A method for detecting a takeoff time of a target machine

Abstract

The application provides a detection method for the take-off time of a target aircraft, which can meet the inspection requirement for the take-off time of the target aircraft in the current target range training, realizes real-time dynamic marking of the target aircraft, and assists the system in commanding the movement of the target aircraft. In the target range training, the continuous movement of the camera causes the continuous update of the image background, so that only one algorithm cannot be used to detect the target aircraft. The application selects a target detection method according to the histogram type of the image, can meet the inspection requirement, and realizes real-time dynamic marking of the target aircraft. The application filters the detected target set, performs target merging, marks the target on the preprocessed image, and detects whether the take-off event occurs. When the average brightness value of the pixel points in the target region changes by more than a certain threshold value, it is judged whether the take-off event of the target aircraft occurs and the corresponding screenshot is obtained.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of system testing and verification, and particularly relates to a method for detecting a take-off moment of a target aircraft. BACKGROUND

[0002] A target range training command system, a forest fire prevention safety system, a city safety monitoring system, etc. are all based on image transmission technology, and information of target elements (objects such as target aircraft, missiles, vehicles, etc. that command and control personnel command control and pay attention to) is acquired through various high-definition cameras or infrared cameras. The information is transmitted and processed, and finally displayed in real time on a screen in the form of digital images. Command and control personnel master the current position, speed and distance information of the target elements through information viewing, searching and calculation, make judgments and decisions accordingly, and then issue command and control instructions to realize the orderly movement of the command and control objects and achieve the purpose of the command and decision system. For the sake of convenience, the above systems are collectively referred to as video command and control systems.

[0003] For a video command and control system, the purpose of image data display is to enable the commander to track and monitor the deployment and state in the current monitoring range in real time, and after algorithm processing, the commander can assist the command and decision personnel to timely master the situation and make judgments and decisions. A large amount of video image data cannot be processed and recognized by a specific algorithm, and the command and decision personnel cannot accurately locate and use it, so it is meaningless. Therefore, image processing is one of the most important capabilities of a video command and control system, and the detection capability for some special events, such as the take-off event, is the most comprehensive, most fundamental and most direct indicator to measure the capability of a video command and control system.

[0004] Regarding the image processing capability, it is generally considered to refer to image enhancement and target detection. The image enhancement algorithm can highlight more details of the image, improve the overall and local contrast of the image, make the enhanced image more consistent with the visual characteristics of the human eye and easy to identify by machines, and the target detection uses the image after image enhancement to extract specific targets from static images or videos. Target elements such as aircraft, missiles and vehicles are in motion, especially the motion speed of target aircraft such as target planes and missiles is high, and the direction and acceleration change sharply, so the video command and control system needs to detect the position and state of the target elements in real time according to the image information obtained, realizes dynamic marking, and cannot rely on the video director to find the target elements. At the same time, when the command personnel view the video image information, they must perform various operations such as recording specific event time, intercepting screen, switching video source, etc. In response to these operations, the entire process needs to consume resources and time to complete cooperatively. If the time required from operation to record completion is too long, it will inevitably affect the command and control personnel's grasp and judgment of the situation, so the real-time response capability of the video command and control system to this series of operations must also be considered. In addition, since the video command and control system has a playback function, it needs to quickly and accurately obtain all target element information during playback. The update speed of the situation information recorded by the director is too fast, which will cause information loss, errors and non-obviousness, so the director's record cannot be used to assist in the take-off task.

[0005] Therefore, there is an urgent need for a target aircraft take-off time detection method that can meet the current target range training requirements for target aircraft take-off time detection, real-time response to system operations, dynamic marking of target aircraft, and assisting system command of target aircraft movement. SUMMARY

[0006] Therefore, there is an urgent need for a target aircraft take-off time detection method that can meet the current target range training requirements for target aircraft take-off time detection, real-time response to system operations, dynamic marking of target aircraft, and assisting system command of target aircraft movement.

[0007] To achieve the above-mentioned purposes, the technical scheme of the present application is as follows:

[0008] A target aircraft take-off time detection method, comprising the following specific steps:

[0009] Step 1, pre-processing the target aircraft image of the current frame to obtain a pre-processed image.

[0010] Step 2, calculating the histogram of the pre-processed image.

[0011] Step 3, judging the type of the histogram according to the characteristic value t of the histogram, selecting a target detection method according to the type of the histogram, and obtaining a target set.

[0012] Step 4, filtering the target set obtained in step 3, and calculating the distance between two targets; when the distance between two targets is less than the distance threshold D and the brightness values are the same, merging the targets and executing step 5; judging whether there is a target aircraft in the merged target set, yes, executing step 5, otherwise updating the current frame image and returning to step 1 for continuous execution.

[0013] Step 5, marking the target on the preprocessed image, and executing step 6.

[0014] Step 6, detecting whether a take-off event occurs, yes, ending the process, otherwise updating the current frame image and returning to step 1 for continuous execution.

[0015] Further, the specific manner of step 1 is: inputting the target aircraft image into the detection model to remove the black edges and characters in the target aircraft image.

[0016] Further, the types of the histogram are seven, which are: type 1 is a single-peak and clean histogram, and the characteristic value t>0.9; type 2 is a single-peak and noisy histogram, and 0.8

[0017] Further, the specific manner of step 3 is: when the histogram belongs to one of types 1 to 6, executing step 3.1; when the histogram belongs to type 7, executing step 3.2.

[0018] Step 3.1, performing background modeling on the preprocessed image and subtracting it to obtain a foreground image; removing the noise and interference targets in the foreground image, solving the connected domain of the image to obtain a target set; performing initial filtering on the target set, judging whether the size, brightness and position of the target after the initial filtering meet the preset values, yes, executing step 4, otherwise updating the current frame image and returning to step 1 for continuous execution.

[0019] Step 3.2, binarizing the preprocessed image to obtain a binarized image; solving the connected domain of the binarized image to obtain a target set, and executing step 4.

[0020] Further, in step 3.1, an adaptive Gaussian mixture background modeling algorithm is used to model the background, and the background is subtracted from the preprocessed image to obtain a foreground image.

[0021] Further, in step 3.1, an image morphological algorithm is used to remove the noise and interference targets in the foreground image.

[0022] Beneficial effects:

[0023] 1. The application provides a target aircraft take-off time detection method, which can meet the demand for target aircraft take-off time detection in current target range training, realizes real-time dynamic marking of the target aircraft, and assists the system in commanding the movement of the target aircraft. In target range training, the continuous movement of the camera causes the image background to be continuously updated, so that only one algorithm cannot be used to detect the target aircraft. The application selects a target detection method according to the histogram type of the image, meets the detection demand, and realizes real-time dynamic marking of the target aircraft. The application filters the detected target set, merges the targets, marks the targets on the preprocessed image, and detects whether a take-off event occurs. When the average brightness value of the target region pixel points changes by more than a certain threshold, it is judged whether it is a target take-off event and the corresponding screenshot is obtained.

[0024] 2. The application preprocesses the target aircraft image, removes the black edges and characters that affect the detection accuracy, and improves the detection precision.

[0025] 3. The application uses two target detection methods to preliminarily extract the target set of the preprocessed image with different feature values, so as to improve the detection precision. For the preprocessed image of type 1-6, the background, noise and interference targets are filtered, the connected domain is solved, the target set is obtained and initially filtered; for the preprocessed image of type 7, the connected domain is solved after binarization, and the target set is obtained.

[0026] 4. The application uses an adaptive Gaussian mixture background modeling algorithm to model the background, subtracts the background from the preprocessed image to obtain a foreground image, and improves the detection precision.

[0027] 5. The application uses an image morphological algorithm to remove the noise and interference targets in the foreground image, and improves the detection precision.

[0028] 6. The method has specific steps, is clear, and has strong operability.

[0029] 7. The application considers all factors that may affect the detection accuracy and faces specific application requirements, has good generalization of the detection algorithm, and can also detect the take-off state of a missile. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 The application provides a method flowchart. DETAILED DESCRIPTION

[0031] The application will be described in detail below in combination with the drawings and examples.

[0032] As shown in the drawings, Figure 1 the application provides a target aircraft take-off time detection method, and the specific steps include:

[0033] Step 1, pre-process the target image of the current frame to obtain a pre-processed image.

[0034] Step 2, calculate the histogram of the pre-processed image.

[0035] Step 3, judge the type of the histogram according to the characteristic value t of the histogram; when the histogram belongs to one of type 1 to type 6, execute step 3.1; when the histogram belongs to type 7, execute step 3.2. Type 1 is a single-peak and clean histogram, and the characteristic value t>0.9; type 2 is a single-peak and noisy histogram, and 0.8

[0036] Step 3.1, perform background modeling by using adaptive Gaussian mixture background modeling algorithm, subtract the background from the pre-processed image to obtain a foreground image. Use image morphological algorithm to remove the noise and interference targets in the foreground image. Solve the connected domain of the foreground image to obtain a target set. Perform primary filtering on the target set, and the filtering condition is the size, brightness and position of the target. Judge whether the size, brightness and position of the target after primary filtering meet the preset value, yes, execute step 4, otherwise, update the current frame image and return to step 1 for continuous execution.

[0037] Step 3.2, binarize the pre-processed image to obtain a binarized image. Solve the connected domain of the binarized image to obtain a target set, and execute step 4. The specific way of binarization is: set the pixel point threshold of the pre-processed image as K=178, if the current pixel point value is greater than K, set the pixel point value as 255, if the current pixel point value is less than and equal to K, set the pixel point value as 0; traverse all pixel points to obtain the binarized image.

[0038] Step 4, perform secondary filtering on the target set after primary filtering, and calculate the distance between two targets. When the distance between two targets is less than the distance threshold D (set D=image resolution x 1%) and the brightness values are the same, merge the targets and execute step 5; otherwise, judge whether there is a target between the two targets, yes, execute step 5, otherwise, update the current frame image and return to step 1 for continuous execution.

[0039] Step 5, mark the target on the pre-processed image, and execute step 6.

[0040] Step 6, detecting whether a take-off event occurs, yes, ending the flow, otherwise updating the current frame image, returning to step 1 for continuous execution. The way to judge whether a take-off event occurs is: binarizing the labeled image, solving its connected domain, presetting a connected domain size threshold S (S = image resolution x 8%) and a brightness concentration threshold. If the connected domain area is greater than S or the average brightness of the pixel points in the connected domain is greater than the brightness concentration threshold, it is judged that a take-off event occurs at this time, the target aircraft image that occurs the take-off event is saved and the current time is recorded as the take-off time; otherwise, the current frame image is updated, and the step 1 is returned for continuous execution.

[0041] The following will be described in detail with the aircraft take-off event detection verification as an example. In this embodiment, the target element is a certain type of target aircraft. Each frame image of the video source is input into the detection model, and the black edges and character-containing areas that may affect the detection accuracy are removed through preprocessing. Since the image background is constantly updated due to the continuous movement of the camera, only one algorithm cannot be used to detect the target aircraft target. The histogram type of the current image frame is judged to select the target aircraft target for real-time detection, the background modeling or the connected domain obtained by binarizing the image is used to obtain the target set, the detected target set is filtered to judge whether the main target meets the requirements, and the video frame is labeled; the target position is constantly updated and the take-off event detection is started, when the average brightness value of the target region pixel points changes more than a certain threshold, it is judged whether it is a take-off event of the target aircraft; the target aircraft image that occurs the take-off event is saved and the current time is recorded. When playing back, the time when the take-off event occurs is directly located to verify the authenticity and objectivity of the detection result.

[0042] In summary, the above is only a preferred embodiment of the present application, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for detecting a takeoff time of a target aircraft, characterized by, The specific steps include: Step 1, pre-processing a target aircraft image of a current frame to obtain a pre-processed image; Step 2, calculating a histogram of the pre-processed image; Step 3, judging a type of the histogram according to a characteristic value t of the histogram, selecting a target detection method according to the type of the histogram, and obtaining a target set; Step 4, filtering the target set obtained in step 3, and calculating distances between targets; when distances between two targets are less than a distance threshold D and the two targets have the same brightness value, merging the two targets and executing step 5; judging whether the merged target set contains a target aircraft, and if so, executing step 5, otherwise, updating the current frame image and returning to step 1 for continuous execution; Step 5, marking the targets on the pre-processed image and executing step 6; Step 6, detecting whether a take-off event occurs, and if so, ending the flow, otherwise, updating the current frame image and returning to step 1 for continuous execution; The type of the histogram is one of seven types, which are: type 1 is a single-peak and clean histogram, and the characteristic value t is greater than 0.9; type 2 is a single-peak and noisy histogram, and 0.8 < t < 0.9; type 3 is a noisy histogram, and 0.2 < t < 0.4; type 4 is a double-peak and clean histogram, and 0.8 < t < 0.9; type 5 is a double-peak and noisy histogram, and 0.9 < t < 0.95; type 6 is a histogram without a peak value, and 0.01 < t < 0.05; and type 7 is a histogram with a peak value, and 0.005 < t < 0.01; The specific manner of step 3 is: when the histogram belongs to one of types 1 to 6, step 3.1 is executed; when the histogram belongs to type 7, step 3.2 is executed; Step 3.1, performing background modeling on the pre-processed image and subtracting the background to obtain a foreground image; removing noise points and interference targets in the foreground image, solving a connected domain of the image to obtain a target set; performing primary filtering on the target set, and judging whether the size, brightness and position of the target after the primary filtering meet preset values, and if so, executing step 4, otherwise, updating the current frame image and returning to step 1 for continuous execution; Step 3.2, binarizing the pre-processed image to obtain a binarized image; solving a connected domain of the binarized image to obtain a target set, and executing step 4.

2. The detection method of claim 1, wherein, The specific manner of step 1 is: inputting the target aircraft image into a detection model to remove black edges and characters in the target aircraft image.

3. The method of claim 1, wherein the detecting is performed by a method selected from the group consisting of mass spectrometry, nuclear magnetic resonance, and chromatography. In step 3.1, an adaptive Gaussian mixture background modeling algorithm is used to model the background and subtract the background from the pre-processed image to obtain a foreground image.

4. The detection method as described in claim 1, characterized in that, In step 3.1, an image morphological algorithm is used to remove noise points and interference targets in the foreground image.

Images