A UAV target detection system, a target detection method, and a computer-readable storage medium

Abstract

The present invention belongs to the technical field of automated target detection, and discloses an unmanned aerial vehicle target detection system, a target detection method, and a computer-readable storage medium. The unmanned aerial vehicle target detection system includes an unmanned aerial vehicle system for carrying an optoelectronic payload system to an observation position and hovering at a fixed point; an optoelectronic payload system for performing fixed-point observation tasks and performing fixed observation of a target range area; a wireless image transmission system for transmitting observation video images using an HHLM type transmission module; a wireless remote control system for controlling the unmanned aerial vehicle system and the optoelectronic payload system, and a target detection integrated task station for receiving, storing, forwarding, and processing video images. The present invention provides a marine target detection system, which is of great significance to improving the level of testing technology and testing equipment, and to improving the actual combat capability and training level of destroyer and frigate naval gun firing. At the same time, it has a strong practical significance for ensuring the quality of naval gun live-fire firing and adapting to battlefield needs.

Description

Technical Field

[0001] The present invention belongs to the technical field of automated target detection, and in particular relates to an unmanned aerial vehicle target detection system, a target detection method, and a computer-readable storage medium. Background Art

[0002] At present, since the 21st century, maritime security issues have attracted more and more attention from all countries. Disputes over maritime territorial and maritime rights and interests between neighboring countries, rampant pirate attacks, and safety issues of free navigation at sea have made all countries gradually pay attention to the construction of naval modernization. Live-fire shooting is an important project in naval militarization training. In the competitive assessment of live-fire shooting, fast and accurate live-fire shooting effect evaluation is the key to completing the comprehensive evaluation of combat performance. The daily training level is judged by the hit rate and miss rate after the naval gun shooting, so it is very necessary to measure it accurately and scientifically. The sea shooting range is different from the land shooting range. The target test traces on the land shooting range are easy to capture, while at sea, the shells falling into the water and stirring up a water column is a short process, and the test traces are fleeting and not easy to capture.

[0003] Therefore, the prior art has not yet applied to an automated target detection method at a maritime shooting range, nor has it been able to implement an automated, intelligent method or system for evaluating maritime naval gun firing.

[0004] Through the above analysis, the problems and defects of the existing technology are as follows: the existing technology has no automated target detection method applied to the offshore shooting range, nor is there a method or system that can realize automated and intelligent evaluation of naval gun firing. The traditional optical measurement method requires manual alignment of two sets of image data time, manual measurement, and orthogonal errors. It is labor-intensive and highly professional, and is not suitable for modern high-intensity training. At the same time, if the large-scale measurement and control equipment with advanced technology, complex structure and high precision in the shooting range is used to obtain data (missile impact deviation), the cost issue cannot be ignored.

[0005] The difficulty of solving the above problems and defects is as follows: From the implementation process of the existing main gun close-range counterattack shooting result determination, there are still the following shortcomings:

[0006] (1) The deployment of a large number of support forces makes coordination difficult. Two support forces, a reference ship and a target ship, are required, and multiple cameras need to be deployed, which means that multiple support personnel are needed. Therefore, constant coordination and communication between the support forces and support personnel is inevitable.

[0007] (2) The recording process is demanding and complex to implement. This is because only by using the two videos from the reference ship and the target ship can the dispersion error of the naval gun be calculated. At the same time, in order to ensure the consistency of the measured targets in the video images, the time base of the two cameras must be unified, and the recording must start at the same time. Otherwise, it is difficult to achieve objective post-video analysis.

[0008] (3) The estimation of dispersion error is slow and the accuracy is low. For the post-processing of the water column video, after confirming the consistency of the measurement target, the distance between the water column and the floating target is measured on the computer display screen using a ruler. After conversion and processing, the dispersion probability error of the naval gun is finally estimated, which is not only time-consuming and labor-intensive, but also has low accuracy.

[0009] The significance of solving the above problems and defects is as follows: the traditional miss amount assessment work is to use two orthogonal cameras to shoot the direction dispersion and distance dispersion from the shooting ship and the support ship respectively. After returning to the shore, the two sets of image data are manually aligned for a long time, and then the distance is measured on the image to calculate the dispersion error and evaluate the assessment results. The whole process is time-consuming and labor-intensive. The application prospect of the drone target detection system is broad. It is not only suitable for the target detection task of ship gun shooting against the sea, but also for ship gun shooting against the shore. With the continuous maturity of drone target detection technology, drone target detection system will be gradually promoted to replace the current target detection system. At the same time, the miss amount assessment of ship gun shooting is completed based on video images. The automatic target detection system of the ship gun shooting drone is used to obtain the video image of the relative position of the shell impact water column and the target from the high-altitude fixed point, and transmit it to the digital processing module. The extraction and positioning of the target and the extraction and positioning of the shell impact water column are completed through digital image processing technology, and the coordinate position of the shell impact water column is calculated. The dispersion error of the shooting is obtained to evaluate the shooting results, and finally the GUI of the miss amount evaluation system is completed. It shortens the evaluation time, reduces manual workload, and improves the accuracy and timeliness of the evaluation. Summary of the invention

[0010] In view of the problems existing in the prior art, the present invention provides an unmanned aerial vehicle target detection system, a target detection method, and a computer-readable storage medium.

[0011] The present invention is implemented as follows: an unmanned aerial vehicle target detection system, the unmanned aerial vehicle target detection system comprising:

[0012] UAV system, optoelectronic payload system, wireless image transmission system, wireless remote control system and integrated target detection task station;

[0013] The UAV system includes a quad-rotor UAV body, a propulsion device, a flight control device, and a power supply device; it is used to carry the optoelectronic payload system to the observation position and hover at a fixed point;

[0014] The optoelectronic load system is mounted on the quad-rotor UAV body; it includes a CCD camera, an infrared detection device and a pod; it is used to perform fixed-point observation tasks and conduct fixed observation of the shooting range area;

[0015] Wireless image transmission system, used to transmit observation video images using HHLM type transmission module;

[0016] Wireless remote control system, used to control the UAV system and optoelectronic payload system;

[0017] The target detection integrated task station is used for receiving, storing, forwarding and processing video images.

[0018] Further, the wireless remote control system comprises:

[0019] A receiving module, used for receiving control instructions;

[0020] A decoding module, used for decoding control instructions using a decoder;

[0021] The control module is used to control the UAV system and the optoelectronic payload system based on the decoded control instructions.

[0022] Another object of the present invention is to provide a drone target detection method applied to the drone target detection system, the drone target detection method comprising:

[0023] The photoelectric load is used to obtain a video image of a naval gun shell hitting a water column, and the video image of the naval gun shell hitting the water column is analyzed and processed by obtaining video information, obtaining an image sequence, playing video information, selecting images frame by frame, inputting a coordinate system rotation angle, extracting the water column and other means to obtain the hit rate and miss rate of the naval gun shooting, and obtain a shooting evaluation result.

[0024] Furthermore, the drone target detection method comprises the following steps:

[0025] Step 1: Generate a control command, and control the quadcopter drone equipped with the optoelectronic payload to fly to a safe height above the target and then hover based on the generated control command; at the same time, control the drone to adjust the shooting posture of the optoelectronic payload based on the generated control command;

[0026] Step 2: Based on the control instructions, the optoelectronic load is controlled to record the gun firing situation in the offshore shooting range, and the video image of the water column aroused by the projectile entering the water is continuously obtained to obtain the relative relationship between the water column and the target in the shooting range, that is, the dispersion of the gun firing;

[0027] Step 3: convert the acquired video image into an electrical signal, sample, quantize, and encode it using an analog / digital converter, and then convert it into a digital signal;

[0028] Step 4: preprocess the converted image to obtain clear and useful image information; and based on the obtained clear and useful image, calculate the relative position of the water column where the bullet hits the floating target, and calculate the projectile dispersion error to evaluate the miss amount of naval gun shooting.

[0029] Furthermore, the safety height calculation method includes:

[0030] First, analyze the hovering height of the drone based on the gun firing height:

[0031] H ≥ h;

[0032] Wherein, H represents the hovering height of the UAV; h represents the height when the projectile reaches the highest point; the height h when the projectile reaches the highest point is determined by the relationship between the firing angle, firing distance and firing height of the naval gun;

[0033] Secondly, the hovering height of the UAV is analyzed based on the observation range of the optoelectronic load: according to the type of artillery, the shooting method and the average shooting distance, the range of the judgment interval is determined according to the corresponding table of shooting performance evaluation parameters for floating targets and simulated targets; the hovering height of the UAV is determined according to the rectangular boundary of the judgment interval; the rectangular boundary of the judgment interval is the rectangular interval boundary of the effective hit area of ​​the naval gun when shooting at floating targets and simulated targets, according to the corresponding shooting performance evaluation parameter table;

[0034] Finally, the safe height of the UAV is determined based on the analysis results of the UAV's hovering height under ship gun firing, the rectangular boundary limit of the judgment interval, the field of view angle range of the camera, and the UAV's own flight height limit.

[0035] Furthermore, the preprocessing of the converted image to obtain clear and useful image information includes:

[0036] (1) Use a color feature-based segmentation method to segment the converted image and identify and extract the target area;

[0037] (2) Based on the target information obtained after segmentation, the least square method is used to perform curve fitting to determine the center and radius of the target for target positioning;

[0038] (3) The extracted and located target image is grayed, median filtered, and contrast enhanced.

[0039] Further, the calculating the relative position of the impact water column and the floating target based on the obtained clear and useful image includes:

[0040] The threshold is calculated by adaptive and iterative methods to segment the water column image, and the water column image with background removed is obtained; the water column image with background removed is binarized; and the improved centroid method is used to locate the water column.

[0041] Further, the segmentation of the water column image by calculating the threshold value by an adaptive and iterative method includes:

[0042] 1.1) Count the minimum gray value T of the water column image min , maximum gray value T max , calculate the binary mean as the initial threshold T:

[0043]

[0044] 1.2) Segment the image according to the threshold T, and obtain two pixel sets:

[0045] G 1 ={f(x, y)≥T}, G 2 ={f(x, y)≤T};

[0046] 1.3) Calculate the pixel set G 1 and G 2 The grayscale average μ 1 and μ 2 :

[0047]

[0048]

[0049] 1.4) According to μ 1 and μ 2 Calculate new threshold Repeat steps 1.2), 1.3), and 1.4) until the threshold T converges to a certain range.

[0050] Further, the positioning of the water column using the improved centroid method includes:

[0051] Perform edge detection on the water column, extract edge information of the water column, and calculate the centroid of the water column by performing arithmetic averaging based on the edge information of the water column.

[0052] Further, in step 4, the calculation of the projectile dispersion error and the evaluation of the miss amount of the naval gun shooting include:

[0053] (1) Calculate the dispersion error:

[0054] Based on the obtained position information of the target and the impact water column in the image, a bullet dispersion coordinate system with the target as the origin is established, and the dispersion error is calculated using the following formula:

[0055]

[0056]

[0057] in, In formula 4.25 (X i , Z i ) represents the actual coordinate value of each bullet impact water column, represents the center coordinates of the water column group, The value of is the average value of the actual coordinates of all the water columns of impact; 0.6745 represents the probability coefficient, and n represents the total number of projectiles;

[0058] (2) Comparison of calculated E X , E Z The value of and the standard value K of the impact point dispersion error determine the naval gun shooting performance, and the evaluation result of the naval gun shooting miss amount is obtained.

[0059] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the function of the drone target detection system.

[0060] In combination with all the above-mentioned technical solutions, the advantages and positive effects of the present invention are as follows: the present invention utilizes the automatic target detection system of the naval gun shooting drone to obtain the video image of the relative position of the water column of the naval gun shooting and the target from a high-altitude fixed point, transmits it to the digital processing module, and completes the extraction and positioning of the target and the extraction and positioning of the water column through digital image processing technology, calculates the coordinate position of the water column, obtains the scattered error of the shooting to evaluate the shooting performance, which is of great significance to improving the level of testing technology and testing equipment, and to improving the actual combat capability and training level of destroyer and frigate naval gun shooting. At the same time, it has a strong practical significance for ensuring the quality of naval gun live-fire shooting and adapting to battlefield needs.

[0061] The present invention adopts the HHLM type wireless image transmission system, and the maximum transmission distance can reach 50 kilometers. The HHLM type microwave image transmission system is a high-performance and high-quality wireless image transmission system specially designed for long-distance or environments without wired transmission conditions. The present invention itself is small in size and light in weight, can transmit high-quality video images in real time without distortion, has stable modulation and demodulation performance, transmits bright and clear images, and is easy to install and debug. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic diagram of a drone target detection system provided by an embodiment of the present invention.

[0063] Figure 2 It is a schematic diagram of the structure of a drone target detection system provided by an embodiment of the present invention;

[0064] In the figure: 1. UAV system; 2. Optoelectronic payload system; 3. Wireless image transmission system; 4. Wireless remote control system; 5. Target detection integrated mission platform.

[0065] Figure 3 It is a schematic diagram of the target detection method for an unmanned aerial vehicle provided in an embodiment of the present invention.

[0066] Figure 4 It is a flow chart of the target detection method of an unmanned aerial vehicle provided in an embodiment of the present invention.

[0067] Figure 5 The figure is a schematic diagram of the position of a target detection UAV provided by an embodiment of the present invention.

[0068] Figure 6 It is a graph of shooting elevation angle and shooting distance provided by an embodiment of the present invention.

[0069] Figure 7 It is a graph of shooting elevation angle and shooting height provided by an embodiment of the present invention.

[0070] Figure 8 It is a shooting curve diagram of different shooting elevation angles provided by an embodiment of the present invention.

[0071] Fig. 9 It is a schematic diagram of a determination interval provided by an embodiment of the present invention.

[0072] Fig.10 It is a detection schematic diagram of a target detection system provided by an embodiment of the present invention.

[0073] Fig.11 It is a schematic diagram of pinhole imaging provided by an embodiment of the present invention.

[0074] Fig.12 The present invention provides a flowchart for automatically extracting and locating the impact water column according to an embodiment of the present invention.

[0075] Fig.13 It is a grayscale histogram showing a single peak provided by an embodiment of the present invention.

[0076] Fig.14 It is a schematic diagram of an image coordinate system provided by an embodiment of the present invention.

[0077] Fig.15 It is a schematic diagram of the principle of calculating the water column coordinates of the impact provided by an embodiment of the present invention.

[0078] Fig.16 It is a schematic diagram of a bullet dispersion coordinate system provided by an embodiment of the present invention.

[0079] Fig.17 It is a planar position relationship diagram of a shooting ship and a UAV provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0080] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0081] In view of the problems existing in the prior art, the present invention provides an unmanned aerial vehicle target detection system, which is described in detail below with reference to the accompanying drawings.

[0082] like Figure 1-Figure 2 As shown, the drone target detection system provided by the embodiment of the present invention includes:

[0083] UAV system 1, optoelectronic payload system 2, wireless image transmission system 3, wireless remote control system 4 and integrated target detection task station 5;

[0084] The UAV system 1 includes a quad-rotor UAV body, a propulsion device, a flight control device, and a power supply device; and is used to carry the optoelectronic payload system to the observation position and hover at a fixed point;

[0085] The optoelectronic load system 2 is mounted on the quad-rotor drone body; it includes a CCD camera, an infrared detection device and a pod; it is used to perform fixed-point observation tasks and conduct fixed observation of the shooting range area;

[0086] A wireless image transmission system 3, used for transmitting observation video images using a HHLM type transmission module;

[0087] Wireless remote control system 4, used to control the UAV system and optoelectronic payload system;

[0088] The target detection integrated task station 5 is used for receiving, storing, forwarding and processing video images.

[0089] The wireless remote control system 4 provided in the embodiment of the present invention comprises:

[0090] A receiving module, used for receiving control instructions;

[0091] A decoding module, used for decoding control instructions using a decoder;

[0092] The control module is used to control the UAV system and the optoelectronic payload system based on the decoded control instructions.

[0093] like Figure 3 As shown, the drone target detection method provided by the embodiment of the present invention includes:

[0094] The photoelectric load is used to obtain a video image of a naval gun shell hitting a water column, and the video image of the naval gun shell hitting the water column is analyzed and processed by obtaining video information, obtaining an image sequence, playing video information, selecting images frame by frame, inputting a coordinate system rotation angle, extracting the water column and other means to obtain the hit rate and miss rate of the naval gun shooting, and obtain a shooting evaluation result.

[0095] like Figure 4 As shown, the drone target detection method provided by the embodiment of the present invention includes the following steps:

[0096] S101, generating a control instruction, and based on the generated control instruction, controlling the quad-rotor drone carrying the optoelectronic payload to fly to a safe height above the target and then hover; and at the same time, based on the generated control instruction, controlling the drone to adjust the shooting posture of the optoelectronic payload;

[0097] S102, based on the control instruction, controlling the optoelectronic load to record the gun firing situation in the offshore shooting range, continuously acquiring the video image of the water column aroused by the projectile entering the water, and acquiring the relative relationship between the water column and the target in the shooting range, that is, the dispersion of the gun firing;

[0098] S103, converting the acquired video image into an electrical signal, sampling, quantizing, and encoding the image using an analog / digital converter, and then converting the image into a digital signal;

[0099] S104, preprocessing the converted image to obtain clear and useful image information; and calculating the relative position of the water column where the bullet hits the floating target based on the obtained clear and useful image, and calculating the projectile dispersion error to evaluate the miss amount of naval gun shooting.

[0100] The safe height calculation method provided by the embodiment of the present invention includes:

[0101] First, analyze the hovering height of the drone based on the gun firing height:

[0102] H ≥ h;

[0103] Wherein, H represents the hovering height of the UAV; h represents the height when the projectile reaches the highest point; the height h when the projectile reaches the highest point is determined by the relationship between the firing angle, firing distance and firing height of the naval gun;

[0104] Secondly, the hovering height of the UAV is analyzed based on the observation range of the optoelectronic load: according to the type of artillery, the shooting method and the average shooting distance, the range of the judgment interval is determined according to the corresponding table of shooting performance evaluation parameters for floating targets and simulated targets; the hovering height of the UAV is determined according to the rectangular boundary of the judgment interval; the rectangular boundary of the judgment interval is the rectangular interval boundary of the effective hit area of ​​the naval gun when shooting at floating targets and simulated targets, according to the corresponding shooting performance evaluation parameter table;

[0105] Finally, the safe height of the UAV is determined based on the analysis results of the UAV's hovering height under ship gun firing, the rectangular boundary limit of the judgment interval, the field of view angle range of the camera, and the UAV's own flight height limit.

[0106] The preprocessing of the converted image provided by the embodiment of the present invention to obtain clear and useful image information includes:

[0107] (1) Use a color feature-based segmentation method to segment the converted image and identify and extract the target area;

[0108] (2) Based on the target information obtained after segmentation, the least square method is used to perform curve fitting to determine the center and radius of the target for target positioning;

[0109] (3) The extracted and located target image is grayed, median filtered, and contrast enhanced.

[0110] The method provided in the embodiment of the present invention for calculating the relative position of the impact water column and the floating target based on the obtained clear and useful image includes:

[0111] The threshold is calculated by adaptive and iterative methods to segment the water column image, and the water column image with background removed is obtained; the water column image with background removed is binarized; and the improved centroid method is used to locate the water column.

[0112] The method provided in the embodiment of the present invention for segmenting the water column image by calculating the threshold value by an adaptive and iterative method includes:

[0113] 1.1) Count the minimum gray value T of the water column image min , maximum gray value T max , calculate the binary mean as the initial threshold T:

[0114]

[0115] 1.2) Segment the image according to the threshold T, and obtain two pixel sets:

[0116] G 1 ={f(x, y)≥T}, G 2 ={f(x, y)≤T};

[0117] 1.3) Calculate the pixel set G 1 and G 2 The grayscale average μ 1 and μ 2 :

[0118]

[0119]

[0120] 1.4) According to μ 1 and μ 2 Calculate new threshold Repeat steps 1.2), 1.3), and 1.4) until the threshold T converges to a certain range.

[0121] The method for locating the water column impacting the bullet using the improved centroid method provided in the embodiment of the present invention includes:

[0122] Perform edge detection on the water column, extract edge information of the water column, and calculate the centroid of the water column by performing arithmetic averaging based on the edge information of the water column.

[0123] The calculation provided by the embodiment of the present invention to obtain the projectile dispersion error and to evaluate the miss amount of naval gun shooting includes:

[0124] (1) Calculate the dispersion error:

[0125] Based on the obtained position information of the target and the impact water column in the image, a bullet dispersion coordinate system with the target as the origin is established, and the dispersion error is calculated using the following formula:

[0126]

[0127]

[0128] in, In formula 4.25 (X i , Z i ) represents the actual coordinate value of each bullet impact water column, represents the center coordinates of the water column group, The value of is the average value of the actual coordinates of all the water columns of impact; 0.6745 represents the probability coefficient, and n represents the total number of projectiles;

[0129] (2) Comparison of calculated E X , E Z The value of and the standard value K of the impact point dispersion error determine the naval gun shooting performance, and the evaluation result of the naval gun shooting miss amount is obtained.

[0130] The technical solution of the present invention is further described below in conjunction with specific embodiments.

[0131] Embodiment 1:

[0132] 1. Composition and working principle of UAV target detection system

[0133] 1.1 System composition

[0134] The components of the UAV target detection system include the UAV system, optoelectronic payload, wireless image transmission system, wireless remote control system and integrated target detection mission platform.

[0135] Unmanned aerial vehicle (UAV) is an unmanned aircraft controlled by a radio remote control device and a self-contained program control device. The UAV platform system is the aerial flight part of the UAV system, including the aircraft body, propulsion device, flight control device, and power supply device. The airborne part of the communication data link and the mission payload are also installed on the aircraft. The present invention uses a quad-rotor UAV and adopts aviation lithium batteries to provide power. This is mainly due to the fact that quad-rotor UAVs are widely used, have a good technical foundation, low cost, good maneuverability, strong survivability, and low environmental requirements. They can overcome observation work that is difficult to perform manually and are easy to use. Therefore, it is convenient to hover and record the water column video of the impact point. In this system, an optical measurement platform can be carried to complete the aerial fixed-point observation task and realize fixed observation of the target area.

[0136] The photoelectric load consists of a CCD camera, an infrared detection device and a pod. CCD (Charge Coupled Device) is made of a highly sensitive semiconductor material that can convert light into electric charge, which is converted into digital signals through an analog-to-digital converter chip. The digital signals are compressed and stored in the flash memory or built-in hard disk card inside the camera, so the data can be easily transmitted to the computer. It is carried on a four-rotor drone to record the firing of naval guns in the offshore shooting range, continuously shoot the water column stirred up by the projectile entering the water, and obtain the relative relationship between the water column and the target in the shooting range, that is, the dispersion of the naval gun firing.

[0137] The wireless image transmission system is used for timely transmission of images. It converts the video image information captured by the unmanned aerial vehicle optical measurement platform into electrical signals, which are converted into digital signals after being sampled, quantized, and encoded by an analog / digital converter and sent to a dynamic memory for temporary storage. Then, the effective data is wirelessly sent to the information receiving and processing system in real time via a wireless data transmission chip for later image processing and data calculation. The present invention adopts an HHLM type wireless image transmission system with a maximum transmission distance of 50 kilometers. The HHLM type microwave image transmission system is a high-performance, high-quality wireless image transmission system specially designed for long-distance or non-wired transmission environments. The present invention itself is small in size and light in weight. It can transmit high-quality video images in real time without distortion. The modulation and demodulation performance is stable, the transmitted image is bright and clear, and the installation and debugging are simple.

[0138] The function of the wireless remote control system is to control the UAV and the optoelectronic payload. The ground control center issues commands, which are transmitted to the receiving module of the wireless remote control system. The decoder reads the commands to control the UAV and the optoelectronic payload.

[0139] The target inspection integrated task station is responsible for receiving, storing and forwarding video images and effectively processing them. It is mainly composed of a PC and some image processing and numerical calculation software. After receiving the information from the image transmission system, it effectively pre-processes the image, removes interference and noise, extracts clear and useful image information, and uses this information to calculate the relative position of the impact water column and the floating target according to a reasonable mathematical model, and calculates the projectile dispersion error.

[0140] 1.2 Working Principle

[0141] The traditional evaluation method is relatively cumbersome, with large accidental errors and heavy workload for support personnel. The automatic target detection system for naval gun shooting drones can change the shortcomings of traditional target detection work from video image information acquisition to later data processing, reduce working time and improve efficiency. The basic process of the system is roughly divided into three steps, as shown in the schematic diagram Figure 1 As shown.

[0142] (1) The UAV is used as a platform to carry the optoelectronic payload. It flies to a safe altitude above the target and then hovers.

[0143] (2) The UAV’s optoelectronic payload adjusts its shooting posture to start working, and the target is imaged on the target surface through the camera.

[0144] (3) The video image information of the naval gun shell hitting the water column is transmitted in real time to the integrated task platform of the target ship through the wireless transmission module for real-time storage and forwarding, and then transmitted to the image acquisition and display board of the video signal through the wireless image transmission system. The video image is then digitized and the video image is quickly and accurately processed through the developed integrated task software to obtain the hit rate and miss rate of the naval gun shooting and evaluate the shooting results.

[0145] 2. Analysis of the target inspection mission area of ​​the UAV system

[0146] In order to ensure the safety of drones and meet the requirements of target inspection missions, the drone target inspection mission area must be clearly defined. When the drone system inspects targets, the drone hovers directly above the floating target, so the drone target inspection mission area is the area directly above the floating target, and the main task is to determine the height of this area.

[0147] 2.1 Analysis of UAV hovering height based on ship gun firing height

[0148] During the naval gun firing process, the trajectory of the projectile after leaving the barrel is approximately a parabola. When the drone target detection system is working, the drone hovers directly above the floating target. In order to ensure the safety of the drone, the height to ensure the safety of the drone must be determined based on the firing range of the naval gun to prevent the drone from being shot down by the projectile. Figure 5 As shown. Figure 5It can be seen that UAVs are almost not threatened in the horizontal direction. The safety threat of UAVs mainly comes from the projectiles that reach directly above the floating target. Therefore, the choice of the hovering height of the UAV is crucial during the naval gun shooting process. Once the position is chosen incorrectly, serious accidents will occur, putting the target inspection work into a passive situation.

[0149] Analyze the relationship between the firing angle, firing range and firing height of the naval gun to find the optimal hovering height.

[0150] It is known that the initial velocity of the 76mm main gun projectile out of the barrel is 980 meters per second, the gravity acceleration g = 9.8, the firing elevation angle is α, and the range of the naval gun firing angle is known to be -15° to 85°. Using Matlab software for simulation, we can obtain the graphs of firing elevation angle and firing distance, firing elevation angle and firing height, and firing distance and firing height at different firing elevation angles. Based on the graphs and simulation results, we can analyze and determine the safe firing range of the naval gun when firing at a floating target.

[0151] First, Matlab is used to simulate and obtain the curve of shooting elevation angle and shooting distance as shown below: Figure 6 As shown. Figure 6 From the graph of firing elevation angle and firing distance, it can be seen that when the firing elevation angle is 45°, the firing distance of the naval gun is the largest; when the distance between the ship and the floating target is 3 nautical miles (5.556 km), and the firing elevation angle is between 0° and 3°, the firing distance is within 10 kilometers, which can meet the shooting requirements under this condition.

[0152] Then use Matlab to simulate and get the curve of shooting elevation angle and maximum shooting height, such as Figure 7 As shown in Figure 744, the elevation angle and distance curve shows that as the shooting elevation angle increases, the maximum shooting height also increases; between 0° and 10°, the maximum shooting height does not change significantly.

[0153] According to Figure 6 Shooting elevation angle and shooting distance curve Figure 7 The analysis results of the shooting elevation angle and maximum shooting height curve are obtained by using Matlab simulation to obtain the corresponding curves of projectile shooting distance and shooting height at different shooting elevation angles. The curves are shown in the figure. Figure 8 shown.

[0154] according to Figure 8 The shooting curves at different shooting elevation angles and the simulation results are used to obtain the parameters of the maximum height that the projectile can reach and the height when it reaches directly above the floating target under different shooting elevation conditions. The parameter table is shown in Table 1.

[0155] Table 1 Projectile shooting distance and height parameters

[0156]

[0157] According to different shooting elevation angles, projectile shooting distance, height parameter table and curve graph, we can intuitively get the danger zone of naval gun shooting, and thus get the firing range of naval gun in height. According to the needs, the distance of naval gun shooting this time is 3 nautical miles (5.556 km). Figure 7 The firing curves at different firing elevation angles can be used to derive the firing range of the naval gun. The firing range is based on the launching ship and is an area below 1.5 times the maximum firing height of the naval gun. Based on the above information, the maximum firing height is 40 meters, so the minimum height of the naval gun's firing range is 1.5 times the maximum firing height, that is, 40×1.5=60 meters.

[0158] Continue to calculate the minimum safe height of the drone. When the drone inspects the target, it hovers directly above the floating target. It is known that the trajectory of the projectile is approximately a parabola. Assuming that the height of the projectile when it reaches the highest point is h, and the hovering height of the drone is H, in order to ensure the safety of the drone, H ≥ h should be set. When the distance between the ship and the floating target is 3 nautical miles (5.556 km), the minimum firing range of the 76 mm naval gun is 60 meters. In order to ensure that the drone is free from the threat of projectiles, the minimum safe hovering height of the drone is determined to be 60 meters, and the drone is directly above the floating target.

[0159] 2.2 Analysis of UAV hovering height based on optoelectronic payload observation range

[0160] The hovering height of the UAV based on the observation range of the optoelectronic payload refers to the height that meets the shooting effect requirements and the shot image contains the elements required for target detection. Therefore, when determining the working height of the UAV, it is not enough to only consider the height that ensures the safety of the UAV, but also the working height of the UAV that can meet the shooting mission requirements.

[0161] The concept of "decision matrix" is introduced here. The rectangular boundary of the "decision interval" is the rectangular interval boundary of the effective hit area determined by the corresponding shooting performance evaluation parameter table when the naval gun fires at floating (reflector) targets and simulated targets.

[0162] According to the type of artillery, shooting method and average shooting distance, and the corresponding table of shooting performance evaluation parameters for floating (reflector) targets and simulated targets, the range of the "judgment interval" is determined. The schematic diagram of the "judgment interval" is as follows Fig. 9 shown.

[0163] Fig. 9In the table, 2X is the distance length of the rectangular judgment interval, and 2Z is ​​the direction length of the rectangular judgment interval. In the parameter table for shooting performance evaluation of sea floating targets, half values ​​X and Z of the distance and direction lengths of the rectangular judgment area are listed. The parameter table for shooting performance evaluation of sea floating targets is shown in Table 2.

[0164] Table 2 Evaluation parameters for shooting at sea floating targets

[0165]

[0166] According to this shooting mission, referring to the parameter table, we can get the boundary value of the "determination interval" rectangular box when the shooting distance is 3 nautical miles (5.556 km), distance X = 114.43 meters, direction Z = 32.651 meters.

[0167] Taking into account the camera's shooting angle and shooting effect, the drone's hovering height must be determined according to the boundary of the "determination interval" rectangular box. Fig.10 As shown, the field of view of the camera is known to be 94° (i.e. ADB = 94°), so half the viewing angle is ACO = BCO = 1 / 2ACB = 47°. During the live-fire assessment of naval guns, the effective area for evaluation is the area of ​​the "decision interval" rectangular frame centered on the floating target O, and the "decision interval" rectangular frame is inscribed with ⊙O.

[0168] Based on the above known conditions, the height of OC can be easily calculated to determine the hovering height of the drone. Using the Pythagorean theorem: c 2 =a 2 +b 2

[0169] Where a = X = 114.43 meters

[0170] b=Z=32.651 m

[0171] have to

[0172] That is, the radius of ⊙OD is equal to 119 meters, OA=OB=OD=OE=OF=OG=119 meters.

[0173] According to the camera's field of view, the length of OC can be calculated using the tangent theorem:

[0174] Where a = OA = 119 meters

[0175] θ=∠ACO=47°

[0176] have to

[0177] That is, OC = b = 111 meters. Therefore, the hovering height of the drone is 111 meters.

[0178] If a water column is observed to be stirred up by a projectile falling outside the rectangular frame area of ​​the "determination interval", the camera's visible range should be expanded. Assuming that the length and width of the visible area are 1.5 times the length and width of the rectangular frame of the "determination interval", under this condition, the boundary value of the rectangular frame is distance X = 1.5*X = 171.645 meters, direction Z = Z = 48.98 meters. According to the method of solving the hovering height of the drone in the rectangular frame area of ​​the "determination interval", the hovering height of the drone under this condition can be obtained by the same logic. Finally, the solution is OC = 166.5 meters, that is, the hovering height of the drone is 166.5 meters.

[0179] The maximum flight altitude of the UAV is 500 meters. Taking into account the height to ensure the safety of the UAV, the rectangular boundary limit of the "determination interval", the field of view of the camera and the flight altitude limit of the UAV itself, the hovering altitude of the UAV when working is finally set at 111 meters.

[0180] 3. Target extraction and positioning and water column measurement

[0181] To achieve the miss-to-target assessment of naval gun shooting based on video images, it is necessary to extract target image information and water column image information from the impact water column image. The target of naval gun firing at sea is a red spherical floating target with a diameter of three meters. In the image, the color contrasts significantly with the ocean background. Therefore, the RGB image of the water column obtained by the drone can be directly used to segment the color area of ​​interest, that is, to extract and locate the target.

[0182] If the water column is to be extracted and located, the input water column image must be preprocessed to obtain clear target information. This chapter first explains the target extraction and location method, the proportional relationship between image distance and actual distance, and preprocesses the water column image to obtain clearer water column information. Then, a threshold segmentation method different from the target extraction method is used to extract the water column, realizing the automatic extraction and location of the water column under simple and complex backgrounds.

[0183] 3.1 Target extraction and positioning and data preprocessing

[0184] Image segmentation is an important step in image processing, which directly determines the quality of subsequent positioning accuracy. First, the target must be extracted and separated from the background in the image. Since the red drift target used for naval gun shooting has a clear color difference from the ocean background, based on the consistency of the target color and the large contrast between the target and the background color, the target can be extracted by directly segmenting the area within the target color range in the RGB image. Therefore, the present invention uses a segmentation method based on color features to segment the image.

[0185] The discrimination rule adopted by the present invention for target extraction is to discriminate a certain pixel as red when the R component among R, G, and B is obviously not less than the other components, and to control the color of the discrimination condition by setting a threshold. The most critical and most difficult to strictly define part of the region-based image segmentation mentioned in the present invention is the RGB color judgment. If the threshold is set too high, the range of target color variation will be reduced, and the tolerance for color distortion caused by video recording will be reduced; if the threshold is set too low, the range of target color variation will be increased, and the target that may be segmented will be larger or an area unrelated to the target will be segmented.

[0186] Then, the position of the target needs to be further determined. Only after the position of the target is determined can the relative position relationship between the impact point and the target be determined and the dispersion error be calculated. The target information obtained after image segmentation is an approximate circle. The positioning of the target is to obtain the center and radius of the target.

[0187] Circle detection is a very important application in digital image processing. There are many methods for extracting the center of a circle or a quasi-circular target, such as the three-point circle determination method, Hough transform method, curve fitting method, etc.

[0188] (1) The three-point circle method is a method for representing the boundary of an external region. The algorithm only focuses on the boundary information of the target circular region. In practical applications, the disadvantage of this algorithm is that the selection of any three points on the boundary of the target circular region will lead to strong randomness in the calculation, which will have a great impact on the calculation results. However, its advantages are that the algorithm can process incomplete or partially defective target circular regions, and the algorithm is relatively simple, easy to code and implement, and has a small amount of calculation.

[0189] (2) Hough transform is a clustering classification method that performs clustering operations on pixel information with certain relationships in the image space and searches for cumulative corresponding points in the parameter space that can connect these pixels in a certain analytical form. The advantages of this algorithm are strong noise resistance and high calculation accuracy. However, its disadvantage is that it requires a large amount of data to be stored, which results in a slow calculation speed of the algorithm.

[0190] (3) The surface fitting method is also a method for representing the boundary of the external region. It is usually based on the least squares method to find the best fitting point set and iterate the solution. The advantage of this algorithm is that it has high accuracy and takes into account the influence of the pixels of each boundary point of the target circular area. It also improves the problem that the selection of three points in the three-point circle determination method is too flexible and the randomness is difficult to control. However, the disadvantage of this algorithm is that when the target circular area is large and there are many boundary pixels in the target area, the algorithm has a large amount of calculation, which is not conducive to rapid center positioning and affects the execution efficiency of the algorithm.

[0191] Based on the comparative analysis of the above methods for extracting the center of various circles or quasi-circles, since the number of target pixels is not large after regional image segmentation during the target extraction process, and the target will be partially obscured by the waves, in order to meet the system's requirements for target position speed and accuracy, the present invention uses the least squares method to perform curve fitting algorithm for the detection of the target center.

[0192] After finding a circular arc, the corresponding parameters are calculated using the circle equation, which is shown in Equation 3.1.

[0193] (xx c ) 2 +(yy c ) 2 =r 2 (Equation 3.1)

[0194] Where (x c ,y c ) is the center of the circle, and the formula 3.2 is obtained by expanding it.

[0195] x 2 +y 2 +ax+by+c=0 (Formula 3.2)

[0196] Where a = -2x c , b = -2y c , c = x c 2 +y c 2 -r 2 , use the least squares method to calculate the parameters a, b, c, and thus find the radius of the circle:

[0197] The target is fitted according to the target center coordinates and radius obtained by the least squares method, and the obtained radius size fits the target radius well.

[0198] After completing the target extraction and positioning, the acquired image information must be preprocessed before extracting the impact water column. Image preprocessing is the basic process of image processing. The water column video image after shooting is obtained through the acquisition system. In order to measure the miss amount of the naval gun, the target impact water column is extracted to overcome some difficulties in the image, especially for the case where the difference between the impact water column and the background in the image is small, so that the extraction of the impact water column target is easier, so the impact water column image should be preprocessed.

[0199] The image preprocessing process mainly includes image grayscale, median filtering to eliminate noise and enhance image contrast, etc., in preparation for the identification of the bullet impact water column.

[0200] 3.2 Relationship between image distance and actual distance

[0201] To evaluate the miss of the shells, it is necessary to calculate the impact deviation based on the actual distance. According to the working requirements of the system, the optical measurement equipment shoots vertically downward from a high altitude to obtain images including the entire offshore range area and the impact water column of the offshore buoy. Based on the principles of perspective geometry and photogrammetry, the relationship between the pixels in the captured image and the actual distance is derived and calculated.

[0202] The present invention performs calculations based on the imaging principle of a pinhole camera, assuming that the image acquisition of the shooting range area by the camera is proportionally scaled, and does not consider the error caused by image distortion to the measured distance, because the closer the image of the bullet hitting the water column is to the edge of the image, the less it is affected by lens distortion.

[0203] like Fig.11 As shown, O is the center of the lens, a is the object distance, which is the distance from the optical center O to the object plane, and f is the focal length of the lens, which is the distance from the optical center O to the image plane.

[0204] It is known that the actual target is a sphere with a diameter of three meters, which can be used as a reference object for non-contact distance measurement. According to the number of pixels occupied by the target in the image, the actual length represented by each pixel can be obtained, that is, the size of the proportional relationship e.

[0205] Calculate the distance between the impact water column and the target by finding the Euclidean distance between the two target centers on the image and obtaining the actual distance based on the proportional relationship between the pixel size and the length of the actual object. Therefore, the relationship between the image pixel and the actual distance is:

[0206] In the previous description, the center coordinates and radius of the target have been determined. The target radius is 14.1032 pixels in size, and the actual radius of the target is 1.5 meters, so e=0.1064 meters / pixel.

[0207] 3.3 Automatic extraction and positioning of the impact water column

[0208] The so-called simple background is an environment that is more conducive to the identification of water columns under ideal circumstances. The ideal condition is that there is only one grayscale ocean background in the water column image. If there is a sky background, the grayscale of the sky is close to the grayscale of the water column, which will make the target segmentation more difficult; if the weather conditions are good and the wind speed is low, the shape of the water column will basically remain cylindrical, and the impact point can be represented by the center of the water column. In addition, the sea conditions are good and there are fewer surges on the sea surface, which will cause less interference to target extraction. The water column image obtained in this case is called a water column image with a simple background. Digital image processing is efficient and has small errors. The automatic extraction and positioning process design of the water column under a simple background is as follows. Fig.12 shown.

[0209] When a naval gun fires at sea, a water column is generated after the shell enters the water, so the impact deviation can be determined based on the position of the impact water column. The extraction of the impact water column mainly separates the target from the ocean background, and then filters out the influence of the waves in the ocean. Due to the difference in contrast between the water column and the background, the target and the background are segmented by an agreed grayscale threshold.

[0210] Assume that there are some bright objects on a dark background of the image f(x, y), so the grayscale of the object and background pixels is divided into two main modes. A common method to extract objects from the background is to select a threshold T to separate the two modes. Then, any point (x, y) that satisfies the condition f(x, y)>T is called an object point, and other points are called background points (and vice versa for dark objects on a bright background). The definition of the thresholded (binary) image g(x, y) is as shown in Equation 4.1.

[0211]

[0212] Therefore, the most difficult part of threshold segmentation is how to determine the threshold T. There are usually two types of threshold calculation methods: global threshold and basic adaptive threshold.

[0213] The global threshold is the most commonly used threshold calculation method. Especially when the grayscale histogram distribution of the image is bimodal, the global threshold can clearly separate the target and background components, and obtain a more ideal image segmentation effect. Basic adaptive threshold is a relatively basic image adaptive segmentation method. It generally performs threshold segmentation based on the characteristics of the grayscale changes of the image pixels themselves and their neighborhoods, and then realizes the binarization of the grayscale image. This method fully considers the characteristics of each pixel neighborhood, so it can generally better highlight the boundary between the target and the background.

[0214] The background of the water column is the sea, and the color of the water column is white. Ideally, the contrast between the water column and the background is relatively large, and the maximum inter-class difference method can be used to separate the water column and the background. However, due to the influence of the environment during image acquisition, such as the reflection of the ocean surface, the shooting angle of the optical measurement platform, and the density of the water column, the grayscale histogram presents a single peak. Fig.13 Therefore, the grayscale histogram of the water column shows a single peak.

[0215] According to the characteristics of the unimodal grayscale histogram, the adaptive threshold calculation steps adopted by the present invention are as follows:

[0216] (1) Initial value. Statistical minimum gray value T of the water column image min , maximum gray value T max , calculate the binary average as the initial threshold T.

[0217]

[0218] (2) Segmentation. Segment the image according to the threshold T and obtain two pixel sets.

[0219] G 1 ={f(x, y)≥T}, G 2 ={f(x, y)≤T} (Formula 4.6)

[0220] (3) Mean. Calculate the pixel set G 1 and G 2 The grayscale average μ 1 and μ 2 .

[0221]

[0222]

[0223] (4) Iteration. According to μ 1 and μ 2 Calculate new threshold Repeat steps 2, 3, and 4 until the threshold T converges to a certain range.

[0224] The impact water column image is segmented by calculating the threshold using an adaptive and iterative method.

[0225] After the image is binarized by the above method, the segmentation of the impact water column and the background is achieved well, but there are still some interference noises. These noises are caused by the fact that the grayscale levels of some areas in the background are close to those of the impact water column, which are false small target points formed after binarization, or the uneven grayscale of the impact water column causes some interference noises in the water column area of ​​the binary image. The existence of these noises affects the extraction of the impact water column and the location of the impact point in the next step. To solve these problems, mathematical morphological operations are used to filter the binary image.

[0226] Mathematical morphology theory is different from the traditional view on numerical modeling and analysis. It mainly uses structural elements to analyze and detect images. Mathematical morphology is composed of a series of algebraic operations in morphology, including 4 basic operations: dilation, erosion, opening, and closing.

[0227] The dilation operation can merge all the background points that the object touches into the object and expand the boundary of the target object outward. Using this operation, the space in the object can be filled; the erosion operation can eliminate the target boundary points and shrink the target boundary inward. Small and meaningless objects in the image can be eliminated by this operation. The filtering of binary images is achieved through two stages to eliminate noise: the first stage is to close the binary image in order to eliminate the black spots in the impact water column and make the impact water column area more complete. The second stage is to open the image in order to remove small interference noise in the background and remove the burrs on the edge of the impact water column, and smooth the edge of the impact water column.

[0228] After obtaining the image of the impact water column, it is necessary to find the coordinates of the center of the impact water column. After the previous image processing, the interference noise in the image is basically eliminated, and the impact water column can be located. Due to the influence of gravity and wind, the image of the impact water column taken from a high altitude is not necessarily a regular circle. The present invention uses the centroid method to find the center of gravity of the irregular image as the center of the impact water column. The centroid method refers to using the horizontal and vertical coordinates of the center of mass of the impact water column to represent its position in the image. Suppose an image f(x, y), x k is the x coordinate of the kth pixel, y k is the y coordinate of the kth pixel, f(x k ,y k ) is the gray value of the kth pixel. The total number of pixels at the centroid coordinate (x, y) is n, so the formula is

[0229]

[0230]

[0231] Since the grayscale values ​​of all pixels in the entire image are involved in the calculation, each pixel in the area not related to the target needs to be calculated, and for an image with irregular contours, the pixel position at the outermost edge has a greater impact on the centroid of the image. Therefore, the centroid algorithm can be improved by first finding the row and column positions of the edge pixels that meet the target, and then performing arithmetic averaging to obtain the centroid of the corresponding target image.

[0232] Edge refers to the set of pixels with step changes or roof changes in the grayscale of the surrounding pixels. Edge detection is mainly the measurement, detection and positioning of the grayscale changes of the image, that is, the image is segmented using the gradient based on the discontinuity of the image. The derivative is usually approximated by the difference in grayscale values ​​on a smaller neighborhood in the image. The following is a 3*3 neighborhood:

[0233] <![CDATA[Z 1 ]]> <![CDATA[Z 2 ]]> <![CDATA[Z 3 ]]> <![CDATA[Z 4 ]]> <![CDATA[Z 5 ]]> <![CDATA[Z 6 ]]> <![CDATA[Z 7 ]]> <![CDATA[Z 8 ]]> <![CDATA[Z 9 ]]>

[0234] Simple edge detection is to perform convolution operations on the original image using first-order differential operators, such as Roberts operator, Sobel operator, and Prewitt operator.

[0235] (1) Roberts operator is the simplest operator, which uses the local difference operator to find edges.

[0236] Its gradient is approximately

[0237] g x =Z 9 -Z 5 (Equation 4.11)

[0238] g y =Z 8 -Z 6 (Equation 4.12)

[0239] The convolution operator is: g x : and g y :

[0240] (2) The Sobel operator approximates the two components of the gradient as

[0241] g x =(Z 7 +2Z 8 +Z 9 )-(Z 1 +2Z 2 +Z 3 )) (Formula 4.13)

[0242] g y =(Z 3 +2Z 6 +Z 9 )-(Z 1 +2Z 4 +Z 7 ) (Formula 4.14)

[0243] The convolution operator is: g x : and g y :

[0244] (3) Prewitt operator

[0245] The Prewitt operator is computationally simpler than the Sobel operator, but the noise generated may be slightly larger. Its gradient is approximately

[0246] g x =(Z 7 +Z8 +Z 9 )-(Z 1 +Z 2 +Z 3 ) (Equation 4.15)

[0247] g y =(Z 3 +Z 6 +Z 9 )-(Z 1 +Z 4 +Z 7 ) (Equation 4.16)

[0248] Convolution operator: g x : and g y :

[0249] Through edge detection, the target edge information is extracted and the centroid position is obtained.

[0250] 4 Off-target Amount Assessment

[0251] 4.1 Calculation and analysis of dispersion error

[0252] Miss-target amount, also known as impact deviation amount. The shooting process of the weapon system will be affected and influenced by various accidental factors, which will continuously produce corresponding errors, causing the impact point to deviate from the target. When evaluating the miss-target amount, the dispersion error, that is, the probability deviation or mean square deviation of the impact point relative to the dispersion center is used to characterize the shooting density. When shooting at targets on the horizontal plane, the distance probability deviation Ex and the direction probability deviation Ez or the relative probability deviation Ex / x and Ez / x are used to characterize the projectile dispersion. The smaller the deviation of the impact point (explosion point) relative to the average impact point or dispersion center, the smaller the projectile dispersion of the weapon, that is, the higher the shooting density.

[0253] Before calculating the dispersion error, it is necessary to discuss the possible coordinate transformation problem, considering that the image coordinate system may not match the bullet dispersion coordinate system.

[0254] The image information obtained by the CCD camera is converted into a digital image after analog-to-digital conversion. The coordinates of the digital image are represented as (u, v), which represents the image coordinate value in pixels. Another coordinate system (x, y) is also required to represent the position of the pixel in the image in physical units. The origin of this coordinate system is a point in the image. 1 , the x-axis is parallel to the u-axis, and the y-axis is parallel to the v-axis.

[0255] In the (x, y) coordinate system, the origin O 1 It is usually located at the center of the image, which is the intersection of the image plane and the camera optical axis.1 In the (u, v) coordinate system (u 0 , v 0 ) position, the physical size of each pixel in the x-axis direction is Δ x , the physical size in the Y-axis direction is Δ y , then the relationship between the two coordinate systems can be expressed by any pixel in the image as shown in Equations 4.17 and 4.18.

[0256]

[0257]

[0258] In the present invention, it is necessary to establish an image coordinate system with the target as the origin, the u axis as the x axis, and the v axis as the y axis. 1 -xy.

[0259] The formula for calculating the coordinates of the water column after the bullet impact is derived below.

[0260] like Fig.15 As shown. Image coordinate system O 1 -xy, the center coordinate of the water column m is (x m ,y m ), the center coordinate of the impact water column M in the world coordinate system O′-X′Y (the X-axis and Z-axis of this coordinate system are not parallel to the impact distance scatter and direction scatter respectively) is (X′ M , Z′ M ), m is the image of M in the image. e is the ratio between the actual distance and the distance on the image.

[0261] X′ M =x m ·e (Equation 4.19)

[0262] Z′ M =y m ·e (Equation 4.20)

[0263] Since the target is far away from the shooting ship, the X-axis of the impact dispersion is difficult to determine, so there may be a situation where the photographic image coordinate system does not coincide with the impact dispersion coordinate system, that is, there is a process in which the impact water column coordinate system O′-X′Y′ rotates relative to the impact dispersion coordinate system O-XZ. Fig.16 As shown. (X M , Z M ) is the coordinate of the impact water column in the impact spread coordinate system.

[0264] Rotation formula for rectangular coordinate system

[0265] X M =X′ M ·cos(θ)+Z′ M·sin(θ) (Equation 4.21)

[0266] Z M =-X′ M sin(θ)+Z′ M ·cos(θ) (Equation 4.22)

[0267] Θ is the rotation angle of the image coordinate system relative to the world coordinate system reflected by the image coordinate system. This rotation angle can be calculated based on the heading of the drone, the heading of the shooting ship, and the shooting side angle, provided that the x-axis of the image coordinate system is in the same direction as the drone. The calculation formula for Θ is shown in Equation 4.23.

[0268] Θ=TC2-TC1-Q (Equation 4.23)

[0269] Where TC1 and TC2 are the headings of the shooting ship and the UAV, respectively, and Q is the shooting side angle. These parameters can be obtained from the UAV and the shooting ship.

[0270] By extracting and locating the target and the impact water column, the position information of the target and the impact water column in the image is obtained, and a projectile dispersion coordinate system with the target as the origin is established. The coordinate data of the impact water column derived from the formula for the impact water column coordinates in the previous section is used to calculate the dispersion error. The formula for calculating the dispersion error is shown in Equations 4.24 and 4.25.

[0271]

[0272]

[0273] In formula 4.24 In formula 4.25

[0274] (X i , Z i ) represents the actual coordinate value of each bullet impact water column, It represents the coordinates of the center of the water column group. Its value is the average of the actual coordinates of all the water columns. 0.6745 is the probability coefficient and n is the total number of projectiles.

[0275] According to the evaluation criteria for naval gun shooting performance in the naval gun shooting table, the dispersion error of the bullet impact point has a standard value K, compared with E X , E Z The shooting result can be determined by comparing the value with the standard value.

[0276] The calculation of the dispersion error has been completed. Based on the knowledge of error theory and combined with the entire system design process, the sources of the error are analyzed as follows:

[0277] (1) Preparation stage: The X-axis of the photoelectric payload image is not parallel or fixed to the heading direction of the UAV; the flight altitude is unreasonable, resulting in a poor shooting angle.

[0278] (2) Shooting stage: When the UAV optoelectronic payload is shooting, the UAV is far away from the control platform and the working electromagnetic environment is complex, which makes the optoelectronic payload of the UAV have certain deviations in its control during operation; the optoelectronic payload itself will produce errors during shooting, the camera is not accurate and the lens is distorted during the lens manufacturing process, or the vibration generated by the wind during shooting will be transmitted to the optical system through the pod's mount fixing point, which will affect the image quality. Secondly, during shooting, since the image obtained by the optoelectronic payload cannot be proportionally scaled to the actual sea area plane, errors will occur when calculating the water column dispersion distance of the bullet.

[0279] (3) Image transmission stage: The acquired image is affected by the performance of the image transmission system during the wireless transmission process. The quality of the image after transmission will cause errors in the subsequent image processing.

[0280] (4) Image processing stage: The quality of image preprocessing will affect the subsequent extraction accuracy of the target and the impact water column. The positioning accuracy of the target and the impact water column is directly related to the error size of the impact water column dispersion distance.

[0281] (5) Calculation stage: The video image and the actual sea area are not completely scaled in the same proportion. When converting the distance on the map to the actual distance, the calculation is based on the same proportion, which will cause certain errors. In addition, in the process of calculating the dispersion distance of the bullet impact water column, the determination of the number of significant digits and the rounding of the numbers will cause errors in the results.

[0282] 4.2 Design of shooting miss distance evaluation software based on MATLAB GUI

[0283] We want to use the rapid modeling and powerful computing power of MATLAB tools to complete the post-analysis of the target acquisition video, including the reading, information acquisition, image sequence acquisition, playback, pause, interactive impact point distance measurement and other functions of the naval gun shooting video. And use the graphical user interface (GUI) provided by MATLAB to realize a visual object-oriented operation interface, realize the selection, processing and result display of video images and frame-by-frame pictures required for the shooting miss amount assessment, as well as the calculation of scattered error and other functions.

[0284] The system stores the acquired video images of the naval gun firing in the form of video frames. When determining the impact point, images are selected from the video frames for water column image processing to locate the impact point. After all the impact points are located, the dispersion error is calculated to determine the shooting results.

[0285] The specific steps are as follows: Figure 3 shown.

[0286] In the design of the system, in order to remind users to operate according to the specifications and reduce operational errors, a system introduction section was added during the software design, and the operation process can be viewed from the software interface. At the same time, if the operation steps are different from the designed process, the system will issue a prompt message.

[0287] The GUI (Graph User Interfaces) platform provided by MATLAB is used to implement the design of the naval gun shooting miss amount evaluation system based on video images of the present invention. The platform provides a variety of control tools for interface design. Users can implement the operation interface required for the design in a convenient and fast manner in the friendly interactive mode provided by the platform. According to the implementation method of the GUI, it is carried out in the following three steps.

[0288] (1) Create a blank GUI.

[0289] (2) Layout and functional design of system modules.

[0290] After completing the interface design required by the system and saving the interface design, MATLAB GUI will generate 2 files related to the interface design.

[0291] .Fig file: This file includes the GUI's image window and a complete description of all sub-objects and the property values ​​of all objects, including user controls and coordinate axes.

[0292] .M file: It contains all the code needed to run the GUI, which can control the GUI and determine how the GUI responds to user operations.

[0293] (3) Compile the callback function.

[0294] Users can write callback functions required by GUI components within the M file framework generated by GUI design. The M file contains a series of sub-functions: main function, opening function, output function and callback function. It should be noted that the main function cannot be modified, otherwise it will cause the failure of GUI interface initialization.

[0295] Test the system according to the operating steps, run the program, enter the miss distance evaluation interface, and input the known rotation angles of the image coordinate system and the impact dispersion coordinate system. The ideal angle is 0. You can view the operating steps of the system by clicking "System Introduction":

[0296] After completing the GUI implementation of the miss distance assessment system, loading the video information and obtaining the video information and image sequence, you can view the video frame-by-frame image information in the current path.

[0297] Select the image information with clear impact water column for image processing, click the "Impact Water Column Extraction" button, the image display area will display the identified target, select the impact water column position with the mouse cursor, and display the mark in the display area, and obtain and display the direction and distance deviation of the impact point relative to the target. Figure 5-6 As shown, a water column extraction operation is performed.

[0298] The image data used in the GUI implementation is captured from the video images obtained from the target ship, reflecting the distance dispersion of the projectile. The result is 61.99m, which is close to the data obtained by traditional measurements and meets the requirements for measuring the miss distance of ship-to-sea gun firing.

[0299] Following the same operation process, select the pictures that best reflect the characteristics of the water column of the shooting times to extract the water column until all the water column information is obtained, and finally click "Dispersion Error Calculation" to obtain the dispersion error of this group of shooting. Then use the dispersion error data to determine the shooting results. 5.

[0301] The evaluation of the miss amount of naval gun shooting plays an indispensable role in the performance appraisal of naval gun weapons and military competition training for the navy. The present invention first introduces the composition and working principle of the drone target detection system and the drone target detection mission area, and focuses on completing the miss amount evaluation of naval gun shooting based on video images. The automatic target detection system of the naval gun shooting drone is used to obtain the video image of the relative position of the water column of the naval gun shooting and the target from a high-altitude fixed point, and transmit it to the digital processing module. The extraction and positioning of the target and the extraction and positioning of the water column are completed through digital image processing technology, the coordinate position of the water column is calculated, and the dispersion error of the shooting is obtained to evaluate the shooting performance. Finally, the GUI of the miss amount evaluation system is completed.

[0302] In the description of the present invention, unless otherwise specified, "plurality" means two or more than two; the orientations or positional relationships indicated by the terms "upper", "lower", "left", "right", "inner", "outer", "front end", "rear end", "head", "tail", etc. are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0303] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0304] 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 modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for detecting a target by using an unmanned aerial vehicle, characterized in that: The unmanned aerial vehicle target detection method comprises: The video image of the naval gun shell hitting the water column is obtained by using the photoelectric load, and the video image of the naval gun shell hitting the water column is analyzed and processed by obtaining video information, obtaining image sequences, playing video information, selecting images frame by frame, inputting coordinate system rotation angles, extracting the water column and other means to obtain the hit rate and miss rate of the naval gun shooting, and obtain the shooting evaluation result; The unmanned aerial vehicle target detection method comprises the following steps: Step 1: Generate a control command, and control the quadcopter drone equipped with the optoelectronic payload to fly to a safe height above the target and then hover based on the generated control command; at the same time, control the drone to adjust the shooting posture of the optoelectronic payload based on the generated control command; Step 2: Based on the control instructions, the optoelectronic load is controlled to record the gun firing situation in the offshore shooting range, and the video image of the water column aroused by the projectile entering the water is continuously obtained to obtain the relative relationship between the water column and the target in the shooting range, that is, the dispersion of the gun firing; Step 3: convert the acquired video image into an electrical signal, sample, quantize, and encode it using an analog / digital converter, and then convert it into a digital signal; Step 4: preprocess the converted image to obtain clear and useful image information; calculate the relative position of the water column where the bullet hits the floating target based on the obtained clear and useful image, and calculate the projectile dispersion error to evaluate the miss amount of naval gun shooting; The improved centroid method for locating the impact water column includes: Performing edge detection on the impact water column, extracting edge information of the impact water column, and calculating the centroid of the impact water column by performing arithmetic averaging based on the edge information of the impact water column; In step 4, the calculation of the projectile dispersion error and the evaluation of the miss amount of the naval gun shooting include: (1) Calculate the dispersion error: Based on the obtained position information of the target and the impact water column in the image, a bullet dispersion coordinate system with the target as the origin is established, and the dispersion error is calculated using the following formula: in, In formula 4.25 (X i , Z i ) represents the actual coordinate value of each bullet impact water column, represents the coordinates of the center of the water column group, The value of is the average value of the actual coordinates of all the water columns of impact; 0.6745 represents the probability coefficient, and n represents the total number of projectiles; (2) Comparison of calculated E X , E Z The value of and the standard value of the impact point dispersion error K determine the naval gun shooting performance, and obtain the naval gun shooting miss amount evaluation result. The unmanned aerial vehicle target detection system for implementing the unmanned aerial vehicle target detection method comprises: UAV system, optoelectronic payload system, wireless image transmission system, wireless remote control system and integrated target detection task station; The UAV system includes a quad-rotor UAV body, a propulsion device, a flight control device, and a power supply device; it is used to carry the optoelectronic payload system to the observation position and hover at a fixed point; The optoelectronic load system is mounted on the quad-rotor UAV body; it includes a CCD camera, an infrared detection device and a pod; it is used to perform fixed-point observation tasks and conduct fixed observation of the shooting range area; Wireless image transmission system, used to transmit observation video images using HHLM type transmission module; Wireless remote control system, used to control the UAV system and optoelectronic payload system; Target detection integrated task station, used for receiving, storing, forwarding and processing video images; The wireless remote control system comprises: A receiving module, used for receiving control instructions; A decoding module, used for decoding control instructions using a decoder; The control module is used to control the UAV system and the optoelectronic payload system based on the decoded control instructions.

2. The target detection method for an unmanned aerial vehicle according to claim 1, characterized in that: The safety height calculation method comprises: First, analyze the hovering height of the drone based on the gun firing height: H ≥ h; Wherein, H represents the hovering height of the UAV; h represents the height when the projectile reaches the highest point; the height h when the projectile reaches the highest point is determined by the relationship between the firing angle, firing distance and firing height of the naval gun; Secondly, the hovering height of the UAV is analyzed based on the observation range of the optoelectronic load: according to the type of artillery, the shooting method and the average shooting distance, the range of the judgment interval is determined according to the corresponding table of shooting performance evaluation parameters for floating targets and simulated targets; the hovering height of the UAV is determined according to the rectangular boundary of the judgment interval; the rectangular boundary of the judgment interval is the rectangular interval boundary of the effective hit area of ​​the naval gun when shooting at floating targets and simulated targets, according to the corresponding shooting performance evaluation parameter table; Finally, the safe height of the UAV is determined based on the analysis results of the UAV's hovering height under ship gun firing, the rectangular boundary limit of the judgment interval, the field of view angle range of the camera, and the UAV's own flight height limit.

3. The target detection method for an unmanned aerial vehicle according to claim 1, characterized in that: The preprocessing of the converted image to obtain clear and useful image information includes: (1) Use a color feature-based segmentation method to segment the converted image and identify and extract the target area; (2) Based on the target information obtained after segmentation, the least square method is used to perform curve fitting to determine the center and radius of the target for target positioning; (3) The extracted and located target image is grayed, median filtered, and contrast enhanced.

4. The target detection method for an unmanned aerial vehicle according to claim 1, characterized in that: The method of calculating the relative position of the impact water column and the floating target based on the obtained clear and useful image comprises: The threshold is calculated by adaptive and iterative methods to segment the water column image, and the water column image with background removed is obtained; the water column image with background removed is binarized; and the improved centroid method is used to locate the water column.

5. The target detection method using an unmanned aerial vehicle as claimed in claim 4, characterized in that: The segmentation of the impact water column image by calculating the threshold value by an adaptive and iterative method includes: 1.1) Count the minimum gray value T of the water column image min , maximum gray value T max , calculate the binary mean as the initial threshold T: 1.2) Segment the image according to the threshold T, and obtain two pixel sets: G1={f(x,y)≥T}, G2={f(x,y)≤T}; 1.3) Calculate the grayscale average μ1 and μ2 of the pixel sets G1 and G2: 1.4) Calculate the new threshold based on μ1 and μ2 Repeat steps 1.2), 1.3), and 1.4) until the threshold T converges to a certain range.

6. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the unmanned aerial vehicle target detection method according to any one of claims 1 to 5.

Images