CN112446874A - Human-computer cooperation autonomous level damage assessment method - Google Patents
Human-computer cooperation autonomous level damage assessment method Download PDFInfo
- Publication number
- CN112446874A CN112446874A CN202011453012.1A CN202011453012A CN112446874A CN 112446874 A CN112446874 A CN 112446874A CN 202011453012 A CN202011453012 A CN 202011453012A CN 112446874 A CN112446874 A CN 112446874A
- Authority
- CN
- China
- Prior art keywords
- damage
- image
- target
- evaluation result
- images before
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 77
- 238000011156 evaluation Methods 0.000 claims abstract description 82
- 230000008859 change Effects 0.000 claims abstract description 70
- 238000004088 simulation Methods 0.000 claims abstract description 66
- 239000002360 explosive Substances 0.000 claims abstract description 45
- 238000001514 detection method Methods 0.000 claims abstract description 35
- 238000005259 measurement Methods 0.000 claims abstract description 19
- 230000011218 segmentation Effects 0.000 claims description 37
- 230000000694 effects Effects 0.000 claims description 25
- 238000012545 processing Methods 0.000 claims description 25
- 238000004422 calculation algorithm Methods 0.000 claims description 21
- 230000000877 morphologic effect Effects 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 11
- 238000012937 correction Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 5
- 239000012634 fragment Substances 0.000 description 31
- 230000035939 shock Effects 0.000 description 29
- 238000004364 calculation method Methods 0.000 description 22
- 230000035515 penetration Effects 0.000 description 22
- 238000004880 explosion Methods 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000003068 static effect Effects 0.000 description 5
- 238000005474 detonation Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000003628 erosive effect Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000008676 import Effects 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 241000566145 Otus Species 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000013467 fragmentation Methods 0.000 description 1
- 238000006062 fragmentation reaction Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/80—Geometric correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
- G06T7/337—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving reference images or patches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20036—Morphological image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a human-computer cooperation autonomous level damage assessment method, which comprises the following steps: s1, acquiring images of a target before and after damage of a specified explosive and presetting simulation parameters; s2, performing model simulation on the damage efficiency of the target under the specified explosive, calculating a first damage evaluation result, performing image change detection on the acquired images before and after damage, constructing difference images of the images before and after damage, calculating a second damage evaluation result according to the difference images respectively, and superposing the change result on the image before the target damage by using an image enhancement mode to receive an externally input auxiliary evaluation result; and S3, fusing the first damage evaluation result, the second damage evaluation result and the auxiliary evaluation result to obtain a final damage measurement result. The method has the advantages of simple implementation method, high precision and efficiency of measurement and evaluation results, good real-time performance and the like.
Description
Technical Field
The invention relates to the technical field of damage efficiency evaluation and measurement, in particular to a man-machine cooperation autonomous level damage evaluation method.
Background
Damage assessment is an important link in an OODA (objective observation, origin judgment, decide decision, act action) loop, and is an important assessment method for the damage efficiency of explosive equipment. The observation result of the damage evaluation stage provides an information basis for judgment, and the judgment result can directly influence the accuracy of decision, thereby determining whether the action is effective. The damage assessment research relates to the aspects of military science, autonomous control, image processing, numerical calculation and the like, and belongs to the comprehensive application field of multiple disciplines.
For the evaluation and measurement of target damage performance, the following methods are mainly adopted at present:
(1) the analysis and simulation method based on the explosive damage efficacy and the target vulnerability is a method for simulating the damage process by using a computer, and comprises the following main steps: modeling target vulnerability and explosive damage effectiveness; then, on the basis of an empirical function method, an engineering calculation formula is fitted by using the existing test result, and a damage result is obtained through computer operation. The method can comprehensively and systematically describe the damage process, but also has the problems of great influence of modeling accuracy on simulation results, large simulation calculation amount, wide related fields and complex simulation.
(2) The manual interpretation method comprises the following steps: generally, the judgment is divided into individual judgment of experts and judgment of expert groups, and currently, the judgment is mainly carried out by the expert groups, namely, the processes of manual processing, analysis and judgment are carried out on the collected information by applying knowledge and experience of personnel groups which are trained professionally and have corresponding professional knowledge. The manual judgment method has the advantages that: firstly, the judgment process is centered on people, and the understanding of the situation is strong; and secondly, the judgment process refers to more expert experiences, and the requirement on interpretation data is lower compared with an automatic algorithm. It still has significant disadvantages: firstly, the composition and the level of an expert group have great influence on the interpretation result; secondly, the subjectivity is strong in the judging process, and the traditional experience can mislead the subjectivity; thirdly, the method is generally only suitable for judgment estimation of the overall situation, and accurate judgment is difficult to support.
The core of damage assessment is target change detection, among the methods, the analysis and simulation method based on the explosive damage efficacy and the target vulnerability has the problem of complex modeling, needs to be respectively modeled and analyzed according to different targets and different explosives, and has poor universality and higher requirement on a database; expert evaluation methods are capable of understanding and evaluating the damage image based on their experience, but generally require a longer time and are less automated. Therefore, it is desirable to provide an evaluation method for measuring the target damage performance, so as to increase the intelligence of the evaluation measurement, reduce the complexity of the evaluation measurement, and ensure the accuracy and efficiency of the evaluation measurement.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: aiming at the technical problems in the prior art, the invention provides the human-computer cooperation autonomous level damage assessment method which is simple in implementation method, high in measurement assessment result precision and efficiency and good in real-time performance, and can integrate an image change detection method and a model simulation method to realize target damage efficiency measurement.
In order to solve the technical problems, the technical scheme provided by the invention is as follows:
a human-computer cooperation autonomous level damage assessment method comprises the following steps:
s1, acquiring images before and after damage of a target under a specified explosive with damage efficacy and presetting simulation parameters;
s2, performing model simulation on the damage efficiency of the target under the specified explosive according to the preset simulation parameters, calculating a first damage evaluation result, performing image change detection on the acquired images before and after damage, constructing a difference image of the images before and after damage, calculating a second damage evaluation result according to the difference image, and superposing the change result on the image before damage of the target in an image enhancement mode according to the difference image so as to receive an externally input auxiliary evaluation result;
and S3, fusing the first damage evaluation result, the second damage evaluation result and the auxiliary evaluation result to obtain a final damage measurement result.
Further, the step of calculating a second damage assessment result in step S2 includes:
image registration: respectively carrying out image registration and correction on the images before and after the damage so as to represent the images before and after the damage in the same coordinate system and obtain the registered images before and after the damage;
target segmentation: respectively carrying out target segmentation on the registered before-damage and after-damage images to obtain target areas in the images;
detecting image change: carrying out image change detection on the target area in the registered images before and after the damage to construct a difference image;
image change recognition: performing morphological processing and gray threshold segmentation on the difference image to obtain change information of the image;
and (3) damage assessment: and extracting damage information from the change information of the image, and counting and calculating the physical damage proportion and the functional damage degree to obtain a second damage evaluation result.
Further, the specific step of receiving the auxiliary evaluation result input from the outside in step S2 includes:
image registration: respectively carrying out image registration and correction on the images before and after the damage so as to represent the images before and after the damage in the same coordinate system and obtain the registered images before and after the damage;
detecting image change: carrying out image change detection on the target area in the registered images before and after the damage to construct a difference image;
image change recognition: performing morphological processing and gray threshold segmentation on the difference image to obtain change information of the image;
image enhancement: superposing the change information of the image on the original damaged image in an image enhancement mode;
and (3) auxiliary evaluation receiving: and receiving an auxiliary evaluation result input from the outside.
Further, the image registration adopts an image registration method based on SURF feature points.
Further, the target segmentation uses a GrabCut (graph cut) algorithm based on a graph theory algorithm.
Further, the image change detection uses an image numerical method, and the specific steps include: dividing corresponding elements in an image matrix based on a ratio method, dually matching the relationship of two input images, limiting the gray scale range of an output image between 0 and 1, expanding the value range to [0, 255] through linear transformation, simultaneously adding 1 to the numerator and denominator of the ratio, and constructing the expression of calculating the difference image, wherein the expression specifically comprises the following steps:
wherein, a (x, y) and B (x, y) are the gray values of the two input images at a corresponding point respectively, and C (x, y) is the gray value of the difference image at the corresponding point.
Further, the morphological processing specifically uses a ROI-based closed-loop operation to process the difference image, and the threshold segmentation adopts a threshold segmentation method based on an OTSU algorithm.
Further, the target functional vulnerability modeling specifically uses Damage Tree Analysis (DTA).
Further, the specific step of calculating the first damage assessment result in step S2 includes: the method comprises the steps of modeling the damage efficacy of various types of explosives in advance to obtain various types of explosive damage efficacy models, modeling the vulnerability of various types of targets to obtain various target vulnerability models, adding the various target vulnerability models into a model database, wherein the target vulnerability models comprise target appearance geometric modeling, physical vulnerability modeling used for modeling physical mechanical parameters of the targets and target functional vulnerability modeling used for analyzing the corresponding relation between the damage types and the damage results in various subsystems, components and units in the targets, calling the required target vulnerability models and explosive damage efficacy models from the model database to perform damage effect simulation, and calculating the first damage assessment result by using the preset simulation parameters.
Compared with the prior art, the invention has the advantages that:
1. according to the invention, a damage evaluation method is constructed by fusing damage evaluation methods based on image change detection, model simulation calculation and external assistance, damage evaluation is carried out based on model simulation and image change detection respectively according to images before and after target damage and preset simulation parameters in a damage scheme, and then a final evaluation result is given by fusing external auxiliary evaluation results, so that the capacity of an autonomous platform can be fully exerted, the effect of 'human in loop' is achieved by considering the supervision effect of human, the advantages based on image change detection, model simulation and external auxiliary evaluation are fully exerted, and accurate and efficient target damage efficiency evaluation measurement under various different conditions can be met.
2. According to the invention, a second damage evaluation result is calculated based on a human-in-loop damage evaluation mode of image change detection, and images before and after damage are represented in the same coordinate system by firstly carrying out image registration and correction; then, performing target segmentation, and after a target area is obtained, performing change detection through an image numerical method to construct a difference image; morphological processing and gray threshold segmentation are carried out on the difference image to extract damage information and carry out statistical calculation, and the physical damage proportion and the functional damage degree can be obtained by utilizing efficient calculation of image change.
3. According to the invention, by introducing image enhancement in image processing, after registration, change detection, morphological processing and threshold segmentation are carried out on the images before and after damage, the global change information of the images is superposed on the original damaged images in an image enhancement mode, an operator is assisted in expert interpretation, and the expert interpretation can be assisted by using the image enhancement mode, so that more reasonable expert-assisted evaluation results can be obtained, and more accurate final evaluation measurement results can be obtained by fusing a method based on image change detection.
Drawings
Fig. 1 is a schematic flow chart illustrating an implementation of the human-computer cooperation autonomous level damage assessment method according to the present embodiment.
Fig. 2 is a schematic diagram illustrating the principle of explosive damage effectiveness modeling employed in the present embodiment.
FIG. 3 is a schematic diagram illustrating a modeling principle of vulnerability of a target in a specific application embodiment.
Fig. 4 is a diagram illustrating a simulation result of the damage effect obtained in the embodiment.
Fig. 5 is a schematic diagram of a scale pyramid constructed in the SURF algorithm employed in this embodiment.
Fig. 6 is a GrabCut undirected graph constructed in this example.
Fig. 7 is a schematic diagram of the morphological processing and OTSU algorithm principle employed in this embodiment.
Detailed Description
The invention is further described below with reference to the drawings and specific preferred embodiments of the description, without thereby limiting the scope of protection of the invention.
As shown in fig. 1, the steps of the human-computer cooperation autonomous level damage assessment method of the present embodiment include:
s1, acquiring images before and after damage of a target under a specified explosive with damage efficacy and presetting simulation parameters;
s2, performing model simulation on damage effectiveness of a target under a specified explosive according to preset simulation parameters, calculating a first damage evaluation result, performing image change detection on the acquired images before and after damage, constructing a difference image of the images before and after damage, calculating a second damage evaluation result according to the difference image, and superposing the change result on the image before the target damage by using an image enhancement mode according to the difference image so as to receive an externally input auxiliary evaluation result;
and S3, fusing the first damage evaluation result, the second damage evaluation result and the auxiliary evaluation result to obtain a final damage measurement result.
The embodiment starts from improving the real-time performance and accuracy of damage assessment, simultaneously solves the problems of real-time performance and complexity in the damage assessment, a set of damage assessment method is constructed by fusing damage assessment methods based on image change detection, model simulation calculation and external assistance (such as expert manual interpretation), carrying out damage assessment based on model simulation and image change detection according to the images before and after the target damage and preset simulation parameters in the damage scheme, fusing external auxiliary assessment results to give a final assessment result, the method can give full play to the ability of the autonomous platform, takes the supervision effect of people into consideration, achieves the effect of 'people in a loop', gives full play to the advantages of image change detection, model simulation and external auxiliary evaluation, and can meet the requirements of accurate and efficient target damage efficiency evaluation and measurement under various different conditions.
The method of the embodiment is applied to a typical unmanned aerial vehicle OODA loop in a real environment, and the specific process is as follows: feeding back target state information after the unmanned aerial vehicle detects a target; determining a damage plan according to the target state, wherein the damage plan comprises information such as explosive throwing position, angle, preset blast height, speed and the like; the unmanned aerial vehicle puts explosives in according to a preset plan; after the damage is implemented, the unmanned aerial vehicle acquires image information of a target, and simultaneously carries out simulation calculation, image change detection and expert-assisted interpretation on an explosive damage efficacy model and a target vulnerability model, and a final evaluation result is given by the system by integrating confidence degrees of the three methods.
The specific step of calculating the first damage evaluation result in step S2 in this embodiment includes: the method comprises the steps of modeling the damage efficacy of various types of explosives in advance to obtain various types of explosive damage efficacy models, modeling the vulnerability of various types of targets to obtain various target vulnerability models, adding the various target vulnerability models into a model database, wherein the target vulnerability models comprise target appearance geometric modeling, physical vulnerability modeling used for modeling physical mechanical parameters of the targets and target functional vulnerability modeling used for corresponding relations between damage types and damage results in various subsystems, components and units in the targets, calling the required target vulnerability models and explosive damage efficacy models from the model database to perform damage effect simulation, and calculating a first damage evaluation result by using preset simulation parameters.
In a specific application embodiment, a first damage evaluation result is calculated by using a mode of building and simulating based on an offline model, specifically, firstly, modeling is performed on the explosive damage efficacy and target vulnerability, and the model is added into a database, wherein a common target such as a building, a vehicle, an airplane, a ship and the like is built in the database to form a general model, and other common target vulnerability models such as a house, a radar and the like and explosive damage efficacy models with various equivalent weights and types are also provided; in the evaluation process, the system automatically builds a simulation example according to a damage scheme (target parameters and expected explosive parameters), calls a corresponding target vulnerability model and an explosive damage efficacy model in a database, and automatically sets simulation parameters to carry out damage evaluation numerical calculation; if the modeling effect is poor due to factors such as poor communication, damaged platform and the like, correcting simulation parameters; after the simulation calculation example is built, simulation calculation is carried out, a preset simulation frequency and a fragment field calculation mode are selected, whether fragment field simulation is carried out or not can be selected according to a target type, and the system automatically carries out data import, numerical value initialization, warhead initialization, fragment field initialization and the like after selection is finished. And the simulation result is dynamically demonstrated by adopting a three-dimensional model, and the result data is stored in a file form. Particularly, fragment simulation is not needed for building targets, and fragment simulation is needed for targets such as vehicles, ships, airplanes and the like. In addition, whether Monte Carlo simulation is carried out or not can be selected according to requirements so as to improve accuracy.
The damage effect simulation specifically comprises shock wave effect damage simulation, fragment effect damage simulation and penetration effect damage simulation, and the modeling of the damage effect of the explosive in the specific application embodiment is specifically shown in fig. 2, wherein the graph (a) corresponds to a shock wave fixed-point waveform, and the graph (b) corresponds to a static fragment scattering model of the cylindrical charge.
The specific steps of the shock wave effect damage simulation in this embodiment include:
calculating the relative relation between each surface element of the target and the explosion center, and judging whether the surface element and the explosion center are in communication view or not and judging the included angle between the surface element and the direction of the shock wave;
establishing a virtual firing line for the shock wave to calculate the relative relation between the surface element and the explosion center, and obtaining parameter data of the shock wave on the target surface by using a shock wave propagation and attenuation model, wherein the parameter data comprises amplitude, an incident angle and the like;
and calculating to obtain the reflected overpressure and reflected overpressure impulse on the target surface according to the obtained parameter data of the shock wave on the target surface.
The specific steps of the fragment effect damage simulation comprise: calculating an initial fragment flying field formed by explosion according to static fragment damage efficacy parameters of the target explosive, wherein the static fragment damage efficacy parameters comprise flight speed and direction data of the target explosive at a trajectory terminal point; calculating whether the fragments meet the target or not, and if so, calculating the position of the meeting point, the penetration speed and the penetration angle during meeting; calculating the damage condition of the fragments to the target by using a classical penetration equation, and counting the result to obtain data such as the damage effect, the damage grade and the like of the target;
the motion condition of the bullet in a penetration medium is directly calculated in penetration effect damage simulation, and a penetration trajectory is given.
The shock wave effect is that explosive explosives explode in air to generate high-temperature and high-pressure detonation products and initial shock waves, then the initial shock waves evolve into explosive air shock waves (or explosion waves for short), and after the explosive waves are formed, the explosive waves are separated from the detonation products and independently spread in the air. The true bookThe model of the change of the shock wave pressure with time constructed by the example is shown in FIG. 2(a), Δ pm＝pm-p0The overpressure (the pressure exceeds the atmospheric pressure), τ + represents the duration of positive pressure, and I + represents the specific impulse, which is specifically defined as the overpressure Δ p of the shock wave in this embodimentmPositive pressure duration τ +, specific impulse I + constitute the three basic parameters of the blast air shock wave.
The fragment effect is the fragments which are generated by the explosive and fly at high speed during explosion, the parameters of the fragments include the quantity, the mass and the distribution thereof, the initial velocity of the fragments, the space fly-off characteristics of the fragments and the like, wherein the initial velocity of the fragments and the space fly-off of the fragments are obtained by calculation, and other parameters can be determined by experiments.
In order to improve the algorithm efficiency, the embodiment specifically adopts the optimized fragment initial velocity formula model as follows:
wherein D is the explosive detonation velocity, and C/M is the metal ratio, namely the ratio of the explosive charging mass C to the mass M of the explosive metal shell.
Assuming that the preformed fragments are arranged only on the side surface of the cylindrical charge, the static explosion and flying characteristics can be described by two parameters of a flying azimuth angle and a flying beam width, and a fragment space flying characteristic model is shown in fig. 2 (b).
The embodiment specifically uses penetration capability to model the fragment penetration effect, including the penetration thickness (penetration limit thickness) of the equivalent target and the penetration speed (penetration limit speed) required to penetrate the equivalent target of a certain thickness.
The target modeling of the embodiment specifically includes appearance geometric modeling, physical vulnerability modeling, and functional vulnerability modeling, and specifically includes:
geometric modeling of the target appearance: specifically, CAD software can be used to perform geometric modeling of the target and to divide the mesh, and then to output data of the geometric shape of the target. The key of geometric modeling is to obtain the shape parameters and the size of a target, and the shape parameters and the size can be obtained through engineering drawings or three-view images. The geometric model of the appearance of the target constructed in the specific application embodiment is shown in fig. 3 (a).
Modeling the target physical vulnerability: the physical vulnerability of the target comprises relevant physical mechanical parameters, damage standards and the like, and specifically comprises equivalent thickness of a part material, tensile strength, a kinetic energy standard of effective fragments, a surface damage density standard, a fragment damage threshold value, a shock wave damage overpressure damage threshold value and the like.
Modeling the vulnerability of the target function: specifically, the DTA method is performed by using a damage tree analysis method, as shown in fig. 3(b), the DTA method starts from a damage (called a top event) of the system, and gradually analyzes downwards what kind of damage is done to subsystems, components, units, etc. constituting the subsystem, which may cause such a result. By adopting the DTA method, the damage of a certain unit can be considered, and the influence on the system when a plurality of units generate a certain level of damage at the same time can also be considered.
The model simulation result obtained in the specific application embodiment is shown in fig. 4, and corresponds to the explosive damage effectiveness modeling, and the damage effect simulation specifically includes shock wave damage simulation, fragmentation damage simulation, and penetration damage simulation, and specifically includes:
simulating the damage of the shock wave: specifically, an optimized explosion shock wave loading algorithm is adopted, the relative relation between each surface element of a target and an explosion center is calculated, whether the surface element and the explosion center are in a see-through state or not and the included angle between the surface element and the direction of the shock wave are judged, then a virtual firing line is established for the shock wave to calculate the relative relation between the surface element and the explosion center, the data such as the amplitude, the incident angle degree and the like of the shock wave on the surface of the target are obtained by utilizing a propagation and attenuation model of the shock wave, the reflected overpressure and the reflected overpressure impulse on the surface of the target are analyzed, the reflected overpressure impulse is given out in a report form or a graphical mode, and the damage condition of the shock wave of the target. By adopting the optimized explosion shock wave loading algorithm to carry out shock wave damage simulation, the load of the shock wave on the surface of a target part can be quickly calculated under the conditions of not calculating a shock wave flow field and not calculating fluid-solid coupling of the shock wave and the target, and data such as an overpressure peak value, an overpressure impulse and the like are given. The simulation demonstration of the damage effect obtained in the specific application embodiment is shown in fig. 4 (a).
Simulation of fragment damage: according to the static rupture disk damage efficacy parameter of the ammunition, considering the flight speed, direction and other data of the ammunition at the trajectory terminal point, and calculating an initial rupture disk flying field formed by explosion by a system; on the basis, whether the fragments meet the target or not is calculated, if the fragments meet the target, the position of the meeting point, the penetration speed and the penetration angle during meeting are calculated, on the basis, the damage condition of the fragments to the target is calculated by using a classical penetration equation, the result is counted, and the data of the damage effect, the damage grade and the like of the target are obtained.
Carrying out penetration trajectory simulation: the method has the advantages that the movement condition of the projectile in a penetration medium is directly calculated, penetration trajectory is given, the division of target body grids is omitted, the complex contact problem and the distortion of the grids when the target body is greatly deformed in calculation are not needed to be considered, and compared with the traditional dynamic software for penetration numerical simulation, the method greatly saves calculation time and improves calculation efficiency. The penetration trajectory obtained in the specific application example is shown in fig. 4 (b).
The specific step of calculating the second damage evaluation result in step S2 in this embodiment includes:
image registration: respectively carrying out image registration on the images before and after the damage so as to represent the images before and after the damage in the same coordinate system and obtain the registered images before and after the damage;
target segmentation: respectively carrying out target segmentation on the registered images before and after the damage to obtain a target area in the images;
detecting image change: carrying out image change detection on the target area in the registered images before and after the damage, and constructing to obtain a difference image;
image change recognition: performing morphological processing and gray threshold segmentation on the difference image to obtain change information of the image;
and (3) damage assessment: and extracting damage information from the change information of the image, and counting and calculating the physical damage proportion and the functional damage degree to obtain a second damage evaluation result.
In the embodiment, a second damage evaluation result is calculated based on a damage evaluation mode of a person in a loop of image change detection, image registration and correction are firstly carried out, and images before and after damage are represented in the same coordinate system; then, performing target segmentation, and after a target area is obtained, performing change detection through an image numerical method to construct a difference image; morphological processing and gray threshold segmentation are carried out on the difference image to extract damage information and carry out statistical calculation, and the physical damage proportion and the functional damage degree can be obtained by utilizing efficient calculation of image processing.
In this embodiment, the target segmentation specifically uses an interactive target segmentation algorithm, and a good segmentation effect can be achieved through a small amount of operations.
The specific step of receiving the auxiliary evaluation result input from the outside in step S2 in this embodiment includes:
image registration: respectively carrying out image registration on the images before and after the damage so as to represent the images before and after the damage in the same coordinate system and obtain the registered images before and after the damage;
detecting image change: carrying out image change detection on the target area in the images before and after the damage after the registration to construct a difference image;
image change recognition: performing morphological processing and gray threshold segmentation on the difference image to obtain change information of the image;
image enhancement: superposing the change information of the image on the original damaged image in an image enhancement mode;
and (3) auxiliary evaluation receiving: and receiving an auxiliary evaluation result input from the outside.
On the basis of manual expert interpretation, by introducing image enhancement in image processing, and by carrying out registration, change detection, morphological processing and threshold segmentation on images before and after damage, the embodiment superimposes global change information of the images on the original damaged images in an image enhancement mode to assist operators in expert interpretation, can assist expert interpretation by using the image enhancement mode, and can obtain more reasonable expert-assisted evaluation results.
In this embodiment, the image registration specifically adopts an image registration method based on SURF feature points, and when detecting and matching the feature points, the positioning of SURF registration points is mainly completed, the core is to construct a Hessian matrix, and find stable edge points, i.e., interest points, in the image. For an image f (x, y), wherein a pixel gray value represents a function value, a scale pyramid is firstly constructed, as shown in fig. 5, box-type filtering templates with different scales are used for filtering, so as to obtain new images D (x, y) with different scales, and the constructed Hessian matrix is specifically as follows:
wherein D isxx(x, y) represents the second derivative of image D (x, y) with respect to the x-direction, the same holds true.
After the weight coefficient of 0.9 is introduced, the discriminant of the Hessian matrix is specifically as follows:
det(H)＝Dxx×Dyy-(0.9Dxy)2 (3)
when the discriminant of the Hessian matrix obtains an extreme value in the airspace and scale space fields, the point can be judged to be brighter or darker than other points in the neighborhood, and therefore the point can be judged to be a key point. Then determining the main characteristic direction of the image through Haar wavelet characteristics; after the characteristic point descriptors are established, the components of the two characteristic point descriptors are respectively used as xiAnd yiAnd (3) representing, calculating Euclidean distances of descriptors of feature points between the images before and after the damage:
the shorter the euclidean distance, the better the similarity between the two feature points, i.e., the better the matching.
Through the feature point matching relationship, a spatial transformation matrix H can be constructed, and the transformation relationship is established as follows:
after the registration image is obtained, the target needs to be segmented for change detection. In this embodiment, the GrabCut algorithm based on the graph theory algorithm is specifically used for target segmentation, and after receiving the substantially rectangular range of the region where the target selected by the operator box is located, all regions outside the rectangle are regarded as the background, and the part inside the rectangle is regarded as the possible foreground or the background. GrabCut uses the Max Flow/Min Cut theory to map an image into an Undirected Graph (Undirected Graph) in Graph theory, as shown in FIG. 6, which includes nodes and edges, and the expression is as follows:
G＝(V,E)
V＝(Vi,S,T)
E＝(Eij,Eis,Eit) (6)
the nodes V are of two types, one type representing each pixel V in the imagei(ii) a The other type is terminal nodes, wherein one S node represents the foreground and one T node represents the background. Edges also include two types: one type represents a connection relation E between adjacent pixelsijAnd the other is the connection relation E between the pixel and the terminalisAnd EitThen, the energy function of the graph can be established as follows:
E(L)＝aR(L)+B(L) (7)
wherein, r (l) is a region term, which represents the weight of the edge from the node to the terminal, and is divided into two parts: one part is the weight Rp (1) of the edge connected to the S node, and the other part is the weight Rp (0) of the edge connected to the T node. When a node is more likely to belong to the foreground, the smaller its Rp (1) is, the larger Rp (0) is. B (L) is a boundary term and represents the weight of the edge between the nodes, and when the difference value of the nodes at the two ends of the edge is larger, namely the edge is more likely to be a cut edge, B (L) is smaller. a is a scale factor representing the relative weight of the region term and the boundary term. The optimization objective of the GrabCut method is to optimize the energy function to a minimum.
After the target area is obtained, a difference image before and after damage needs to be constructed, and the embodiment uses a ratio method in an image numerical method to perform change detection and makes corresponding improvement. The ratio method uses division in image arithmetic operation as a basic processing method, and is also called ratio conversion. The ratio method divides corresponding elements in the image matrix, and the expression specifically comprises:
wherein, a (x, y) and B (x, y) are the gray values of the two input images at a corresponding point respectively, and C (x, y) is the gray value of the difference image at the point.
In order to avoid the problem of different outputs caused by the denominator of different images, the embodiment limits the gray scale range of the output image to be between 0 and 1 according to the relationship between the two input images. In order to extract damage information by subsequent threshold segmentation, the value range is expanded to [0, 255] by linear transformation, and meanwhile, in order to avoid that the ratio is not established because the gray value of the original image is 0, 1 is added to the numerator and denominator of the ratio. The expression of the improved ratio method adopted in this embodiment is specifically:
the constructed difference image usually contains noise and dark spot defects, and in order to extract change information more accurately, morphological processing is usually used to reduce the influence of the noise and the dark spot. In order to improve the computation efficiency when the difference image ROI (Region Of Interest) occupies a small area in the damage evaluation, the morphological processing Of this embodiment specifically uses a closing operation based on ROI to process the difference image, as shown in fig. 7, where a represents the ROI portion Of the original image, B represents the convolution kernel, first obtains a' through an image expansion (decomposition) operation, and then obtains a "through an Erosion (Erosion) operation.
After the processed difference image is obtained, image segmentation is performed on the image to obtain damage information. In this embodiment, a threshold segmentation method based on an OTSU algorithm is specifically adopted, the OTUS algorithm is a maximum inter-class variance method, and the inter-class variance between the foreground and the background segmented by the OTSU threshold is made to be maximum by solving the OTSU threshold. The inter-class variance is an index for describing distribution uniformity in a pixel gray space, and the threshold segmentation is performed through an OTSU algorithm so that the probability of wrong segmentation can be minimized. The principle is as follows:
σ2＝p1(g1-gG)2+p2(g2-gG)2 (10)
wherein the OTSU threshold divides the gray scale into a background portion that is less than the threshold and a foreground portion that is greater than the threshold. The probability of the background part is p1, and the average gray scale is g 1; the probability of the foreground part is p2, and the average gray scale is g 2; the global average gray scale is gG. Therefore, the optimization goal of the OTSU algorithm is to solve TH, so that the between-class variance σ is2And max.
As shown in fig. 1, the detailed steps for implementing the target damage performance measurement in the embodiment of the present application are as follows:
the system firstly extracts images before and after the target damage and preset simulation parameters, and then sequentially enters three evaluation method flows.
Firstly, modeling the damage efficacy and target vulnerability of explosives and adding the models into a database by a damage evaluation module based on model simulation calculation, and then automatically calling target parameters and expected explosive parameters in the damage plan by a system to build a simulation example; confirming and correcting the model and parameters in the example, saving the example by the system, entering a calculation mode, and selecting a preset simulation frequency and a fragment field calculation mode; after the selection is finished, the system automatically conducts data import, numerical value initialization and the like; and the simulation result is dynamically demonstrated by adopting a three-dimensional model, and the result data is stored in a file form.
Firstly, image registration and correction are carried out by a person based on image change detection in a loop damage evaluation module, and images before and after damage are represented in the same coordinate system; secondly, performing target segmentation by using an interactive target segmentation algorithm, after a target area is obtained, performing change detection by using an image numerical method, and constructing a difference image; morphological processing and gray threshold segmentation are carried out on the difference image to extract damage information and statistical calculation is carried out on the damage information to obtain the physical damage proportion and the functional damage degree.
The expert interpretation module based on image enhancement performs registration, change detection, morphological processing and threshold segmentation on the images before and after the damage, and the global change information of the images is superimposed on the original damaged images in an image enhancement mode to assist operators in expert interpretation, and the system receives the auxiliary evaluation result of expert interpretation.
After the damage evaluation results of the three modules are obtained, the system fuses the results according to the preset weight to obtain the final damage evaluation result.
By the method, the image registration correction, the optimization ratio method change detection, the morphological processing and the gray threshold segmentation processing are sequentially carried out on the image, the evaluation modes based on the image change detection, the model simulation and the external auxiliary evaluation can be fully fused, the precision of the final target damage performance evaluation measurement is ensured, the efficiency and the real-time performance of the evaluation measurement are ensured, the complexity of the whole evaluation measurement implementation process is low, and the complex modeling or image processing algorithm is not needed.
The foregoing is considered as illustrative of the preferred embodiments of the invention and is not to be construed as limiting the invention in any way. Although the present invention has been described with reference to the preferred embodiments, it is not intended to be limited thereto. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical spirit of the present invention should fall within the protection scope of the technical scheme of the present invention, unless the technical spirit of the present invention departs from the content of the technical scheme of the present invention.
Claims (8)
1. A human-computer cooperation autonomous level damage assessment method is characterized by comprising the following steps:
s1, acquiring images before and after damage of a target under a specified explosive with damage efficacy and presetting simulation parameters;
s2, performing model simulation on the damage efficiency of the target under the specified explosive according to the preset simulation parameters, calculating a first damage evaluation result, performing image change detection on the acquired images before and after damage, constructing a difference image of the images before and after damage, calculating a second damage evaluation result according to the difference image, and superposing the change result on the image before damage of the target in an image enhancement mode according to the difference image so as to receive an externally input auxiliary evaluation result;
and S3, fusing the first damage evaluation result, the second damage evaluation result and the auxiliary evaluation result to obtain a final damage measurement result.
2. The human-computer cooperative autonomous level damage assessment method according to claim 1, wherein: the step of calculating a second damage evaluation result in step S2 includes:
image registration: respectively carrying out image registration and correction on the images before and after the damage so as to represent the images before and after the damage in the same coordinate system and obtain the registered images before and after the damage;
target segmentation: respectively carrying out target segmentation on the registered before-damage and after-damage images to obtain target areas in the images;
detecting image change: carrying out image change detection on the target area in the registered images before and after the damage to construct a difference image;
image change recognition: performing morphological processing and gray threshold segmentation on the difference image to obtain change information of the image;
and (3) damage assessment: and extracting damage information from the change information of the image, and counting and calculating the physical damage proportion and the functional damage degree to obtain a second damage evaluation result.
3. The human-computer cooperative autonomous level damage assessment method according to claim 1, wherein: the specific step of receiving the auxiliary evaluation result input from the outside in step S2 includes:
image registration: respectively carrying out image registration and correction on the images before and after the damage so as to represent the images before and after the damage in the same coordinate system and obtain the registered images before and after the damage;
detecting image change: carrying out image change detection on the target area in the registered images before and after the damage to construct a difference image;
image change recognition: performing morphological processing and gray threshold segmentation on the difference image to obtain change information of the image;
image enhancement: superposing the change information of the image on the original damaged image in an image enhancement mode;
and (3) auxiliary evaluation receiving: and receiving an auxiliary evaluation result input from the outside.
4. The human-computer cooperation autonomous level damage assessment method according to claim 2 or 3, characterized in that: the image registration adopts an image registration method based on SURF characteristic points.
5. The human-computer cooperative autonomous level damage assessment method according to claim 2, wherein: the target segmentation uses GrabCut algorithm based on graph theory algorithm.
6. The human-computer cooperation autonomous level damage assessment method according to claim 2 or 3, characterized in that: the image change detection uses an image numerical method, and comprises the following specific steps: dividing corresponding elements in an image matrix based on a ratio method, dually matching the relationship of two input images, limiting the gray scale range of an output image between 0 and 1, expanding the value range to [0, 255] through linear transformation, simultaneously adding 1 to the numerator and denominator of the ratio, and constructing the expression of calculating the difference image, wherein the expression specifically comprises the following steps:
wherein, a (x, y) and B (x, y) are the gray values of the two input images at a corresponding point respectively, and C (x, y) is the gray value of the difference image at the corresponding point.
7. The human-computer cooperation autonomous level damage assessment method according to claim 2 or 3, characterized in that: the morphological processing specifically uses ROI-based closed operation to process the difference image, and the threshold segmentation adopts an OTSU algorithm-based threshold segmentation method.
8. The human-computer cooperation autonomous level damage assessment method according to claim 1, 2 or 3, characterized in that: the specific step of calculating the first damage evaluation result in step S2 includes: the method comprises the steps of modeling the damage efficacy of various types of explosives in advance to obtain various types of explosive damage efficacy models, modeling the vulnerability of various types of targets to obtain various target vulnerability models, adding the various target vulnerability models into a model database, wherein the target vulnerability models comprise target appearance geometric modeling, physical vulnerability modeling used for modeling physical mechanical parameters of the targets and target functional vulnerability modeling used for analyzing the corresponding relation between the damage types and the damage results in various subsystems, components and units in the targets, calling the required target vulnerability models and explosive damage efficacy models from the model database to perform damage effect simulation, and calculating the first damage assessment result by using the preset simulation parameters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011453012.1A CN112446874A (en) | 2020-12-11 | 2020-12-11 | Human-computer cooperation autonomous level damage assessment method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011453012.1A CN112446874A (en) | 2020-12-11 | 2020-12-11 | Human-computer cooperation autonomous level damage assessment method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112446874A true CN112446874A (en) | 2021-03-05 |
Family
ID=74740354
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011453012.1A Pending CN112446874A (en) | 2020-12-11 | 2020-12-11 | Human-computer cooperation autonomous level damage assessment method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112446874A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113868759A (en) * | 2021-09-10 | 2021-12-31 | 中国兵器科学研究院 | Method, device, equipment and medium for establishing shelter vehicle equivalent model |
| CN114037650A (en) * | 2021-05-17 | 2022-02-11 | 西北工业大学 | Ground target visible light damage image processing method for change detection and target detection |
| CN114048637A (en) * | 2022-01-11 | 2022-02-15 | 中国人民解放军96901部队 | Meta-normal-form-based rapid target damage calculation method |
| CN114386766A (en) * | 2021-12-08 | 2022-04-22 | 西安近代化学研究所 | Main battle tank damage assessment method based on multi-index high-dimensional capacity space volume comprehensive analysis |
| CN115984590A (en) * | 2022-12-27 | 2023-04-18 | 中船重工奥蓝托无锡软件技术有限公司 | Target vulnerability assessment method and device based on image recognition and electronic equipment |
| CN116363491A (en) * | 2023-06-01 | 2023-06-30 | 南京理工大学 | Damage assessment method and system based on optimal target set and artificial intelligence |
| CN117392311A (en) * | 2023-09-28 | 2024-01-12 | 北京化工大学 | SAR image simulation method and device for damage scene |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104268879A (en) * | 2014-09-28 | 2015-01-07 | 民政部国家减灾中心 | Physical building quantity damage evaluation method based on remote sensing multi-spectral images |
| CN108733925A (en) * | 2018-05-22 | 2018-11-02 | 中国人民解放军军事科学院评估论证研究中心 | A method of power is injured based on numerical simulation assessment nature fragmentation type howitzer |
| CN110119580A (en) * | 2019-05-17 | 2019-08-13 | 中国人民解放军军事科学院国防工程研究院 | A kind of ground surface works target Damage assessment system and method |
-
2020
- 2020-12-11 CN CN202011453012.1A patent/CN112446874A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104268879A (en) * | 2014-09-28 | 2015-01-07 | 民政部国家减灾中心 | Physical building quantity damage evaluation method based on remote sensing multi-spectral images |
| CN108733925A (en) * | 2018-05-22 | 2018-11-02 | 中国人民解放军军事科学院评估论证研究中心 | A method of power is injured based on numerical simulation assessment nature fragmentation type howitzer |
| CN110119580A (en) * | 2019-05-17 | 2019-08-13 | 中国人民解放军军事科学院国防工程研究院 | A kind of ground surface works target Damage assessment system and method |
Non-Patent Citations (1)
| Title |
|---|
| 牛轶峰 等: "基于无人机侦察与模型仿真的人机协同毁伤评估", 《指挥与控制学报》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114037650A (en) * | 2021-05-17 | 2022-02-11 | 西北工业大学 | Ground target visible light damage image processing method for change detection and target detection |
| CN114037650B (en) * | 2021-05-17 | 2024-03-19 | 西北工业大学 | Ground target visible light damage image processing method for change detection and target detection |
| CN113868759A (en) * | 2021-09-10 | 2021-12-31 | 中国兵器科学研究院 | Method, device, equipment and medium for establishing shelter vehicle equivalent model |
| CN114386766A (en) * | 2021-12-08 | 2022-04-22 | 西安近代化学研究所 | Main battle tank damage assessment method based on multi-index high-dimensional capacity space volume comprehensive analysis |
| CN114048637A (en) * | 2022-01-11 | 2022-02-15 | 中国人民解放军96901部队 | Meta-normal-form-based rapid target damage calculation method |
| CN114048637B (en) * | 2022-01-11 | 2022-04-08 | 中国人民解放军96901部队 | Meta-normal-form-based rapid target damage calculation method |
| CN115984590A (en) * | 2022-12-27 | 2023-04-18 | 中船重工奥蓝托无锡软件技术有限公司 | Target vulnerability assessment method and device based on image recognition and electronic equipment |
| CN116363491A (en) * | 2023-06-01 | 2023-06-30 | 南京理工大学 | Damage assessment method and system based on optimal target set and artificial intelligence |
| CN116363491B (en) * | 2023-06-01 | 2023-08-11 | 南京理工大学 | Damage assessment method and system based on optimal target set and artificial intelligence |
| CN117392311A (en) * | 2023-09-28 | 2024-01-12 | 北京化工大学 | SAR image simulation method and device for damage scene |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112446874A (en) | Human-computer cooperation autonomous level damage assessment method | |
| CN113033520B (en) | Tree nematode disease wood identification method and system based on deep learning | |
| CN110363151A (en) | Based on the controllable radar target detection method of binary channels convolutional neural networks false-alarm | |
| CN108921030B (en) | SAR automatic target recognition method | |
| CN109255286B (en) | Unmanned aerial vehicle optical rapid detection and identification method based on deep learning network framework | |
| Shi et al. | Objects detection of UAV for anti-UAV based on YOLOv4 | |
| CN112749761A (en) | Enemy combat intention identification method and system based on attention mechanism and recurrent neural network | |
| CN110334584B (en) | Gesture recognition method based on regional full convolution network | |
| CN108256436A (en) | A kind of radar HRRP target identification methods based on joint classification | |
| Peng et al. | Battlefield image situational awareness application based on deep learning | |
| CN112288026B (en) | Infrared weak and small target detection method based on class activation diagram | |
| CN109255339B (en) | Classification method based on self-adaptive deep forest human gait energy map | |
| CN112861257A (en) | Aircraft fire control system precision sensitivity analysis method based on neural network | |
| CN112101189A (en) | SAR image target detection method and test platform based on attention mechanism | |
| CN112613504A (en) | Sonar underwater target detection method | |
| Guo et al. | Aircraft detection in high-resolution SAR images using scattering feature information | |
| CN115690545B (en) | Method and device for training target tracking model and target tracking | |
| Tian et al. | Performance evaluation of deception against synthetic aperture radar based on multifeature fusion | |
| CN115661576A (en) | Method for identifying airplane group intention under sample imbalance | |
| CN114049551B (en) | ResNet 18-based SAR raw data target identification method | |
| CN112215902B (en) | Eagle eye-imitated midbrain loop return inhibition mechanism unmanned target machine target detection method | |
| CN115457423A (en) | Flame smoke detection method combining efficient sampling enhancement | |
| CN114296067A (en) | Pulse Doppler radar low-slow small target identification method based on LSTM model | |
| Tang et al. | Target recognition method of laser imaging fuze based on deep transfer learning | |
| CN114943251B (en) | Unmanned aerial vehicle target recognition method based on fusion attention mechanism |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210305 |
|
| RJ01 | Rejection of invention patent application after publication |