Unmanned aerial vehicle completion acceptance material statistical method and system based on Kalman filtering

Abstract

The invention discloses an unmanned aerial vehicle completion acceptance material statistical method and system based on Kalman filtering, and the method comprises the steps: carrying out the collection of an acceptance video of a target tower based on an unmanned aerial vehicle, and recording the position information of the position of the unmanned aerial vehicle; determining an identification frame sequence containing a plurality of video frame data based on the acceptance video; and traversing the recognition frame sequence, estimating the number of the target type of materials according to the video frame data and the corresponding position information based on Kalman filtering, and determining a material estimation value. According to the unmanned aerial vehicle completion acceptance work scene of the overhead distribution line, a video real-time target detection technology is utilized to identify the target material to be counted, and material statistics is performed in combination with the position information of the unmanned aerial vehicle, so that an effective support means can be provided for overhead distribution line engineering construction workload statistics and equipment ledger management.

Description

Technical Field

[0001] The present invention relates to the technical field of overhead power distribution line project acceptance, and more specifically, to a Kalman filtering-based unmanned aerial vehicle completion acceptance material statistics method and system. Background Art

[0002] For new distribution network construction projects, the traditional acceptance method is for personnel to count materials and check construction processes on each tower. This not only requires climbing towers for high-altitude operations, but also has problems such as low efficiency of manual counting, which can no longer meet the needs of new power system construction. With the large-scale application of drones in the power industry, drones equipped with visible light imaging equipment have carried out daily inspections of overhead transmission and distribution lines, achieving good results. With the construction of a strong power grid, the length of newly built transmission and distribution lines has increased year by year. The acceptance and commissioning of the expansion and renovation of new lines is an important part of power grid construction.

[0003] The core of the final acceptance is the material statistics, that is, to check the usage of various materials during the construction process, so as to facilitate the subsequent project settlement. If a single shot is used, when the tower to be accepted and counted in three-dimensional space is photographed by a drone, the image will be compressed to a two-dimensional plane, and the materials on the other side of the tower will often be blocked and some materials will be omitted from the statistics. If two or more shots are used, the same target material will appear in multiple images, resulting in repeated material statistics.

[0004] Therefore, it is urgent to propose a material statistics technology for the completion acceptance of overhead distribution lines by drones to effectively improve the accuracy of material statistics. Summary of the invention

[0005] The present invention proposes a material statistics method and system for unmanned aerial vehicle completion acceptance based on Kalman filtering to solve the problem of how to perform material statistics efficiently and accurately.

[0006] In order to solve the above problem, according to one aspect of the present invention, a material statistics method for completion acceptance of an unmanned aerial vehicle based on Kalman filtering is provided, and the method comprises:

[0007] Collect the acceptance video of the target tower based on the drone, and record the location information of the drone;

[0008] Determining an identification frame sequence including a plurality of video frame data based on the acceptance video;

[0009] The recognition frame sequence is traversed, and based on Kalman filtering, the quantity of target type materials is estimated according to the video frame data and the corresponding position information to determine the material estimation value.

[0010] Preferably, the traversing of the recognition frame sequence, estimating the quantity of target type materials based on the video frame data and the corresponding position information, and determining the material estimation value comprises:

[0011] S31, initialization t=1, setting the initial value of material estimation and the initial value of estimation error;

[0012] S32, detecting the quantity of target type materials in the t-th video frame data in the recognition frame sequence, and obtaining the material measurement value;

[0013] S33, determining the confidence level based on the position information of the drone when collecting the t-th video frame data;

[0014] S34, determining a measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data;

[0015] S35, calculating a gain coefficient based on a measurement value error corresponding to the t-th video frame data and an estimated value error corresponding to the t-1-th video frame data;

[0016] S36, calculating the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data;

[0017] S37, calculating the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data;

[0018] S38, if the current estimated value error is less than or equal to the preset threshold, the current material estimated value is determined to be the final material estimated value, and the process ends; if the current estimated value error is greater than the preset threshold, t=t+1 is updated, and the process returns to step S32 for recalculation;

[0019] Among them, when t=1, the material estimation value corresponding to the t-1th video frame data is the material estimation initial value, and the estimation value error corresponding to the t-1th video frame data is the estimation value error initial value.

[0020] Preferably, the step of determining the confidence level based on the position information of the drone when collecting the t-th video frame data comprises:

[0021] An initial confidence level is set, a normal angle α is determined according to the position information, and a confidence level is determined by increasing a first preset percentage based on the initial confidence level every time the normal angle increases by a first preset degree;

[0022] When the confidence level is greater than or equal to the second preset percentage, it is determined that the material actually exists, and the measurement value error is adjusted at the same time.

[0023] Preferably, the step of determining the measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data comprises:

[0024] E c (t) = (1-η) * N c (t),

[0025] Among them, E c (t) is the measurement error corresponding to the t-th video frame data; η is the confidence level; N c (t) is the material measurement value corresponding to the tth video frame data.

[0026] Preferably, the step of calculating the gain coefficient based on the measurement value error corresponding to the t-th video frame data and the estimation value error corresponding to the t-1-th video frame data includes:

[0027]

[0028] Wherein, K(t) is the gain coefficient corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; E c (t) is the measurement value error corresponding to the t-th video frame data.

[0029] Preferably, the step of calculating the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data comprises:

[0030] N g (t) = N g (t-1)+K(t)×[N c (t)-N g (t-1)],

[0031] Among them, N g (t) is the material estimation value corresponding to the t-th video frame data; N g (t-1) is the material estimation value corresponding to the t-1th video frame data; N c (t) is the material measurement value corresponding to the t-th video frame data; K(t) is the gain coefficient corresponding to the t-th video frame data.

[0032] Preferably, the step of calculating the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data comprises:

[0033] E g (t)=(1-K(t))*E g(t-1),

[0034] Among them, E g (t) is the estimated value error corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; K(t) is the gain coefficient corresponding to the tth video frame data.

[0035] According to another aspect of the present invention, a UAV completion acceptance material statistics system based on Kalman filtering is provided, and the system comprises:

[0036] A data acquisition unit, used to collect the acceptance video of the target tower based on the drone, and record the location information of the drone;

[0037] An identification frame sequence determining unit, configured to determine an identification frame sequence including a plurality of video frame data based on the acceptance video;

[0038] The material estimation value determination unit is used to traverse the identification frame sequence, estimate the quantity of the target type of material based on the video frame data and the corresponding position information based on the Kalman filter, and determine the material estimation value.

[0039] Preferably, the material estimation value determination unit traverses the recognition frame sequence, estimates the quantity of target type materials based on the video frame data and the corresponding position information based on the Kalman filter, and determines the material estimation value, including:

[0040] S31, initialization t=1, setting the initial value of material estimation and the initial value of estimation error;

[0041] S32, detecting the quantity of target type materials in the t-th video frame data in the recognition frame sequence, and obtaining the material measurement value;

[0042] S33, determining the confidence level based on the position information of the drone when collecting the t-th video frame data;

[0043] S34, determining a measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data;

[0044] S35, calculating a gain coefficient based on a measurement value error corresponding to the t-th video frame data and an estimated value error corresponding to the t-1-th video frame data;

[0045] S36, calculating the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data;

[0046] S37, calculating the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data;

[0047] S38, if the current estimated value error is less than or equal to the preset threshold, the current material estimated value is determined to be the final material estimated value, and the process ends; if the current estimated value error is greater than the preset threshold, t=t+1 is updated, and the process returns to step S32 for recalculation;

[0048] Among them, when t=1, the material estimation value corresponding to the t-1th video frame data is the material estimation initial value, and the estimation value error corresponding to the t-1th video frame data is the estimation value error initial value.

[0049] Preferably, the material estimation value determination unit determines the confidence level based on the position information of the drone when collecting the t-th video frame data, including:

[0050] An initial confidence level is set, a normal angle α is determined according to the position information, and a confidence level is determined by increasing a first preset percentage based on the initial confidence level every time the normal angle increases by a first preset degree;

[0051] When the confidence level is greater than or equal to the second preset percentage, it is determined that the material actually exists, and the measurement value error is adjusted at the same time.

[0052] Preferably, the material estimation value determination unit determines the measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data, including:

[0053] E c (t) = (1-η) * N c (t),

[0054] Among them, E c (t) is the measurement error corresponding to the t-th video frame data; η is the confidence level; N c (t) is the material measurement value corresponding to the tth video frame data.

[0055] Preferably, the material estimated value determination unit calculates the gain coefficient based on the measurement value error corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data, including:

[0056]

[0057] Wherein, K(t) is the gain coefficient corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; E c (t) is the measurement value error corresponding to the t-th video frame data.

[0058] Preferably, the material estimation value determination unit calculates the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data, including:

[0059] N g (t) = N g (t-1)+K(t)×[N c (t)-N g (t-1)],

[0060] Among them, N g (t) is the material estimation value corresponding to the t-th video frame data; N g (t-1) is the material estimation value corresponding to the t-1th video frame data; N c (t) is the material measurement value corresponding to the t-th video frame data; K(t) is the gain coefficient corresponding to the t-th video frame data.

[0061] Preferably, the material estimated value determining unit calculates the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data, including:

[0062] E g (t)=(1-K(t))*E g (t-1),

[0063] Among them, E g (t) is the estimated value error corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; K(t) is the gain coefficient corresponding to the tth video frame data.

[0064] The present invention provides a method and system for counting materials for completion acceptance of unmanned aerial vehicles based on Kalman filtering, including: collecting acceptance videos of target poles and towers based on unmanned aerial vehicles, and recording the location information of the unmanned aerial vehicles at the same time; determining an identification frame sequence containing multiple video frame data based on the acceptance video; traversing the identification frame sequence, estimating the quantity of target type materials based on the video frame data and the corresponding location information based on Kalman filtering, and determining the material estimation value. The present invention is aimed at the completion acceptance operation scenario of unmanned aerial vehicles for overhead distribution lines, and uses video real-time target detection technology to identify the target materials to be counted, and at the same time combines the location information of the unmanned aerial vehicles to count the materials, which can provide effective support means for the construction workload statistics and equipment ledger management of overhead distribution line projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] A more complete understanding of exemplary embodiments of the present invention may be obtained by referring to the following drawings:

[0066] Figure 1 It is a flow chart of a material statistics method 100 for completion acceptance of an unmanned aerial vehicle based on Kalman filtering according to an embodiment of the present invention;

[0067] Figure 2 A schematic diagram of video collection for completion acceptance of an empty power distribution line by a drone according to an embodiment of the present invention;

[0068] Figure 3 A schematic diagram of an imaging angle for completion acceptance of an overhead power distribution line by a drone according to an embodiment of the present invention;

[0069] Figure 4 4 is a schematic diagram of the structure of a UAV completion acceptance material statistics system 400 based on Kalman filtering according to an embodiment of the present invention. DETAILED DESCRIPTION

[0070] Now, exemplary embodiments of the present invention are described with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely and to fully convey the scope of the present invention to those skilled in the art. The terms used in the exemplary embodiments shown in the accompanying drawings are not intended to limit the present invention. In the accompanying drawings, the same units / elements are marked with the same reference numerals.

[0071] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0072] Figure 1 FIG. 1 is a flow chart of a material statistics method 100 for UAV completion acceptance based on Kalman filtering according to an embodiment of the present invention. Figure 1 As shown, the material statistics method for drone completion acceptance based on Kalman filtering provided by the embodiment of the present invention is aimed at the drone completion acceptance operation scenario of overhead distribution lines, uses video real-time target detection technology to identify the target materials to be counted, and combines the location information of the drone to count the materials, which can provide effective support for the workload statistics and equipment ledger management of overhead distribution line engineering construction. The material statistics method 100 for drone completion acceptance based on Kalman filtering provided by the embodiment of the present invention starts from step 101. In step 101, the acceptance video of the target tower is collected based on the drone, and the location information of the drone is recorded at the same time.

[0073] Combination Figure 2 As shown, in the present invention, a drone with high-precision positioning function is controlled to fly 180° around the tower slowly and steadily from any position to collect video data. There should be no trees or other obstacles blocking the view within the selected flight range. The visible light acquisition mount is turned on to collect the video of the tower to be accepted, and the video is synchronously recognized and calculated using the onboard front-end computing power or edge computing power.

[0074] In the present invention, the high-precision position information of the drone when the target material is located in the identification frame is recorded, including: positioning data and pan / tilt rotation angle information, and the imaging surface of the visible light video acquisition device and its normal angle α are calculated, such as Figure 3 The figure shows a schematic diagram of the imaging angle of the drone completion acceptance of overhead distribution lines when looking down.

[0075] In step 102, an identification frame sequence including a plurality of video frame data is determined based on the acceptance video.

[0076] In the present invention, a plurality of identification frame data are determined based on the collected acceptance video, and an identification frame sequence is formed to perform material detection frame by frame.

[0077] In step 103, the recognition frame sequence is traversed, and based on Kalman filtering, the quantity of target type materials is estimated according to the video frame data and the corresponding position information to determine the material estimation value.

[0078] Preferably, the traversing of the recognition frame sequence, estimating the quantity of target type materials based on the video frame data and the corresponding position information, and determining the material estimation value comprises:

[0079] S31, initialization t=1, setting the initial value of material estimation and the initial value of estimation error;

[0080] S32, detecting the quantity of target type materials in the t-th video frame data in the recognition frame sequence, and obtaining the material measurement value;

[0081] S33, determining the confidence level based on the position information of the drone when collecting the t-th video frame data;

[0082] S34, determining a measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data;

[0083] S35, calculating a gain coefficient based on a measurement value error corresponding to the t-th video frame data and an estimated value error corresponding to the t-1-th video frame data;

[0084] S36, calculating the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data;

[0085] S37, calculating the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data;

[0086] S38, if the current estimated value error is less than or equal to the preset threshold, the current material estimated value is determined to be the final material estimated value, and the process ends; if the current estimated value error is greater than the preset threshold, t=t+1 is updated, and the process returns to step S32 for recalculation;

[0087] Among them, when t=1, the material estimation value corresponding to the t-1th video frame data is the material estimation initial value, and the estimation value error corresponding to the t-1th video frame data is the estimation value error initial value.

[0088] Preferably, the step of determining the confidence level based on the position information of the drone when collecting the t-th video frame data comprises:

[0089] An initial confidence level is set, a normal angle α is determined according to the position information, and a confidence level is determined by increasing a first preset percentage based on the initial confidence level every time the normal angle increases by a first preset degree;

[0090] When the confidence level is greater than or equal to the second preset percentage, it is determined that the material actually exists, and the measurement value error is adjusted at the same time.

[0091] Preferably, the step of determining the measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data comprises:

[0092] E c (t) = (1-η) * N c (t),

[0093] Among them, E c (t) is the measurement error corresponding to the t-th video frame data; η is the confidence level; N c (t) is the material measurement value corresponding to the tth video frame data.

[0094] Preferably, the step of calculating the gain coefficient based on the measurement value error corresponding to the t-th video frame data and the estimation value error corresponding to the t-1-th video frame data includes:

[0095]

[0096] Wherein, K(t) is the gain coefficient corresponding to the t-th video frame data; Eg (t-1) is the estimated value error corresponding to the t-1th video frame data; E c (t) is the measurement value error corresponding to the t-th video frame data.

[0097] Preferably, the step of calculating the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data comprises:

[0098] N g (t) = N g (t-1)+K(t)×[N c (t)-N g (t-1)],

[0099] Among them, N g (t) is the material estimation value corresponding to the t-th video frame data; N g (t-1) is the material estimation value corresponding to the t-1th video frame data; N c (t) is the material measurement value corresponding to the t-th video frame data; K(t) is the gain coefficient corresponding to the t-th video frame data.

[0100] Preferably, the step of calculating the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data comprises:

[0101] E g (t)=(1-K(t))*E g (t-1),

[0102] Among them, E g (t) is the estimated value error corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; K(t) is the gain coefficient corresponding to the tth video frame data.

[0103] In the present invention, based on the YOLO v5 architecture, optimization and customization are performed to form an efficient and lightweight target detection algorithm model, which can be deployed on the front end of the drone or the handheld remote control to achieve real-time and accurate identification of multiple key materials, thereby determining the measured value of the material quantity. Among them, the types of materials include but are not limited to transformers, insulators, wire clamps, lightning arresters, fuses, and pole tower crossarms. Through sparse matrix or vector technology, the size and computational complexity of the model are reduced, so that the algorithm model can adapt to the environment with limited front-end computing power, ensuring that the ideal frame rate performance can be achieved on hardware platforms with different computing capabilities, thereby not affecting the detection efficiency and quality.

[0104] In order to further enhance the practical value of the target detection results and facilitate the analysis of the material statistics module, drone technology is integrated. The material target detection module dynamically calls the precise rotation angle information of the gimbal carried by the drone, and combines the high-precision positioning data of the drone to record the high-precision coordinate position of the drone corresponding to each target material detection frame during the entire material statistics process, and maps it in three-dimensional space.

[0105] In the present invention, the real-time quantity of the target detection material in each recognition frame is counted as a measurement value.

[0106] Since the deviation of the target material between adjacent recognition frames in the image is small, the repeatability of the target detected material can be judged by the deviation of the target material between multiple adjacent recognition frames. When the pixel position of the target material in the image of adjacent recognition frames changes little, it is considered that the material is the same as the target material of the previous recognition frame and no repeated statistics are performed.

[0107] The present invention is based on the UAV completion acceptance material statistics method based on the Kalman filter principle. The material target detection module and the material statistics module complete the quantity measurement value N of the target type material of the video acquisition frame. c (t) is calculated, and according to the current measurement value and its confidence, the material estimation value N is continuously iterated through the Kalman filter principle statistical module g (t).

[0108] In the present invention, according to the application characteristics of the on-site completion acceptance of the overhead distribution line UAV, the material statistics optimal estimation equation is constructed as follows:

[0109]

[0110] Among them, N g (t) represents the quantity measurement value of the target type material corresponding to the t-th video frame data; N g (t-1) is the material estimation value corresponding to the t-1th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; N c (t) is the material measurement value corresponding to the t-th video frame data; E c (t) is the measurement value error corresponding to the t-th video frame data.

[0111] For the single-base tower being accepted, when the tower type is determined, the quantity range of its specific materials is determined, so the initial value of the estimated value and the initial value of the estimated value error can be obtained based on prior experience.

[0112] The gain coefficient K in the Kalman filter system is:

[0113]

[0114] Therefore, formula (1) can be rewritten as:

[0115] E g (t)=(1-K(t))*E g (t-1) (3)

[0116] N g (t) = N g (t-1)+K(t)×[N c (t)-N g (t-1)] (4)

[0117] In the present invention, the measured value of the number of target type materials is calculated by the target detection module and the material statistics module; the measurement value error E c (t) can be calculated based on the confidence level η, as shown in formula (6):

[0118] E c (t) = (1-η) * N c (t) (6)

[0119] Therefore, in the present invention, the specific process of determining the estimated value of the material includes:

[0120] S31, initialization t=1, setting the initial value of material estimation and the initial value of estimation error;

[0121] S32, detecting the number of target type materials in the t-th video frame data in the recognition frame sequence, and obtaining the material measurement value N c (t);

[0122] S33, determining the confidence η based on the position information of the drone when collecting the tth video frame data;

[0123] S34, determining the measurement error E based on the material measurement value and confidence level corresponding to the t-th video frame data c (t);

[0124] S35, calculating a gain coefficient K(t) based on a measurement value error corresponding to the t-th video frame data and an estimated value error corresponding to the t-1-th video frame data;

[0125] S36, calculating the material estimation value N corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data g (t);

[0126] S37, calculating the estimated value error E corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame datag (t);

[0127] S38, if the current estimated value error is less than or equal to the preset threshold, the current material estimated value is determined to be the final material estimated value, and the process ends; if the current estimated value error is greater than the preset threshold, t=t+1 is updated, and the process returns to step S32 for recalculation;

[0128] Among them, when t=1, the material estimation value corresponding to the t-1th video frame data is the material estimation initial value, and the estimation value error corresponding to the t-1th video frame data is the estimation value error initial value.

[0129] In the present invention, the initial confidence level can be set to 50%, the first preset degree to 2°, the first preset percentage to 1%, and the second preset percentage to 90%. That is, when the target material is identified by the target detection module, its initial confidence level is set to 50%, the change of the normal angle α is recorded, and the confidence level increases by 1% for every 2° increase in the angle. When the confidence level is greater than or equal to 90%, it is considered that the material really exists, and the measurement value error is adjusted at the same time.

[0130] The method of the present invention estimates the quantity of target materials through past estimated values ​​and their errors, current measured values ​​and their errors, and can improve the accuracy of material statistics for overhead distribution lines. The confidence level is determined based on the position information of the drone, which can effectively avoid the problem of inaccurate target material statistics due to line of sight obstruction and recurrence after obstruction, and effectively improve the accuracy of material statistics.

[0131] Figure 4 FIG. 4 is a schematic diagram of a Kalman filter-based UAV completion acceptance material statistics system 400 according to an embodiment of the present invention. Figure 4 As shown, the UAV completion acceptance material statistics system 400 based on Kalman filtering provided in an embodiment of the present invention includes: a data acquisition unit 401, an identification frame sequence determination unit 402 and a material estimation value determination unit 403.

[0132] Preferably, the data acquisition unit 401 is used to collect the acceptance video of the target tower based on the drone, and simultaneously record the location information of the drone.

[0133] Preferably, the recognition frame sequence determining unit 402 is used to determine a recognition frame sequence including a plurality of video frame data based on the acceptance video.

[0134] Preferably, the material estimation value determination unit 403 is used to traverse the identification frame sequence, estimate the quantity of the target type of material based on the video frame data and the corresponding position information based on the Kalman filter, and determine the material estimation value.

[0135] Preferably, the material estimation value determination unit 403 traverses the recognition frame sequence, estimates the quantity of target type materials based on the video frame data and the corresponding position information based on the Kalman filter, and determines the material estimation value, including:

[0136] S31, initialization t=1, setting the initial value of material estimation and the initial value of estimation error;

[0137] S32, detecting the quantity of target type materials in the t-th video frame data in the recognition frame sequence, and obtaining the material measurement value;

[0138] S33, determining the confidence level based on the position information of the drone when collecting the t-th video frame data;

[0139] S34, determining a measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data;

[0140] S35, calculating a gain coefficient based on a measurement value error corresponding to the t-th video frame data and an estimated value error corresponding to the t-1-th video frame data;

[0141] S36, calculating the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data;

[0142] S37, calculating the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data;

[0143] S38, if the current estimated value error is less than or equal to the preset threshold, the current material estimated value is determined to be the final material estimated value, and the process ends; if the current estimated value error is greater than the preset threshold, t=t+1 is updated, and the process returns to step S32 for recalculation;

[0144] Among them, when t=1, the material estimation value corresponding to the t-1th video frame data is the material estimation initial value, and the estimation value error corresponding to the t-1th video frame data is the estimation value error initial value.

[0145] Preferably, the material estimation value determination unit determines the confidence level based on the position information of the drone when collecting the t-th video frame data, including:

[0146] An initial confidence level is set, a normal angle α is determined according to the position information, and a confidence level is determined by increasing the initial confidence level by a first preset percentage every time the normal angle increases by a first preset degree;

[0147] When the confidence level is greater than or equal to the second preset percentage, it is determined that the material actually exists, and the measurement value error is adjusted at the same time.

[0148] Preferably, the material estimation value determining unit 403 determines the measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data, including:

[0149] E c (t) = (1-η) * N c (t),

[0150] Among them, E c (t) is the measurement error corresponding to the t-th video frame data; η is the confidence level; N c (t) is the material measurement value corresponding to the tth video frame data.

[0151] Preferably, the material estimated value determining unit 403 calculates the gain coefficient based on the measurement value error corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data, including:

[0152]

[0153] Wherein, K(t) is the gain coefficient corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; E c (t) is the measurement value error corresponding to the t-th video frame data.

[0154] Preferably, the material estimation value determining unit 403 calculates the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data, including:

[0155] N g (t) = N g (t-1)+K(t)×[N c (t)-N g (t-1)],

[0156] Among them, N g (t) is the material estimation value corresponding to the t-th video frame data; N g (t-1) is the material estimation value corresponding to the t-1th video frame data; N c (t) is the material measurement value corresponding to the t-th video frame data; K(t) is the gain coefficient corresponding to the t-th video frame data.

[0157] Preferably, the material estimated value determining unit 403 calculates the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data, including:

[0158] E g (t)=(1-K(t))*E g (t-1),

[0159] Among them, E g (t) is the estimated value error corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; K(t) is the gain coefficient corresponding to the tth video frame data.

[0160] The UAV completion acceptance material statistics system 400 based on Kalman filtering in an embodiment of the present invention corresponds to the UAV completion acceptance material statistics method 100 based on Kalman filtering in another embodiment of the present invention, which will not be repeated here.

[0161] The invention has been described above with reference to a few embodiments. However, it is readily apparent to a person skilled in the art that other embodiments than the ones disclosed above are equally within the scope of the invention, as defined by the appended patent claims.

[0162] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise therein. All references to "a / said / the [means, components, etc.]" are to be openly interpreted as at least one instance of said means, components, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not necessarily have to be performed in the exact order disclosed, unless explicitly stated otherwise.

[0163] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0164] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0165] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0166] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0167] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A material statistics method for completion acceptance of unmanned aerial vehicles based on Kalman filtering, characterized in that: The method comprises: Collect the acceptance video of the target tower based on the drone, and record the location information of the drone; Determining an identification frame sequence including a plurality of video frame data based on the acceptance video; The recognition frame sequence is traversed, and based on Kalman filtering, the quantity of target type materials is estimated according to the video frame data and the corresponding position information to determine the material estimation value.

2. The method according to claim 1, characterized in that The traversing the recognition frame sequence, estimating the quantity of target type materials according to the video frame data and the corresponding position information based on Kalman filtering, and determining the material estimation value includes: S31, initialization t=1, setting the initial value of material estimation and the initial value of estimation error; S32, detecting the quantity of target type materials in the t-th video frame data in the recognition frame sequence, and obtaining the material measurement value; S33, determining the confidence level based on the position information of the drone when collecting the t-th video frame data; S34, determining a measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data; S35, calculating a gain coefficient based on a measurement value error corresponding to the t-th video frame data and an estimated value error corresponding to the t-1-th video frame data; S36, calculating the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data; S37, calculating the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data; S38, if the current estimated value error is less than or equal to the preset threshold, the current material estimated value is determined to be the final material estimated value, and the process ends; if the current estimated value error is greater than the preset threshold, t=t+1 is updated, and the process returns to step S32 for recalculation; Among them, when t=1, the material estimation value corresponding to the t-1th video frame data is the material estimation initial value, and the estimation value error corresponding to the t-1th video frame data is the estimation value error initial value.

3. The method according to claim 2, characterized in that The step of determining the confidence level based on the position information of the drone when collecting the t-th video frame data includes: An initial confidence level is set, a normal angle α is determined according to the position information, and a confidence level is determined by increasing a first preset percentage based on the initial confidence level every time the normal angle increases by a first preset degree; When the confidence level is greater than or equal to the second preset percentage, it is determined that the material actually exists, and the measurement value error is adjusted at the same time.

4. The method according to claim 2, characterized in that: The determining of the measurement value error based on the material measurement value and the confidence level corresponding to the t-th video frame data includes: E c (t)=(1-η)*N c (t), Among them, E c (t) is the measurement error corresponding to the t-th video frame data; η is the confidence level; N c (t) is the material measurement value corresponding to the tth video frame data.

5. The method according to claim 2, characterized in that: The step of calculating the gain coefficient based on the measurement value error corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data includes: Wherein, K(t) is the gain coefficient corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; E c (t) is the measurement value error corresponding to the t-th video frame data.

6. The method according to claim 2, characterized in that The calculating the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data includes: N g (t)=N g (t-1)+K(t)×[N c (t)-N g (t-1)], Among them, N g (t) is the material estimation value corresponding to the t-th video frame data; N g (t-1) is the material estimation value corresponding to the t-1th video frame data; N c (t) is the material measurement value corresponding to the t-th video frame data; K(t) is the gain coefficient corresponding to the t-th video frame data.

7. The method according to claim 2, characterized in that The calculating the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data includes: E g (t)=(1-K(t))*E g (t-1), Among them, E g (t) is the estimated value error corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; K(t) is the gain coefficient corresponding to the tth video frame data.

8. A UAV completion acceptance material statistics system based on Kalman filtering, characterized in that: The system comprises: A data acquisition unit, used to collect the acceptance video of the target tower based on the drone, and record the location information of the drone; An identification frame sequence determining unit, configured to determine an identification frame sequence including a plurality of video frame data based on the acceptance video; The material estimation value determination unit is used to traverse the identification frame sequence, estimate the quantity of the target type of material based on the video frame data and the corresponding position information based on the Kalman filter, and determine the material estimation value.

9. The system according to claim 8, characterized in that The material estimation value determination unit traverses the recognition frame sequence, estimates the quantity of target type materials according to the video frame data and the corresponding position information based on Kalman filtering, and determines the material estimation value, including: S31, initialization t=1, setting the initial value of material estimation and the initial value of estimation error; S32, detecting the quantity of target type materials in the t-th video frame data in the recognition frame sequence, and obtaining the material measurement value; S33, determining the confidence level based on the position information of the drone when collecting the t-th video frame data; S34, determining a measurement value error based on the material measurement value and confidence level corresponding to the t-th video frame data; S35, calculating a gain coefficient based on a measurement value error corresponding to the t-th video frame data and an estimated value error corresponding to the t-1-th video frame data; S36, calculating the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data; S37, calculating the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data; S38, if the current estimated value error is less than or equal to the preset threshold, the current material estimated value is determined to be the final material estimated value, and the process ends; if the current estimated value error is greater than the preset threshold, t=t+1 is updated, and the process returns to step S32 for recalculation; Among them, when t=1, the material estimation value corresponding to the t-1th video frame data is the material estimation initial value, and the estimation value error corresponding to the t-1th video frame data is the estimation value error initial value.

10. The system according to claim 9, characterized in that The material estimation value determination unit determines the confidence based on the position information of the drone when collecting the t-th video frame data, including: An initial confidence level is set, a normal angle α is determined according to the position information, and a confidence level is determined by increasing a first preset percentage based on the initial confidence level every time the normal angle increases by a first preset degree; When the confidence level is greater than or equal to the second preset percentage, it is determined that the material actually exists, and the measurement value error is adjusted at the same time.

11. The system according to claim 9, characterized in that The material estimation value determination unit determines the measurement value error based on the material measurement value and the confidence level corresponding to the t-th video frame data, including: E c (t)=(1-η)*N c (t), Among them, E c (t) is the measurement error corresponding to the t-th video frame data; η is the confidence level; N c (t) is the material measurement value corresponding to the tth video frame data.

12. The system according to claim 9, characterized in that The material estimated value determination unit calculates a gain coefficient based on a measurement value error corresponding to the t-th video frame data and an estimated value error corresponding to the t-1-th video frame data, including: Wherein, K(t) is the gain coefficient corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; E c (t) is the measurement value error corresponding to the t-th video frame data.

13. The system according to claim 9, characterized in that The material estimation value determination unit calculates the material estimation value corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data, the material measurement value and the material estimation value corresponding to the t-1-th video frame data, including: N g (t)=N g (t-1)+K(t)×[N c (t)-N g (t-1)], Among them, N g (t) is the material estimation value corresponding to the t-th video frame data; N g (t-1) is the material estimation value corresponding to the t-1th video frame data; N c (t) is the material measurement value corresponding to the t-th video frame data; K(t) is the gain coefficient corresponding to the t-th video frame data.

14. The system according to claim 9, characterized in that The material estimated value determination unit calculates the estimated value error corresponding to the t-th video frame data according to the gain coefficient corresponding to the t-th video frame data and the estimated value error corresponding to the t-1-th video frame data, including: E g (t)=(1-K(t))*E g (t-1), Among them, E g (t) is the estimated value error corresponding to the t-th video frame data; E g (t-1) is the estimated value error corresponding to the t-1th video frame data; K(t) is the gain coefficient corresponding to the tth video frame data.

Images