A power emergency command system

Calculating the merge ratio through the particle swarm algorithm improves the accuracy of aerial image noise reduction processing, solves the problem of inaccurate setting of the merge ratio, and shortens the power recovery time.

CN118691831BActive Publication Date: 2024-11-19国网四川省电力公司电力应急中心 +1
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202411189363.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-11-19
Estimated Expiration
2044-08-28

AI Technical Summary

Technical Problem

In the prior art, in the noise reduction process of aerial images, inaccurate setting of the combined proportion leads to poor image quality, affecting the accuracy of the recognition of the degree of disaster, and prolonging the power recovery time.

Method used

The particle swarm algorithm is used to calculate the merge ratio of each processed image. The particle swarm is initialized by the initial proportion, the particle position and velocity are updated, and the conditions are satisfied to determine the optimal merge ratio and improve the quality of the noise-reducing image.

Benefits of technology

Improves the quality of noise-reducing images and the accuracy of disaster-affected situations, and shortens the total time for power recovery.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118691831B_ABST
    Figure CN118691831B_ABST
Patent Text Reader

Abstract

The present invention belongs to the field of emergency command, and discloses an electric power emergency command system, including an image recognition device; the image recognition device includes an image noise reduction module; the image noise reduction module includes a noise reduction unit, a calculation unit and a merging unit; the noise reduction unit is used to use multiple types of noise reduction algorithms to reduce noise on aerial images of the disaster area respectively, and obtain multiple processed images; the calculation unit is used to use a particle swarm algorithm to obtain the merging ratio of each processed image respectively; the merging unit is used to merge the processed images based on the merging ratio to obtain a noise-reduced image. The present invention can further improve the quality of the obtained noise-reduced image, obtain a more accurate disaster situation, and effectively shorten the total time required for the overall restoration of power supply.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of emergency command, and in particular to an electric power emergency command system. Background Art

[0002] When natural disasters such as typhoons and floods occur, in order to restore power supply as soon as possible, it is usually necessary to first obtain the extent of damage to each local area of ​​the disaster area through aerial images, and then the emergency rescue commander will command the rescue team to conduct on-site inspections of the severely damaged local areas first, and carry out emergency rescue and repair of damaged power facilities to ensure that power supply is restored as soon as possible. Aerial images can usually be obtained by taking pictures of the disaster area with drones.

[0003] The judgment result of the degree of damage in a local area is generally obtained based on image recognition. Compared with the manual recognition method, the image recognition method has lower requirements on the judgment experience of the on-site commander and can obtain more accurate judgment results more quickly. The process of obtaining the judgment result based on image recognition generally includes the steps of noise reduction, image feature extraction, and judging the degree of damage based on the extracted image features.

[0004] In order to improve the quality of image features obtained by feature extraction of aerial images, the denoising process generally uses multiple types of denoising algorithms to perform denoising on the aerial images separately to obtain multiple processed images, and then merge the multiple processed images according to a set merging ratio (for example, a ratio obtained based on experience) to obtain the final denoised image.

[0005] Since the merging ratios corresponding to different denoised images are usually set in advance, there is a high probability that the denoised images obtained based on these merging ratios have significantly different image quality compared to the denoised images obtained with the optimal merging ratio (i.e., the merging ratio that can obtain the highest quality merged image), thereby affecting the accuracy of the final identified degree of damage, increasing the probability that the repair team will give priority to repairing the disaster-stricken areas with the least degree of damage, and extending the total time from the start of the repair to the restoration of power supply to all local areas in the disaster area. Summary of the invention

[0006] The purpose of the present invention is to disclose an electric power emergency command system to solve the technical problem of how to improve the gap between the improved merging ratio and the optimal merging ratio after the denoising processing of aerial images by using different types of denoising algorithms, so that the denoised image obtained based on the merging ratio can have higher quality, thereby improving the accuracy of the recognition result of the degree of damage and shortening the total time for overall restoration of power supply.

[0007] In order to achieve the above-mentioned purpose, the present invention adopts the following technical scheme: the present invention provides an electric power emergency command system, comprising an image recognition device; the image recognition device comprises an image denoising module; the image denoising module comprises a denoising unit, a calculating unit and a merging unit; the denoising unit is used to adopt multiple types of denoising algorithms to denoise the aerial images of the disaster area respectively, and obtain multiple processed images; the calculating unit is used to adopt a particle swarm algorithm to respectively obtain the merging ratio of each processed image; the merging unit is used to merge the processed images based on the merging ratio to obtain the denoised image; wherein, the particle swarm algorithm is used to respectively obtain the merging ratio of each processed image, including: a first step, calculating the initial ratio based on all processed images; a second step, initializing the particle swarm based on the initial ratio; a third step, calculating the fitness of each particle; a fourth step, updating the historical optimal position of each particle according to the fitness, and updating the global historical optimal position of the particle swarm; a fifth step, updating the speed and position of each particle; a sixth step, judging whether the iteration end condition is met, if so, entering the seventh step; if not, entering the third step; the seventh step, taking the values ​​of each dimension in the position of the particle corresponding to the global historical optimal position as the final merging ratio of each processed image.

[0008] Preferably, the multiple types of noise reduction algorithms include an algorithm for reducing noise for Gaussian noise, an algorithm for reducing noise for salt and pepper noise, and an algorithm for reducing noise for periodic noise.

[0009] Preferably, calculating the initial ratio based on all processed images includes: calculating a comparison coefficient of each processed image respectively; and calculating the initial ratio of each processed image based on the comparison coefficient.

[0010] Preferably, the calculation formula of the comparison coefficient is: ; Represents the comparison coefficient of the processed image, migy and magy represent the lower limit and upper limit of the grayscale value of the pixel in the processed image respectively; It indicates the total number of pixels in the processed image whose gray value is equal to z, nuals indicates the total number of pixels in the processed image, cg indicates the comparison value, represents the total number of pixel points at the edge of the processed image, w1 and w2 represent the first weight and the second weight respectively.

[0011] Preferably, calculating the initial ratio of each processed image based on the comparison coefficient includes: using N to represent the total number of processed images, then for the nth processed image, the calculation formula of its initial ratio is: ; represents the initial scale of the nth processed image, Represents the comparison coefficient of the nth processed image, n∈[1,N].

[0012] Preferably, the particle swarm is initialized based on the initial ratio, including: using M to represent the total number of particles in the particle swarm, using N to represent the total number of processed images, and using positive integers from 1 to N to number the N processed images; then for the mth particle, the value of the nth dimension in its velocity is a random number generated in the interval (-0.1, 0.1); m∈[1, M]; when m is equal to 1, for the mth particle, the value of the nth dimension in its position is equal to the initial ratio corresponding to the processed image numbered n; when m is greater than 1, for the mth particle, the value of the nth dimension in its position is a random number generated in the interval The random number generated in Represents the value of the nth dimension in the position of the first particle, n∈[1,N].

[0013] Preferably, the image recognition device further comprises a feature extraction module; the feature extraction module is used to partition the denoised image into a plurality of local images; and extract the image features contained in each local image respectively.

[0014] Preferably, the image recognition device also includes a disaster severity coefficient calculation module; the disaster severity coefficient calculation module is used to obtain the set of disaster types contained in each local image and the distribution area of ​​each type of disaster based on image features, and is used to calculate the disaster severity coefficient of each local image based on the set of disaster types and the distribution area of ​​each type of disaster.

[0015] Preferably, it also includes a command device; the command device includes a sorting module and a scheduling module; the sorting module is used to sort the emergency repair order of the area corresponding to each local image in the order of the disaster degree coefficient from large to small; the scheduling module is used to generate an emergency repair plan according to the total number of local images Q1 and the total number of emergency repair teams Q2, and send the emergency repair plan to each emergency repair team.

[0016] Preferably, it also includes a drone device; the drone device includes a drone control center and a drone; the drone control center is used to control the drone to fly to the sky above the center of the disaster-stricken area, take a bird's-eye view of the disaster-stricken area, and obtain an aerial image of the disaster-stricken area.

[0017] Beneficial effects: When merging processed images obtained by different types of denoising algorithms, the present invention does not directly use a pre-set merging ratio to merge, but first calculates an initial ratio based on the processed image, and then initializes the position in the particle swarm algorithm based on the initial ratio, and then runs the particle swarm algorithm to obtain the global historical optimal position after satisfying the iteration condition; the values ​​of each dimension in the position of the particle corresponding to the global historical optimal position are used as the merging ratio of each processed image finally determined; by initializing the particle swarm algorithm based on the initial ratio, the relationship between the particle swarm algorithm and the processed image is improved, so that the particle swarm algorithm can initialize the initialized particle group closer to the optimal merging ratio faster according to the actual situation of the obtained processed image, thereby being able to reduce the early fall into the local optimal result far from the optimal merging ratio, while improving the convergence speed of the present invention, and improving the efficiency of obtaining a merging ratio closer to the optimal merging ratio; the present invention can further improve the quality of the obtained denoised image, obtain a more accurate disaster situation, and effectively shorten the total time required for the overall restoration of power supply. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A schematic diagram of the power emergency command system.

[0020] Figure 2 A schematic diagram of the process of obtaining the merging ratio of a processed image. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments; the components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations; therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention; based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts belong to the scope of protection of the present invention.

[0022] like Figure 1In one embodiment shown in the figure, the present invention provides an electric power emergency command system, comprising an image recognition device; the image recognition device comprises an image noise reduction module; the image noise reduction module comprises a noise reduction unit, a calculation unit and a merging unit; the noise reduction unit is used to use multiple types of noise reduction algorithms to respectively reduce the noise of the aerial images of the disaster area to obtain multiple processed images; the calculation unit is used to use a particle swarm algorithm to respectively obtain the merging ratio of each processed image; the merging unit is used to merge the processed images based on the merging ratio to obtain a noise-reduced image; Figure 2 As shown, the particle swarm algorithm is used to obtain the merging ratio of each processed image, including: the first step, calculating the initial ratio based on all processed images; the second step, initializing the particle swarm based on the initial ratio; the third step, calculating the fitness of each particle; the fourth step, updating the historical optimal position of each particle according to the fitness, and updating the global historical optimal position of the particle swarm; the fifth step, updating the speed and position of each particle; the sixth step, judging whether the iteration end condition is met, if so, entering the seventh step; if not, entering the third step; the seventh step, taking the values ​​of each dimension in the position of the particle corresponding to the global historical optimal position as the final merging ratio of each processed image.

[0023] When obtaining the merging ratio, the present invention calculates the initial ratio based on the processed image, and then initializes the position in the particle swarm algorithm based on the initial ratio, and then runs the particle swarm algorithm to obtain the global historical optimal position after satisfying the iteration condition; the values ​​of each dimension in the position of the particle corresponding to the global historical optimal position are used as the merging ratio of each processed image finally determined; by initializing the particle swarm algorithm based on the initial ratio, the relationship between the particle swarm algorithm and the processed image is improved, so that the particle swarm algorithm can initialize the initialized particle group closer to the optimal merging ratio more quickly according to the actual situation of the obtained processed image, thereby being able to reduce the risk of falling into a local optimal result far from the optimal merging ratio in advance, while improving the convergence speed of the present invention and the efficiency of obtaining a merging ratio closer to the optimal merging ratio; the present invention can further improve the quality of the obtained denoised image, obtain a more accurate disaster situation, and effectively shorten the total time required for the overall restoration of power supply.

[0024] By adopting multiple types of noise reduction algorithms, the present invention can effectively suppress different types of noise, and can also combine the advantages of different noise reduction algorithms (for example, the noise reduction results of algorithms with strong noise reduction capabilities and noise reduction algorithms with strong edge preservation capabilities of the image after noise reduction), and finally merge the noise reduction results, thereby improving the quality of the denoised image.

[0025] Preferably, the multiple types of noise reduction algorithms include an algorithm for reducing noise for Gaussian noise, an algorithm for reducing noise for salt and pepper noise, and an algorithm for reducing noise for periodic noise.

[0026] Gaussian noise follows a normal distribution and is commonly found in thermal noise and electronic noise in image sensors.

[0027] Salt and pepper noise is random black and white spots that appear in an image, often caused by sensor failure or data transmission errors.

[0028] Periodic noise is periodic, such as the striped interference that appears in an image.

[0029] Causes of periodic noise: (1) Interference from electronic equipment and power supply: Unstable power supply or electromagnetic interference near power lines (such as 50 Hz or 60 Hz power supply frequency) can introduce periodic noise into the image.

[0030] Other electronic devices: Electromagnetic interference on the circuit boards or data transmission lines of cameras, sensors and other electronic devices can also cause periodic noise.

[0031] (2) Mechanical vibration: The camera equipment or sensor is affected by mechanical vibration (such as motor vibration, environmental vibration), resulting in periodic interference in the image.

[0032] (3) Data transmission, data transmission interference: During the transmission of image data, periodic noise may also be introduced due to poor contact of the data line or transmission interface, electromagnetic interference, etc.

[0033] Preferably, the algorithms for reducing Gaussian noise include Gaussian noise reduction algorithm, mean noise reduction algorithm and bilateral noise reduction algorithm.

[0034] Preferably, the algorithm for reducing the salt and pepper noise includes a median noise reduction algorithm and an adaptive median noise reduction algorithm.

[0035] Preferably, the algorithm for reducing the periodic noise includes a deblocking effect noise reduction algorithm and a wavelet transform noise reduction algorithm.

[0036] Preferably, when performing noise reduction on aerial photography, a noise reduction algorithm is selected for each type of noise, and the total number of noise reduction algorithms is equal to the total number of noise types.

[0037] Preferably, calculating the initial ratio based on all processed images includes: calculating a comparison coefficient of each processed image respectively; and calculating the initial ratio of each processed image based on the comparison coefficient.

[0038] By calculating the initial ratio by comparing the coefficients, the initial ratio can be calculated based on the difference in information included in different processed images, so that an initial position closer to the optimal merging ratio can be obtained based on the initial ratio.

[0039] Preferably, the calculation formula of the comparison coefficient is: ; Represents the comparison coefficient of the processed image, migy and magy represent the lower limit and upper limit of the grayscale value of the pixel in the processed image respectively; It indicates the number of pixels in the processed image whose gray value is equal to z, nuals indicates the total number of pixels in the processed image, cg indicates the comparison value, represents the total number of pixel points at the edge of the processed image, w1 and w2 represent the first weight and the second weight respectively.

[0040] The comparison coefficient of the present invention is, on the one hand, a weighted value of the appearance ratio of pixels of different gray values, and on the other hand, a total number of pixels at the edge of the image, which accurately represents the content of data contained in the processed image that can provide assistance for image feature extraction from two different perspectives; the larger the absolute value of the difference between the lower limit and the upper limit of the gray value of the pixel points in the processed image, the more uniform the distribution between the pixels of different gray values, the larger the number of pixels at the edge of the image, and the larger the comparison coefficient of the image. In the subsequent merging process, the larger the weight, which is conducive to obtaining an initialization result that is closer to the optimal merging ratio.

[0041] The lower limit and upper limit of the gray value of the pixel are the minimum gray value and the maximum gray value in the processed image respectively.

[0042] Preferably, the value of cg is 10.

[0043] The pixels at the edge of the image can be obtained by Roberts algorithm.

[0044] Preferably, the values ​​of w1 and w2 are 0.66 and 0.34 respectively.

[0045] Preferably, calculating the initial ratio of each processed image based on the comparison coefficient includes: using N to represent the total number of processed images, then for the nth processed image, the calculation formula of its initial ratio is: ; represents the initial scale of the nth processed image, Represents the comparison coefficient of the nth processed image, n∈[1,N].

[0046] In the process of calculating the initial ratio, the larger the comparison coefficient of the processed image, the larger the corresponding initial ratio; in this way, the processed image containing more data that can help image feature extraction can have a larger value at the time of initialization, and is closer to the optimal merging ratio.

[0047] Preferably, merging the processed images based on the merging ratio to obtain the denoised image includes: merging using the following formula: ; Represents the grayscale value of the pixel at the coordinate (x, y) in the denoised image merimg. Indicates the merging ratio of the nth processed image, Represents the grayscale value of the pixel at coordinate (x, y) in the nth processed image.

[0048] The merged result is the weighted result of the grayscale values ​​of the pixels at the same position in each processed image.

[0049] Preferably, the particle swarm is initialized based on the initial ratio, including: using M to represent the total number of particles in the particle swarm, using N to represent the total number of processed images, and using positive integers from 1 to N to number the N processed images; then for the mth particle, the value of the nth dimension in its velocity is a random number generated in the interval (-0.1, 0.1); m∈[1, M]; when m is equal to 1, for the mth particle, the value of the nth dimension in its position is equal to the initial ratio corresponding to the processed image numbered n; when m is greater than 1, for the mth particle, the value of the nth dimension in its position is a random number generated in the interval The random number generated in Represents the value of the nth dimension in the position of the first particle, n∈[1,N].

[0050] Different from the existing initialization process, the present invention does not use a random generation method when initializing the position of the particle, because the data generated in this way may deviate far from the optimal merging ratio, resulting in that even after a sufficient number of iterations, a result close enough to the optimal merging ratio cannot be obtained; therefore, when the present invention updates the position, the position of the first particle is directly initialized by the initial ratio, and the remaining particles are randomly generated based on the fluctuation range determined by the initial ratio, so that the probability that the randomly generated position is close to the optimal merging ratio is greater.

[0051] Preferably, the calculation function of the particle's fitness is: ; ; represents the fitness of the mth particle in the particle swarm, represents the value of the nth dimension in the position of the mth particle in the particle group, Indicates normalization operation, mapping the value in brackets to the range of [0,1], It represents the score calculated by using the BRISQUE algorithm for the processed image numbered n. Indicates the upper limit of the score calculated by the BRISQUE algorithm, It represents the score calculated by using the NIQE algorithm for the processed image numbered n. Indicates the upper limit of the score calculated by the NIQE algorithm; MAX means obtaining the maximum value of the result in the brackets. It indicates the score calculated by the RankIQA algorithm for the image merImg. Indicates the upper limit of the score calculated by the RankIQA algorithm. represents discrete weights, represents the merging weight, merImg is the image obtained by merging the processed images with the position of the mth particle in the particle group as the merging ratio, and the value of the nth dimension of the position of the mth particle is the merging ratio of the processed image numbered n.

[0052] The fitness of the present invention not only considers the quality of the image after merging, but also considers the weighted values ​​of the quality of different processed images in discrete states; thereby, more additional reference information can be introduced into the fitness calculation function, reducing the probability of the historical optimal position of the iteration falling into the local optimal state too early, and improving the accuracy of the iteration result.

[0053] Preferably, the discrete weights are , the combined weight is .

[0054] Preferably, updating the historical optimal position of each particle according to the fitness includes: comparing the latest acquired fitness with the fitness corresponding to the historical optimal position of the particle, if the latest acquired fitness is greater than the fitness corresponding to the historical optimal position of the particle, then taking the solution corresponding to the latest acquired fitness (i.e., the values ​​of each dimension of the particle's position) as the new historical optimal position of the particle.

[0055] Preferably, updating the global historical optimal position of the particle swarm includes: comparing the maximum value of the fitness of all particles newly obtained with the fitness corresponding to the global historical optimal position; if the maximum value of the fitness is greater than the fitness corresponding to the global historical optimal position, then taking the solution corresponding to the maximum fitness (i.e., the value of each dimension of the particle position) as the new global historical optimal position.

[0056] Preferably, The value of is 100.

[0057] Preferably, The value of is 100.

[0058] Preferably, The value of is 100.

[0059] Preferably, the updating formula of the particle velocity is: ; and Respectively represent the speed of the particle after the s+1th and sth iterations, represents the inertia weight at the s+1th iteration, c1 and c2 represent the first acceleration coefficient and the second acceleration coefficient respectively, and are all values ​​obtained by randomly selecting values ​​from the interval [0,1]. represents the historical optimal position of the particle after the sth iteration, represents the global historical optimal position of the particle swarm after the sth iteration; Represents the velocity of the particle after the sth iteration; the update formula of the particle position is: ; represents the position of the particle after the s+1th iteration; , Represents the value of the nth dimension of the particle's position after the sth iteration, n∈[1,N]; , Represents the value of the nth dimension of the particle's velocity after the sth iteration.

[0060] Preferably, The calculation formula is as follows: ; represents the minimum value of the inertia weight, represents the change of inertia weight, It represents the fitness of the particle after the sth iteration calculation. It represents the minimum fitness of all particles in the particle swarm after the sth iteration calculation. It represents the average fitness of all particles in the particle swarm after the sth iteration calculation. Indicates the maximum number of iterations set.

[0061] The present invention improves the inertia weight. The prior art usually sets a fixed inertia weight or an inertia weight that decreases linearly as the number of iterations increases. The fixed inertia weight is prone to being too large or too small. When the inertia weight is too large, the iterative result still cannot effectively converge after multiple iterations. If a linearly decreasing inertia weight is set, the probability of particles falling into a local optimum prematurely will increase.

[0062] Therefore, on the basis that the inertia weight decreases with the increase of the number of iterations, the present invention further optimizes the change of the inertia weight by comparing the fitness of the particle with the fitness of other particles in the particle group after the previous iteration. When the fitness of the particle is smaller than that of other particles, the particle can be given a larger inertia weight, thereby increasing the range of the particle to obtain the solution, and effectively reducing the probability of falling into the local optimum.

[0063] Preferably, the variation range of the inertia weight is equal to the maximum value of the inertia weight minus the minimum value.

[0064] When the latest calculated inertia weight is greater than the maximum value of the inertia weight, the maximum value of the inertia weight replaces the latest calculated inertia weight; when the latest calculated inertia weight is less than the minimum value of the inertia weight, the minimum value of the inertia weight replaces the latest calculated inertia weight.

[0065] Preferably, the maximum value of the inertia weight is 0.8 and the minimum value is 0.3.

[0066] Preferably, the maximum number of iterations is set to 1000.

[0067] Preferably, the iteration end condition is that the number of iterations is greater than a set maximum number of iterations Sma.

[0068] Preferably, the values ​​of each dimension in the position of the particle corresponding to the global historical optimal position are used as the merging ratio of each processed image finally determined, including: for the processed image numbered n, its merging ratio is the value of the nth dimension in the position of the particle corresponding to the global historical optimal position.

[0069] Preferably, the image recognition device further comprises a feature extraction module; the feature extraction module is used to partition the denoised image into a plurality of local images; and extract the image features contained in each local image respectively.

[0070] Preferably, the denoised image is partitioned into a plurality of local images, including: decomposing the denoised image according to administrative scopes, dividing the denoised image into a plurality of local images, each administrative scope corresponding to a local image.

[0071] For example, the denoised image is decomposed according to the administrative scope of each village, so that the areas belonging to the same village in the denoised image are divided into the same local image.

[0072] For another example, the denoised image is decomposed according to the administrative scope of each town, so that in the denoised image, areas belonging to the same town are grouped into the same local image.

[0073] Preferably, the noise reduction image is partitioned into multiple local images, including: partitioning the noise reduction image according to the repair efficiency of the repair team, dividing the noise reduction image into multiple local images, and the range corresponding to each local image is the average value of the area that can be repaired by all repair teams in one day.

[0074] Preferably, the image recognition device also includes a disaster severity coefficient calculation module; the disaster severity coefficient calculation module is used to obtain the set of disaster types contained in each local image and the distribution area of ​​each type of disaster based on image features, and is used to calculate the disaster severity coefficient of each local image based on the set of disaster types and the distribution area of ​​each type of disaster.

[0075] Preferably, the types of disasters include floods, mudslides, building collapses and ground fissures.

[0076] Preferably, the calculation formula of the disaster degree coefficient is: ; represents the disaster degree coefficient of the qth local image, sumu represents the total number of disaster types, represents the weight of disaster type i, Represents the distribution area of ​​disaster type i in the local image.

[0077] The disaster degree coefficient is the weighted result of the area occupied by various types of disasters. The more types of disasters included in the local image and the larger the distribution range of the disasters, the greater the disaster degree coefficient.

[0078] Specifically, The value of is related to the destructive power of the disaster. The greater the destructive power, the higher the weight can be set.

[0079] For example, the weights corresponding to disasters such as floods, mudslides, house collapses and ground fissures can be set to 0.4, 0.2, 0.2 and 0.2 respectively.

[0080] Preferably, it also includes a command device; the command device includes a sorting module and a scheduling module; the sorting module is used to sort the emergency repair order of the area corresponding to each local image in the order of the disaster degree coefficient from large to small; the scheduling module is used to generate an emergency repair plan according to the total number of local images Q1 and the total number of emergency repair teams Q2, and send the emergency repair plan to each emergency repair team.

[0081] Specifically, generating a repair plan based on the total number of local images Q1 and the total number of repair teams Q2 includes: if Q1 is less than Q2, then sorting the repair teams from strong to weak according to their repair capabilities, and the ones ranked higher are responsible for repairing the areas corresponding to the local images with larger disaster severity coefficients; the first round of arrangements: firstly select the first Q1 repair teams with the strongest repair capabilities, and assign corresponding repair areas to each repair team; the second round of arrangements: evenly distribute the personnel in the remaining Q2-Q1 repair teams to the Q repair teams arranged in the first round.

[0082] If Q1 is equal to Q2, a repair area corresponding to a local image is directly allocated to each team. The repair team with stronger repair ability will be assigned an area corresponding to a local image with a greater disaster degree coefficient as its repair area.

[0083] If Q1 is greater than Q2, the area corresponding to the first Q1 local images with the largest disaster degree coefficient will be repaired first, and then the repair team that has completed the repair work will be dispatched to the remaining unrepaired areas for repair.

[0084] Preferably, sending the emergency repair plan to each emergency repair team includes: sending the emergency repair scope corresponding to each emergency repair team to the terminal device used by the person in charge of the emergency repair team.

[0085] Preferably, it also includes a drone device; the drone device includes a drone control center and a drone; the drone control center is used to control the drone to fly to the sky above the center of the disaster-stricken area, take a bird's-eye view of the disaster-stricken area, and obtain an aerial image of the disaster-stricken area.

[0086] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention; therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A power emergency command system, characterized in that: It includes an image recognition device; the image recognition device includes an image noise reduction module; the image noise reduction module includes a noise reduction unit, a calculation unit and a merging unit; the noise reduction unit is used to use multiple types of noise reduction algorithms to reduce noise on the aerial images of the disaster area respectively, and obtain multiple processed images; The calculation unit is used to respectively obtain the merging ratio of each processed image by using a particle swarm algorithm; The merging unit is used to merge the processed images based on the merging ratio to obtain the denoised image; wherein, the merging ratio of each processed image is obtained respectively by using the particle swarm algorithm, including: the first step, calculating the initial ratio based on all the processed images; the second step, initializing the particle swarm based on the initial ratio; the third step, calculating the fitness of each particle; the fourth step, updating the historical optimal position of each particle according to the fitness, and updating the global historical optimal position of the particle swarm; the fifth step, updating the speed and position of each particle; the sixth step, judging whether the iteration end condition is met, if so, entering the seventh step; if not, entering the third step; the seventh step, taking the values ​​of each dimension in the position of the particle corresponding to the global historical optimal position as the final merging ratio of each processed image; calculating the initial ratio based on all the processed images, including: calculating the comparison coefficient of each processed image respectively; calculating the initial ratio of each processed image based on the comparison coefficient; the calculation formula of the comparison coefficient is: ; Represents the comparison coefficient of the processed image, migy and magy represent the lower limit and upper limit of the grayscale value of the pixel in the processed image respectively; It indicates the total number of pixels in the processed image whose gray value is equal to z, nuals indicates the total number of pixels in the processed image, cg indicates the comparison value, represents the total number of pixels at the edge of the processed image, w1 and w2 represent the first weight and the second weight respectively; calculating the initial proportion of each processed image based on the comparison coefficient includes: using N to represent the total number of processed images, then for the nth processed image, the calculation formula of its initial proportion is: ; represents the initial scale of the nth processed image, Represents the comparison coefficient of the nth processed image, n∈[1,N]; Initialize the particle swarm based on the initial ratio, including: using M to represent the total number of particles in the particle swarm, using N to represent the total number of processed images, and using positive integers from 1 to N to number the N processed images; then for the mth particle, the value of the nth dimension in its velocity is a random number generated in the interval (-0.1,0.1); m∈[1,M]; when m is equal to 1, for the mth particle, the value of the nth dimension in its position is equal to the initial ratio corresponding to the processed image numbered n; when m is greater than 1, for the mth particle, the value of the nth dimension in its position is a random number generated in the interval The random number generated in Represents the value of the nth dimension in the position of the first particle, n∈[1,N]; the image recognition device also includes a feature extraction module; the feature extraction module is used to partition the denoised image and divide the denoised image into multiple local images; the image features contained in each local image are extracted respectively; the image recognition device also includes a disaster degree coefficient calculation module; the disaster degree coefficient calculation module is used to obtain the set of disaster types contained in each local image and the distribution area of ​​each type of disaster based on the image features, and is used to calculate the disaster degree coefficient of each local image based on the set of disaster types and the distribution area of ​​each type of disaster; it also includes a command device; the command device includes a sorting module and a scheduling module; the sorting module is used to sort the repair order of the area corresponding to each local image according to the order of the disaster degree coefficient from large to small; the scheduling module is used to generate a repair plan according to the total number of local images Q1 and the total number of repair teams Q2, and send the repair plan to each repair team; the calculation function of the fitness of the particle is: ; ; represents the fitness of the mth particle in the particle swarm, represents the value of the nth dimension in the position of the mth particle in the particle group, Indicates normalization operation, mapping the value in brackets to the range of [0,1], It represents the score calculated by using the BRISQUE algorithm for the processed image numbered n. Indicates the upper limit of the score calculated by the BRISQUE algorithm, It represents the score calculated by using the NIQE algorithm for the processed image numbered n. Indicates the upper limit of the score calculated by the NIQE algorithm; MAX means obtaining the maximum value of the result in the brackets. It indicates the score calculated by RankIQA algorithm for image merImg. Indicates the upper limit of the score calculated by the RankIQA algorithm. represents discrete weights, represents the merging weight, merImg is the image obtained by merging the processed images with the position of the mth particle in the particle swarm as the merging ratio, and the value of the nth dimension of the position of the mth particle is the merging ratio of the processed image numbered n; the updating formula of the particle speed is: ; and Respectively represent the speed of the particle after the s+1th and sth iterations, represents the inertia weight at the s+1th iteration, c1 and c2 represent the first acceleration coefficient and the second acceleration coefficient respectively, and are all values ​​obtained by randomly selecting values ​​from the interval [0,1]. represents the historical optimal position of the particle after the sth iteration, represents the global historical optimal position of the particle swarm after the sth iteration; Represents the velocity of the particle after the sth iteration; the update formula of the particle position is: ; represents the position of the particle after the s+1th iteration; , Represents the value of the nth dimension of the particle's position after the sth iteration, n∈[1,N]; , Represents the value of the nth dimension of the particle's velocity after the sth iteration; The calculation formula is as follows: ; represents the minimum value of the inertia weight, represents the change of inertia weight, It represents the fitness of the particle after the sth iteration calculation. It represents the average fitness of all particles in the particle swarm after the sth iteration calculation. Indicates the maximum number of iterations set.

2. The power emergency command system according to claim 1, characterized in that: The various types of noise reduction algorithms include algorithms for reducing Gaussian noise, algorithms for reducing salt and pepper noise, and algorithms for reducing periodic noise.

3. The power emergency command system according to claim 1, characterized in that: It also includes a drone device; the drone device includes a drone control center and a drone; the drone control center is used to control the drone to fly to the sky above the center of the disaster-stricken area, take a bird's-eye view of the disaster-stricken area, and obtain aerial images of the disaster-stricken area.

Citation Information

Patent Citations

  • Image processing method and system thereof

    CN103778599A

  • Automatic image segmentation method of continuous quantum goose group algorithm evolution pulse coupling neural network system parameters

    CN103824291A

  • Self-adaptive compressed sensing reconstruction method based on image information content difference

    CN106530365A

  • Image fractal interpolation method for optimizing particle swarm

    CN108009988A

  • Image segmentation method and device based on ALSLCV model

    CN114596320A