through the seed region growth method; Perform clustering again to update the seed region features; Update the rust labels again through the seed region growth method; End when the clustering center no longer changes or the iteration termination signal is reached.

4. The method for measuring the rust of the storage tank top according to claim 1, wherein The flight altitude of the drone is greater than 200 meters.

5. The method for measuring the rusting of the storage tank roof according to claim 1, characterized in that, The proportion of the image of the storage tank to be measured in the inspection pictures is between 0.2 and 0.

4.

6. The method for measuring the rust of the storage tank roof according to claim 1, characterized in that, It also includes: The inspection data also includes the flight parameters of the drone, and the flight parameters include longitude, latitude, yaw angle, pitch angle, roll angle, and flight altitude when taking the inspection pictures; Unify the coordinate systems of the inspection pictures according to the flight parameters corresponding to the inspection pictures through image rotation; Combined with the actual positions of the storage tanks in the storage tank area, identify each storage tank in the inspection pictures; According to the identification result, label each of the tank top rectangular images respectively.

7. A corrosion measurement system for the tank top of a storage tank, characterized in that, It includes: A drone and a remote processing center; The drone is used to obtain inspection data of the tank area; the inspection data includes inspection pictures of the tank area obtained by an image acquisition device arranged on the drone; the inspection pictures are orthophotos; The remote processing center includes a prediction model, a preliminary rust recognition sub-model, a photo rust recognition unit, and an area ratio calculation unit; The prediction model is used to identify the rust on the tank tops of each storage tank to be measured according to the inspection pictures; the prediction model includes: A rectangle recognition sub-model, which is used to generate rectangle box labels of the storage tanks to be measured in the inspection pictures according to the inspection pictures; the rectangle recognition sub-model is trained and generated by combining the original inspection pictures with the rectangle box labels; the rectangle box labels of the tank top patterns are obtained by frame selection of the original inspection pictures as historical data, and the tank top mask labels and rust labels are generated; the original inspection pictures are obtained by an image acquisition device arranged on the drone; A mask recognition sub-model, which is used to generate the tank top mask labels of the storage tanks to be measured in the tank top rectangular images according to the tank top rectangular images; the mask recognition sub-model is trained and generated by combining the tank top rectangular images with the tank top mask labels; the tank top rectangular images are obtained according to the rectangle labels; The preliminary rust recognition sub-model is used to obtain a preliminary rust recognition result according to the tank top rectangular images; the preliminary rust recognition sub-model is trained and generated by combining the tank top rectangular images with the rust labels; The photo rust recognition unit is used to determine the photo rust recognition result of the storage tank to be measured according to the tank top mask labels generated by the mask recognition sub-model and the preliminary rust recognition result; The area ratio calculation unit is used to determine the corrosion area ratio of the storage tank to be measured according to the photo rust recognition result; The inspection data also includes scanning data; the drone is also equipped with a synthetic aperture radar for obtaining scanning data synchronized with the image acquisition device; the scanning data is synchronized and one-to-one corresponding to the inspection pictures obtained by the image acquisition device; The remote processing center also includes: A coordinate conversion unit, configured to generate a SAR image consistent with the inspection picture coordinate system corresponding to the current scan data according to the current scan data; A reflection rust determination unit, configured to determine a reflection rust area corresponding to the rust area in the photo rust recognition result in the SAR image; A rust depth determination unit, configured to determine the rust depth of the reflection rust area according to the gray value of the reflection rust area; The determining the rust depth of the reflection rust area according to the gray value of the reflection rust area includes: Presetting the correspondence between the gray value of the SAR image and the rust depth; Determining the rust depth of the reflection rust area according to the gray value of the reflection rust area in the SAR image and the correspondence; The generating the tank top mask label includes: Selecting a tank top candidate area through edge detection; Selecting the largest polygon object as the rough label; Obtaining the final tank top mask label after manual fine-tuning.

8. A storage tank top rust determination device, comprising: A memory, configured to store a computer program; A processor, configured to call and execute the computer program to implement each step of the storage tank top rust determination method according to any one of claims 1-6.

9. A storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each step of the storage tank top rust determination method according to any one of claims 1-6 is implemented.

Images