A power inspection robot positioning image construction method and device, electronic equipment and storage medium

Abstract

The present application relates to a kind of electric power inspection robot positioning image construction method, device and equipment and storage medium, obtain the video selected shooting object main body characteristic picture and the static photo of shooting object main body, carry out fusion processing to obtain target picture, the target picture of different target area corresponds to different database function, while obtaining target picture, obtain the location information of shooting object main body and the positioning information of inspection robot, receive the same target picture of different target area in different time range, form history database in cloud, by using the method of the present application, can obviously improve the safety and precision and efficiency problem of power grid system inspection in the working process of inspection robot, reduce the inspection cost and risk problem, solve the problem that need artificial fault judgment and marking etc., improve the image construction technology and fault positioning accuracy of relevant inspection robot.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power inspection, in particular to a power inspection robot positioning image construction method and device, electronic equipment and a storage medium. BACKGROUND

[0002] Power and power grid system inspection has always relied on manual work. Due to the geographical complexity of power grid equipment construction, as well as the high-risk nature of the equipment and power grid system itself, the inspection cost is high, the risk is great, and other problems such as the difficulty of inspection in mountainous areas, sparsely populated areas, and the like, the potential danger of high voltage and high current of power grid equipment itself, and the labor cost and personal safety problem of equipment inspection, the existing technology has adopted devices such as patrol robots and unmanned aerial vehicles for inspection, but in the existing technology, the inspection effect is still not good, and manual fault judgment and marking are required, and other problems, therefore, how to improve the image construction technology of the related inspection robot and the fault positioning accuracy, improve the calculation precision and efficiency of the inspection, and the like, has become a problem to be solved. SUMMARY

[0003] In order to overcome the defects of the background art, the present application provides a power inspection robot positioning image construction method, device and system, and the scheme is as follows:

[0004] A power inspection robot positioning image construction method periodically acquires inspection dynamic video information of a target area; divides the target area scene according to the inspection dynamic video information, and selects a shooting object main feature picture from the video, and the target area scene division at least includes two-dimensional scene videos in different light conditions, different camera angles, and different time points of the video shooting object main body;

[0005] Periodically acquire a static photo of the shooting object main body in the target area; the static photo is obtained by detail sharpening enhancement and composition processing of multiple shooting photos under multiple exposure parameters,

[0006] The shooting object main feature picture selected from the video and the static photo of the shooting object main body are acquired, and a target picture is obtained by fusion processing, and the target pictures of different target areas correspond to different database functions Q m At the same time of acquiring the target picture, the position information L m of the shooting object main body and the inspection robot positioning information I m are acquired, the database function Q m has a one-to-one correspondence with the position information L m and I m , and a historical database is formed in the cloud according to the same target picture of different target areas in different time ranges [t1, t2].

[0007] Further, when the inspection discovers an anomaly, the most recent time t is called m The target picture Gm is compared with the inspection anomaly picture Bm, and before the comparison, the positioning information of the inspection robot is used to capture the position information L of the subject m of the target area m The cloud forms a historical database sequence L m = (L1, L2,..., L n ); I m = (I1, I2,..., I n ), and the corresponding L m and I m values are called through the database function Q m The corresponding target picture Gm is called, the target anomaly position accurate image is obtained through the comparison between the target picture Gm and the inspection anomaly picture Bm, and the target loss key area image historical database is formed in the cloud.

[0008] Further, according to the target loss key area image historical database, the target vulnerable loss area is divided: the area of the target picture is Sm, the area of the vulnerable loss area is Sn, the target anomaly position coordinate point C(x, y) is recorded, and the two target anomaly positions (X1, Y1) and (X2, Y2) farthest from each other are determined as the diameter of the Sn area Thus Sn = π*(R / 2) 2 The center position K(X0, Y0) of the vulnerable loss area is obtained, wherein

[0009] Further, by recording the target anomaly position coordinate point C(x, y), a sample clustering method is used, the sample set of the target anomaly position coordinate point is calculated, the area of the vulnerable loss area Sn is obtained, and the function f(x) of the center position K of the vulnerable loss area can also be calculated in the following way:

[0010]

[0011] Wherein C is the input target anomaly position coordinate point, C = {C1, C2,..., Ck}, D is the output target anomaly position coordinate point, D = {D1, D2,..., Dk}.

[0012] Further, when the inspection robot captures the inspection dynamic video information of the target area and the static photo of the subject in the target area, the positioning information can be sent to other nearby inspection robots and asked to go to the target area for shooting assistance.

[0013] Further, each abnormal position in the target picture historical state is identified, and the subject of the photographed object in the picture is clustered and identified according to different components, and the clustering identification information at least includes the time, position information and defect type information of the defect position of the target picture.

[0014] The power inspection robot positioning image construction device comprises a video acquisition module, a periodic target area inspection dynamic video information acquisition module, a target area scene division module, a video shooting object main feature picture selection module, a target area scene division module, and a target area scene division module.

[0015] The photo acquisition module periodically acquires static photos of the target area shooting object main body, and the static photos are obtained by detail sharpening enhancement and synthesis processing of multiple shooting photos under multiple exposure parameters,

[0016] The cloud processing module acquires the video shooting object main feature picture and the static photo of the shooting object main body, and performs fusion processing to obtain a target picture, and different target pictures of different target areas correspond to different database functions Q m At the same time of acquiring the target picture, the position information L m of the shooting object main body is acquired, and the inspection robot positioning information I m The database function Q m and the position information L m and I m have a one-to-one correspondence, and the same target picture of different target areas in different time ranges [t1, t2] is formed into a historical database in the cloud.

[0017] An electronic device comprises a processor and a memory storing a program, the program comprising instructions that, when executed by the processor, cause the processor to perform the power inspection robot positioning image construction method.

[0018] A computer-readable storage medium storing a program, the program comprising instructions that, when executed by a processor of an electronic device, cause the electronic device to perform the power inspection robot positioning image construction method.

[0019] By adopting the method, device, equipment and storage medium, the safety, accuracy and efficiency of the power grid system inspection during the working process of the inspection robot can be obviously improved, the inspection cost and risk problem is reduced, the problem of manual fault judgment and marking is solved, and therefore the image construction technology and fault positioning accuracy of the related inspection robot are improved, and the calculation accuracy and efficiency of the power grid inspection are improved. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 is a method flowchart diagram according to an embodiment of the present application;

[0021] Figure 2 is a derivation diagram of the range of the vulnerable area according to an embodiment of the present application, wherein Figure 2 a-2b is a calculation formula derivation process diagram of the range of the vulnerable area of the present application;

[0022] Figure 3 is a working state diagram of the inspection robot according to an embodiment of the present application;

[0023] Figure 4 is a target picture abnormal position identification method diagram according to an embodiment of the present application;

[0024] Figure 5 is a power inspection robot positioning image construction device diagram according to an embodiment of the present application; DETAILED DESCRIPTION

[0025] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail with reference to the drawings. The description is introduced by way of examples and is not limited, the specific embodiments consistent with the principles of the present application are introduced in sufficient detail so that the persons skilled in the art can practice the present application, other embodiments can be used and the structure of each element can be changed and / or replaced without departing from the scope and spirit of the present application. Therefore, the following detailed description should not be understood in a limiting sense.

[0026] In the first embodiment

[0027] A power inspection robot positioning image construction method, periodically acquires inspection dynamic video information of a target area; the target area can be an environmental area including the surroundings of the power equipment, or a power equipment area, divides the target area scene according to the inspection dynamic video information, for example, can shoot the target area video under different light conditions such as sunny day, cloudy day, different shooting angles such as top view, side view, different time such as daytime, night, and select the shooting object main characteristic picture from the video, for example, the shooting object main characteristic picture reflects the main structure of a certain power equipment or the key area of the power equipment itself, the target area scene division at least includes two-dimensional scene video in different light conditions, different camera angles, different time of the video shooting object main body;

[0028] The system periodically acquires still photographs of the main subject within a target area. These still photographs are obtained by combining multiple photographs taken under various exposure parameters with detail sharpening and enhancement processes. For example, photographs can be taken simultaneously with video recording. The final still photograph of the main subject is obtained by fusing multiple photographs, with the aim of obtaining high-resolution still images.

[0029] The target image is obtained by fusing selected images of the main subject from the video with still photos of the same subject. Through multi-dimensional processing of both dynamic video and still photos, high-quality, high-definition, and high-fidelity target image information is finally obtained, which can be stored in a cloud database. Target images of different target areas correspond to different database functions Q. m While acquiring the target image, obtain the location information of the main subject L. m With the inspection robot's positioning information I m Database function Q m Location information L m and I m There is a one-to-one correspondence to obtain the location of the subject being photographed (e.g., power equipment). For example, the location can be obtained by installing GPS (Global Positioning System) on the power equipment itself or by using relevant electronic map positioning information. Inspection robots are generally equipped with GPS devices, which transmit their own geographical location information during the inspection process. By establishing Lm and Im related location information in the database, a complete positioning information database can be formed. The location information is corrected by using the fixed positioning information of the equipment itself and the dynamic positioning information of the inspection robot, which improves the accuracy of the location information in the database. Then, the accurate historical location information of the relevant power equipment can be retrieved through the database function Qm. At the same time, by receiving the same target image in different target areas at different time ranges [t1, t2], since the location of the inspection robot may be different at different time ranges, the location information can be corrected in real time by receiving location information from multiple time periods. A historical database is formed in the cloud, that is, the database information corresponding to the target image and the precise location is obtained, and the high-precision positioning effect is achieved. Moreover, due to the periodic reception and processing of images and videos, the timeliness and accuracy of the historical database can be guaranteed.

[0030] Furthermore, when an anomaly is detected during the inspection, the most recent time t is invoked. m The target image Gm is compared with the inspection anomaly image Bm. Before the comparison, the location information L of the main object is captured based on the positioning information of the inspection robot. m With the inspection robot's positioning information I m Traversing the cloud to form a historical database sequence L m = (L1, L2, ..., Ln ) ; I m = (I1, I2,..., In), find the corresponding L n ), find the corresponding L m and I m value through the database function Q m call the corresponding target image Gm, through the comparison of the target image Gm and the inspection abnormal picture Bm, the target abnormal position accurate image is obtained, which can be, for example, the position of the abnormal and damaged power equipment, and the target loss key area image history database is formed in the cloud, through the formation of the database of the places prone to damage or abnormality, in the future inspection process, the focus can be more focused on the related key area position.

[0031] Further, according to the target loss key area image history database, the target vulnerable loss area is divided: the area of the target picture is Sm, the area of the vulnerable loss area is Sn, the coordinates of the target abnormal position C (x, y) are recorded, and the two target abnormal positions (X1, Y1) and (X2, Y2) farthest from each other are determined as the diameter of the Sn area Thus Sn = π * (R / 2) 2 At the same time, the center position K (X0, Y0) of the vulnerable loss area is obtained, wherein

[0032] Further, by recording the target abnormal position coordinate point C (x, y), a sample clustering method is adopted, the sample set of the target abnormal position coordinate point is calculated, the area of the vulnerable loss area Sn and the function f (x) of the center position K of the vulnerable loss area are obtained, and the calculation method is as follows:

[0033]

[0034] , wherein C is the input target abnormal position coordinate point, C = {C1, C2,... Ck}, D is the output target abnormal position coordinate point, D = {D1, D2,... Dk}, the above calculation method is through a computer model, randomly setting the center position K, statistically collecting the input target abnormal position coordinate point C set, each point of the coordinate point C set is allocated to the D set, through calculating the correlation degree of C and the center position K, output to different D, and then update the actual center position of the whole D set, so as to obtain the above calculation method of the function f (x) of the center position K of the vulnerable loss area.

[0035] Further, when the inspection robot captures the inspection dynamic video information of the target area and the static photo of the target area shooting object, the positioning information can be sent to other inspection robots nearby, and the other inspection robots are required to go to the target area for shooting assistance. The other inspection robots can also upload the positioning information while shooting, so as to provide more positioning reference data, correct the positioning information of the database, and provide more accurate positioning information.

[0036] Further, each abnormal position in the historical state of the target picture is identified, and the shooting object in the picture is clustered and identified according to different components. The clustering identification information at least includes the time, position information and defect type information of the defect part of the target picture shooting.

[0037] An electric power inspection robot positioning image construction device, a video acquisition module periodically acquires inspection dynamic video information of a target area; the target area scene is divided according to the inspection dynamic video information, and a shooting object characteristic picture is selected from the video. The target area scene division at least includes two-dimensional scene videos in different light conditions, different camera angles and different time points of the video shooting object;

[0038] A photo acquisition module periodically acquires static photos of the target area shooting object; the static photos are obtained by detail sharpening enhancement and synthesis processing of multiple shooting photos under multiple exposure parameters,

[0039] A cloud processing module acquires the shooting object characteristic picture selected from the video and the static photo of the shooting object, performs fusion processing to obtain a target picture, and different target pictures of different target areas correspond to different database functions Q m At the same time of acquiring the target picture, the position information L m of the shooting object and the positioning information I m of the inspection robot are acquired, the database function Q m has a one-to-one correspondence with the position information L m and the positioning information I m , and a historical database is formed in the cloud according to the same target picture of different target areas in different time ranges [t1, t2].

[0040] In the second embodiment

[0041] A method for constructing localization images for a power inspection robot involves periodically acquiring dynamic video information of a target area. The target area can be the surrounding environment of the power equipment or the equipment itself. The target area scene is divided based on the dynamic video information. For example, videos of the target area can be captured under different lighting conditions (sunny or cloudy), different shooting angles (top or side view), and different times of day (day or night). Images of the main features of the target object are selected from the videos. These images may reflect the main structure of a power equipment or key areas of the equipment itself. The target area scene division includes at least two dimensions: different lighting conditions, different shooting angles, and different times of day. The selected images of the main features of the target object are acquired and fused with static photographs of the target object to obtain the target image. Through multi-dimensional processing of both dynamic video and static photographs, high-quality, high-definition, and high-fidelity target image information can be stored in a cloud database. Target images for different target areas correspond to different database functions Q. m While acquiring the target image, obtain the location information of the main subject L. m With the inspection robot's positioning information I m Database function Q m Location information L m and I m There is a one-to-one correspondence to obtain the location of the subject being photographed (e.g., power equipment). For example, this can be obtained by installing GPS (Global Positioning System) on the power equipment itself or by using relevant electronic map positioning information. Inspection robots are generally equipped with GPS devices, which transmit their own geographical location information during the inspection process. By establishing Lm and Im related location information in the database, a complete positioning information database can be formed. The location information is corrected by using the fixed positioning information of the equipment itself and the dynamic positioning information of the inspection robot, which improves the accuracy of the location information in the database. Then, the accurate historical location information of the relevant power equipment can be retrieved through the database function Qm. When an anomaly is detected during power inspection using the power inspection positioning and image construction method and system described in this application, the most recent time t is called. m The target image Gm is compared with the inspection anomaly image Bm. Before the comparison, the location information L of the main object is captured based on the positioning information of the inspection robot. m With the inspection robot's positioning information I m Traversing the cloud to form a historical database sequence L m = (L1, L2, ..., L n );I m = (I1, I2, ..., I n), find the corresponding L m With I m value through the database function Q m Call the corresponding target picture Gm, through the comparison of the target picture Gm and the inspection abnormal picture Bm, the target abnormal position accurate image is obtained, which can be, for example, the position of the abnormal and damaged power equipment, and the target loss key area image history database is formed in the cloud, through the formation of the database of the places prone to damage or abnormality, in the future inspection process, the relevant key area position can be focused on.

[0042] Further, according to the target loss key area image history database, the target loss area is divided, and in the future inspection process, when the key image of the relevant area is shot, the relevant loss area can be inspected, and the loss area image data is extracted for comparison, so that in some cases, the whole picture can be compared, and the relevant loss information can be efficiently mastered by comparing the key area, and at the same time, due to the demarcation of the loss area range, the focal length and shooting angle of the camera can be set according to the loss area range of the image in the actual inspection process of the robot, which reduces the manual remote operation focusing and angle adjustment (in order to clearly see the relevant area or manual operation for further viewing): the loss area range can also be obtained by formula calculation, assuming that the area of the target picture is Sm, and the area of the loss area is Sn, by recording the target abnormal position coordinate point C(x, y), the two target abnormal positions (X1, Y1) and (X2, Y2) farthest apart are demarcated as the diameter of the Sn area

[0043] So Sn=π*(R / 2) 2 At the same time, the center position K(X0, Y0) of the loss area is obtained, wherein

[0044] Further, by recording the target abnormal position coordinate point C(x, y), a sample clustering method is adopted, the sample set of the sample target abnormal position coordinate point is calculated, the area of the loss area Sn is obtained, and the function f(x) of the center position K of the loss area can also be calculated in the following way:

[0045]

[0046] , C is the input target abnormal position coordinate point in the model, C={C1, C2,...Ck}, D is the output target abnormal position coordinate point in the model, D={D1, D2,...Dk}.

[0047] As shown in Figure 2 As shown in Figure 2In a-2b, the calculation formula of the range of the vulnerable area is summarized by recording the target damage point distribution, (power equipment, etc. in the form of an image), recording target abnormal coordinate points C1, C2, C3, C4, …, and as the number of recorded coordinate points increases, the final damage point distribution will show a certain rule, such as Figure 2 In c, the recorded damage points increase to a certain extent, and the main abnormal position area shows a point-to-plane area rule distribution. At this time, in the position area where the damage points are concentrated, it is defined as Sn, and the two target abnormal positions CA(XA, YA) and CB(XB, YB) with the farthest distance are determined. At this time, the Sn area presents a circular shape, and the line connecting the two target abnormal positions CA(XA, YA) and CB(XB, YB) with the farthest distance can be regarded as the diameter of the Sn area. The center point is the center position K(X0, Y0) of the vulnerable area. For further reference Figure 2 d, the corresponding relationship between the entire area of the target picture Sm and the area of the vulnerable area Sn can be obtained intuitively, so that the circular area calculation formula can be used, for example, Sn = π * (R / 2)2, and the center position K(X0, Y0) of the vulnerable area is obtained, where K After obtaining the calculation formula of the vulnerable area, it can be substituted into the computer system model, and the sample clustering method is used to obtain the function f(x) of the vulnerable area area Sn and the vulnerable area center position K by calculating the sample set of the target abnormal position coordinate points.

[0048]

[0049] where C is the input target abnormal position coordinate point in the model, C = {C1, C2, … Ck}, and D is the output target abnormal position coordinate point in the model, D = {D1, D2, … Dk}. Thus, in the system, a target loss key area image history database is formed, and by forming a database of vulnerable or abnormal places, when shooting key images of related areas in the future inspection process, the magnified picture of the vulnerable area Sn can be directly shot according to the recorded vulnerable area Sn, which facilitates key comparison during inspection and improves work efficiency, such as the recent time t mWhen the target picture Gm is compared with the abnormal inspection picture Bm, the area comparison picture of the area Sn can be focused on, and in some cases, the entire picture does not need to be compared, which facilitates comparison of the key area. At the same time, due to the demarcation of the easy-wear area, the focal length and shooting angle of the camera can be set according to the easy-wear area of the image during the actual inspection of the robot, which avoids the need to repeatedly adjust the focal length and shooting angle of the camera when viewing the relevant area or needing manual operation to further view the specific area, thereby reducing the manual remote control operation of focusing and angle adjustment in some inspection cases.

[0050] In the third embodiment

[0051] Reference Figure 5 An electric power inspection robot positioning image construction device, a video acquisition module periodically acquires inspection dynamic video information of a target area; the target area scene is divided according to the inspection dynamic video information, and a shooting object main feature picture is selected from the video, and the target area scene division includes at least two dimensions of scene video of different light conditions, different camera angles, and different time points of the video shooting object main body;

[0052] A photo acquisition module periodically acquires a static photo of a shooting object main body in a target area; the static photo is obtained by detail sharpening enhancement and synthesis processing of multiple shooting photos under multiple exposure parameters,

[0053] A cloud processing module acquires the shooting object main feature picture selected from the video and the static photo of the shooting object main body, performs fusion processing to obtain a target picture, and different target pictures of different target areas correspond to different database functions Q m At the same time of acquiring the target picture, the position information L m of the shooting object main body and the inspection robot positioning information I m are acquired, the database function Q m and the position information L m and I m have a one-to-one correspondence, and according to the same target picture of different target areas in different time ranges [t1, t2], a historical database is formed in the cloud.

[0054] Further, when the inspection robot shoots the inspection dynamic video information of the target area and the static photo of the shooting object main body of the target area, the positioning information can be sent to other nearby inspection robots, and shooting assistance is required to go to the target area, for example Figure 3In the scenario shown, when the inspection robot 100 captures the inspection dynamic video information of the target area and the static photos of the target area shooting object, the positioning information can be sent to other nearby inspection robots 110 and 120, so as to capture dynamic videos and static photos of the target area shooting object from multiple angles and multiple collection subjects, provide multi-dimensional information verification, and improve the accuracy of picture information in the database.

[0055] Further, if an abnormal situation is found, the target picture abnormal position 2000 is identified, and the information thereof is uploaded as an abnormal position historical state identification. In the next inspection process, the target picture abnormal position 2000 is focused on for shooting. When the inspection robot 100 captures the inspection dynamic video information of the target area and the static photos of the target area shooting object, the positioning information can be sent to other nearby inspection robots 110 and 120, so as to capture dynamic videos and static photos of the target area shooting object from multiple angles and multiple collection subjects. In this way, the defect position clustering identification can be performed according to different components, and the clustering identification information at least includes the time, position information of the target picture shooting, and the defect type information of the defect position.

[0056] In the fourth embodiment

[0057] A power inspection robot positioning image construction method periodically acquires inspection dynamic video information of a target area. The target area can be an environment area around a power equipment or a power equipment area. The target area scene is divided according to the inspection dynamic video information. For example, the target area video under different light conditions such as sunny day and cloudy day, different shooting angles such as top view and side view, and different time such as day and night can be captured, and the shooting object characteristic picture is selected from the video. For example, the shooting object characteristic picture reflects the main structure of a certain power equipment or the key area of the power equipment itself. The target area scene division at least includes two-dimensional scene videos in different light conditions, different camera angles, and different time of the video shooting object.

[0058] Periodically acquire static photos of the target area shooting object. The static photos are obtained by detail sharpening enhancement synthesis processing of multiple shooting photos under multiple exposure parameters. For example, the video shooting can be performed at the same time as the photo shooting. The final subject static photo is obtained by fusion processing of multiple photos. The purpose is to obtain high-definition static pictures.

[0059] Reference Figure 4, further, if the abnormal situation is found, the collected target picture abnormal position is identified, and the information is uploaded as the abnormal position history state identification, forming the power equipment different component identification information library, so as to carry out the defect part clustering identification, and the clustering identification information at least includes the time, position information and defect type information of the target picture shooting, such as Figure 4 As shown in FIG. 8, through the multiple inspection process, it is found that the main vulnerable components of the power equipment include the vulnerable components 200, the components 300 and the components 400, then in the future inspection process, the target vulnerable components 200, the components 300 and the components 400 of the power equipment are inspected dynamically, the video information and the static photo of the target area shooting object main body, and the defect information of the vulnerable components 200, the components 300 and the components 400 at different time is recorded, including the further subdivided component defect position, so that the image information of the vulnerable components 200, the components 300 and the components 400 in the target picture is focused on in the next shooting process, and the vulnerable area area Sn of the vulnerable components 200, the components 300 and the components 400 is formed, and further, the vulnerable component information database of different types of power grid equipment can be formed, for example, it can be obtained by the above method that the most vulnerable components of large power transformer are: porcelain bushing, explosion-proof membrane and other positions, accordingly, the existing vulnerable component information database of power grid equipment can also be used to synchronize the power inspection robot, so that it focuses on shooting the related vulnerable components according to the database information feedback in the process of inspection and shooting.

[0060] Those skilled in the art can realize the method steps and units described in connection with the embodiments disclosed herein can be realized by electronic hardware, computer software, or a combination of both. The steps and units of the embodiments described above are generally described in the above description in terms of functional generalities. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application. The methods or steps described in connection with the embodiments disclosed herein can be implemented by hardware, a software program executed by a processor, or a combination of both. The software program can be stored in a storage medium such as a Random Access Memory (RAM), a memory, a Read-Only Memory (ROM), an Electrically Programmable Read-Only Memory (EPROM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), a register, a hard disk, or a removable disk.

[0061] Moreover, other implementations of the application will be apparent from the disclosure, and the examples shown herein should be considered to be completely illustrative only, and not restrictive of the present application, as the true scope and spirit of the application is indicated by the appended claims.

Claims

1. A method for constructing a positioning image of a power inspection robot, characterized in that, periodically acquiring inspection dynamic video information of a target area; dividing a scene of the target area according to the inspection dynamic video information, and selecting a subject characteristic picture of a shooting object from the video, the scene division of the target area including at least two-dimensional scene videos of different light conditions, different camera angles, and different time points of the shooting object in the video; periodically acquiring a static photo of a shooting object of the target area; the static photo being obtained by detail sharpening enhancement and composition processing of multiple shooting photos under multiple exposure parameters, Acquire the selected shooting object body characteristic picture in the video and the static photo of the shooting object body, perform fusion processing to obtain a target picture, and different target regions of the target picture correspond to different database functions Q m , Acquire the target picture, acquire the location information L m of the shooting object body at the same time m , the database function Q m and the location information L m and I m have a one-to-one correspondence, according to the same target picture of different target regions in different time ranges [t1, t2], form a historical database in the cloud, when the inspection finds an anomaly, call the target picture Gm of the nearest time t m and the inspection abnormal picture Bm for comparison, before comparison, according to the location information of the shooting object body L m and the positioning information of the inspection robot I m, , traverse the historical database sequence L m formed in the cloud = (L1, L2,..., L n ); I m = (I1, I2,..., I n ), find the corresponding L m and I m value through the database function Q m , call the corresponding target picture Gm, through the comparison of the target picture Gm and the inspection abnormal picture Bm, obtain the target abnormal position accurate image, and form the target loss key region image historical database in the cloud.

2. The image construction method of claim 1, wherein, According to the target loss focus area image history database, the target vulnerable area is divided: the area of the target picture is Sm, the area of the vulnerable area is Sn, the coordinates of the target abnormal position C(x, y) are recorded, and the two farthest target abnormal positions (X1, Y1) and (X2, Y2) are determined as the diameter of the Sn area Thus Sn = π * (R / 2) 2 At the same time, the center position K(X0, Y0) of the vulnerable area is obtained, wherein K 3. The image construction method according to claim 1 or 2, characterized by, when the inspection robot shoots the inspection dynamic video information of the target area and the static photo of the shooting object of the target area, positioning information can be sent to other nearby inspection robots and asked to shoot and assist in the target area.

4. The image construction method according to claim 1 or 2, characterized by, each abnormal position in the historical state of the target picture is identified, and the shooting object in the picture is clustered and identified according to different components, and the clustering identification information includes at least time, position information of the shooting of the target picture, and defect type information of the defect part. 5.A device for constructing a positioning image of a power inspection robot, characterized in that, a video acquisition module periodically acquires inspection dynamic video information of a target area; divides a scene of the target area according to the inspection dynamic video information, and selects a subject characteristic picture of a shooting object from the video, the scene division of the target area including at least two-dimensional scene videos of different light conditions, different camera angles, and different time points of the shooting object in the video; a photo acquisition module periodically acquires a static photo of a shooting object of the target area; the static photo being obtained by detail sharpening enhancement and composition processing of multiple shooting photos under multiple exposure parameters, The cloud processing module acquires a selected image of the subject's characteristics and a still photo of the subject from the video, and performs a fusion process to obtain the target image. Target images for different target areas correspond to different database functions Q. m While acquiring the target image, obtain the location information of the main subject L. m With the inspection robot's positioning information I m Database function Q m Location information L m and I m There is a one-to-one correspondence. Based on receiving images of the same target in different target areas within different time ranges [t1, t2], a historical database is formed in the cloud. When an anomaly is detected during inspection, the most recent time t is called. m The target image Gm is compared with the inspection anomaly image Bm. Before the comparison, the location information L of the main object is captured based on the positioning information of the inspection robot. m With the inspection robot's positioning information I m, Traversing the cloud to form a historical database sequence L m = (L1, L2, ..., L n );I m = (I1, I2, ..., I n Find the corresponding L. m with I m The value is obtained through the database function Q. m The corresponding target image Gm is retrieved. By comparing the target image Gm with the inspection anomaly image Bm, the precise image of the target anomaly location is obtained, and a historical database of images of key areas of target loss is formed in the cloud.

6. An electronic device, comprising: comprising: a processor; and a memory storing programs, the programs including instructions that, when executed by the processor, cause the processor to perform the power inspection robot positioning image construction method according to any one of claims 1-4.

7. A computer readable storage medium storing a program, characterized by The programs include instructions that, when executed by the processor of the electronic device, cause the electronic device to perform the power inspection robot positioning image construction method according to any one of claims 1-4.

Images