Large-field-of-view swing-scanning type drop point miss distance extraction method based on background splicing

Abstract

The invention relates to the field of image processing, and discloses a large-view-field sweep type drop point miss distance extraction method based on background stitching. According to the implementation scheme, the method comprises the following steps: 1, before a task, determining and searching a large-view-field area, and calculating a single-view-field view field angle; 2, slowly and horizontally scanning from left to right by the photoelectric theodolite, and splicing a background; 3, after the task is started, staring a preset drop point view field by the photoelectric theodolite, detecting a smoke column in a neighborhood on a horizon line, and extracting a drop point; and 4, outputting the drop point miss distance during staring or sweeping. According to the method, a background splicing-based large-field-of-view frame difference method and a strategy of staring detection first and then sweep detection are adopted, and the drop point miss distance is extracted by using the smoke column position. The method improves the detection capability of the target drop point of the visible light target in the target range.

Description

Technical Field

[0001] This invention relates to the field of image processing, and specifically to a method for extracting the miss distance of a large field of view sweeping impact point based on background stitching. Background Technology

[0002] With the continuous development of digital image processing technology, the target detection technology of visible light photoelectric theodolites is also constantly evolving. Photoelectric theodolites utilize optical images to detect and track targets. Currently, the commonly used detection method is moving target detection based on frame difference. For example, another Chinese patent application of mine, 2024114851563, discloses "A method for extracting the miss distance of a projectile impact point in a staring visible light photoelectric theodolite," application publication number CN119559251A. Its implementation scheme is as follows: 1. The photoelectric theodolite acquires visible light images in real time and buffers background frames; 2. The real-time acquired current frame image and the background frame are differentially analyzed to obtain a difference image; 3. The difference image target is extracted based on multi-feature joint filtering; 4. The target centroid and lowest point coordinates extracted in the current frame are calculated; 5. It is determined whether the target centroid ordinate in the current frame is at its minimum value, and the miss distance of the target's lowest point when the target centroid ordinate is at its minimum value is output. However, this method can detect moving targets within a fixed field of view that does not exceed the visibility range of the photoelectric theodolite. For larger fields of view that exceed the visibility range of the photoelectric theodolite, dynamic scene detection is required. Another Chinese patent application, 202510414539.X, discloses a "Large Field of View Situation Stitching System and Method Based on Multi-device Joint Imaging," with publication number CN119941505A. This system includes an image acquisition module that acquires image data from camera subsystems of multiple measurement stations in real time; a communication module that communicates with multiple measurement stations in real time and acquires data information from multiple measurement stations in real time; an image processing module that constructs a situation image in real time based on the image data and data information from multiple measurement stations; an image display module that displays the image data and situation image from each measurement station in real time; and an image output module that converts the situation image into SD SDI or HD SDI format via an image output card and outputs the converted situation image in segments to the display system of the command and control center, realizing real-time monitoring of each target. The system requires the setup of multiple measurement stations and real-time communication, resulting in a complex structure. The data processing module needs to collect and process image data from multiple measurement stations, leading to a large data processing volume and slow operation speed. Summary of the Invention

[0003] To address the challenges of projectile deviation from the field of view, searching for the projectile's impact point, and extracting the miss distance, this invention proposes a large field-of-view sweeping method for extracting the miss distance based on background stitching. This method can utilize smoke plumes from both the staring field of view and the large field-of-view search area to detect the impact point, thereby improving the detection capability of visible light targets in the firing range.

[0004] The present invention discloses a method for extracting the miss distance of a large field of view sweeping impact point based on background stitching, comprising the following steps: 1) Capture high-speed sequence images, determine the search area with a large field of view, divide the large field of view area into multiple horizontally arranged single field of view areas based on the horizontal distance of the field of view, and each single field of view area has an equal horizontal distance of the single field of view area. Calculate the single field of view angle based on the horizontal distance of the single field of view and the shooting distance. 2) The photoelectric theodolite scans the background image of each single field of view area, and stitches the background images of the single field of view areas together to obtain a large field of view background image; 3) At the trigger moment, the photoelectric theodolite gazes at the preset landing point field of view and detects smoke columns in the narrow band region of the horizon. The narrow band region of the horizon is a horizontal strip-shaped area within a certain width above the horizon of the preset landing point field of view. If a smoke column is detected in the narrow band region of the horizon at the trigger moment, the abscissa of the centroid of the smoke column is marked as the abscissa of the projectile's landing point. If no smoke column is detected in the narrow band region of the horizon at the trigger moment, the oscillating smoke column detection algorithm is activated. The oscillating method is used to find the smoke column until it is detected. The landing point of the smoke column is extracted, and the abscissa of the centroid of the smoke column is marked as the abscissa of the projectile's landing point. The above method for detecting smoke columns is to perform background frame difference detection on the background area corresponding to the azimuth angle of the preset landing point field of view, set a grayscale segmentation threshold T, erode the background difference image multiple times, extract the largest connected region, and calculate the abscissa of the centroid of the connected region. 4) Output the miss distance of the projectile during staring or sweeping based on the shooting distance and the horizontal coordinate of the projectile's impact point.

[0005] In step 1), the single field of view angle is calculated using the following formula. Theta = 2·arctan(D / 2L) In the above formula, Theta is the field of view angle of a single field of view; D represents the horizontal distance of a single field of view; L represents the shooting distance.

[0006] In step 2), the photoelectric theodolite starts from the zero point of the field of view and rotates from left to right, with the azimuth angle gradually increasing. Each time it rotates through a field of view angle, it caches a single field of view background image until the entire large field of view area is scanned to obtain multiple single field of view background images. The multiple single field of view background images are then stitched together to obtain the large field of view background image.

[0007] In step 3), the sweeping search for the smoke column involves using the photoelectric theodolite with the current field of view as the center, searching first to the right and then to the left.

[0008] The advantages of this invention compared to the prior art are: The system searches for and detects the impact points of projectiles that deviate from the field of view, thereby assessing the accuracy of such impact points.

[0009] Background stitching technology was used to achieve target detection capability with a large field of view and a large area.

[0010] By utilizing the background frame difference detection in the narrow strip region of the horizon, the detection range is narrowed, focusing on the root of the smoke column and improving the accuracy of target detection. Attached Figure Description

[0011] Figure 1 This is a flowchart of a large field-of-view sweeping method for extracting the miss distance of a target based on background stitching, according to the present invention.

[0012] Figure 2 This is a schematic diagram for calculating a single field of view. Detailed Implementation

[0013] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0014] This invention provides a large field-of-view sweeping method for extracting the miss distance of projectiles based on background stitching. This method enables users to detect and extract the miss distance of deviated projectiles. The process is as follows: Figure 1 The steps shown are as follows: 1) Capture high-speed image sequences to determine the large field of view region to search for. Divide this large field of view region into multiple horizontally arranged single field of view regions based on the horizontal distance of the field of view. For example, the horizontal distance of the large field of view can be divided into 3-6 equal parts. Figure 2 As shown, each equal division represents a single field of view region; each single field of view region has an equal horizontal distance, and the single field of view angle is calculated based on the horizontal distance and the shooting distance; wherein, the single field of view angle is calculated using the following formula. Theta = 2·arctan(D / 2L) In the above formula, Theta is the field of view angle of a single field of view; D represents the horizontal distance of a single field of view; L represents the shooting distance.

[0015] The photoelectric theodolite scans the background image of each single field of view region, and then stitches the background images of the single field of view regions together to obtain a large field of view background image. When scanning the background image of each single field of view region, the photoelectric theodolite starts from the zero point of the field of view and rotates from left to right, with the azimuth angle gradually increasing; after rotating through each field of view angle, one single field of view background image is cached, until the entire large field of view region is scanned and multiple single field of view background images are obtained. Then, the multiple single field of view background images are stitched together to obtain the large field of view background image.

[0016] At the trigger moment, the photoelectric theodolite gazes at the preset landing point field of view and detects a smoke column in a narrow band area on the horizon. The narrow band area on the horizon is a horizontal strip-shaped area within a certain width above the horizon of the preset landing point field of view. If a smoke column is detected in the narrow band area on the horizon at the trigger moment, the abscissa of the centroid of the smoke column is marked as the abscissa of the projectile's landing point. If no smoke column is detected in the narrow band area on the horizon at the trigger moment, the oscillating smoke column detection algorithm is activated. The oscillating method is used to search for the smoke column until it is detected. The landing point of the smoke column is extracted, and the abscissa of the centroid of the smoke column is marked as the abscissa of the projectile's landing point. The oscillating search for the smoke column involves the photoelectric theodolite searching in the direction of first searching to the right and then searching to the left, with the current field of view as the center.

[0017] The above method for detecting smoke columns involves performing background frame difference detection on the background region corresponding to the azimuth angle of the preset landing point. A grayscale segmentation threshold T is set, and the background difference image is eroded multiple times to extract the largest connected region and calculate the centroid abscissa of the connected region.

[0018] Let I(x,y) be the grayscale image of the current frame, and B(x,y) be the grayscale image of the corresponding azimuth background image. Then the background difference image is: D(x,y)=|I(x,y)−B(x,y)|. Using a grayscale segmentation threshold T, the image is segmented to obtain a binary image expressed by the following formula: The background difference image is eroded multiple times to remove small areas of noise. Let the structuring element be S (such as a 3×3 rectangular structure), and the erosion operation be denoted as ⊖.

[0019] One erosion cycle can be represented as E1(x,y)=B bin (x,y)⊖S.

[0020] E represents the nth corrosion. n (x,y)=(((B bin ⊖S)⊖S)⋯⊖S)(total n times).

[0021] Finally, extract the largest connected component, and let the set of connected components be: {C1,C2,…,Ck}. Where each C... i Given a connected region (set of pixels), calculate the area (number of pixels) of each region: Take the area with the largest area: Calculate the x-coordinate of its centroid. Let C be the largest connected region. max The set of pixel coordinates in the data is: Given {(x1,y1),(x2,y2),…,(xN,yN)}, the x-coordinate of the centroid is: 4) Output the miss distance of the projectile during staring or sweeping based on the shooting distance and the horizontal coordinate of the projectile's impact point.

[0022] The invention will now be described in detail using a 1280*1024 resolution visible light photoelectric theodolite.

[0023] Step 1: Before the task, determine the large field of view area to search and calculate the single field of view angle Theta; Step 2: The photoelectric theodolite starts from the far left of the field of view and slowly scans horizontally from left to right. For each rotation of the azimuth angle by the Theta angle, a background image is cached until the entire field of view is covered by stitching from left to right.

[0024] Step 3: After the mission begins, at the trigger moment, the photoelectric theodolite gazes at the preset landing point field of view. The narrow horizon region is a horizontal stripe area of ​​100 pixels above the horizon. If a smoke column is detected in the narrow horizon region at the trigger moment, the background frame difference method is applied to the background region corresponding to that azimuth angle. A grayscale segmentation threshold of 80 is set, and the background difference image is eroded five times to extract the largest connected region. Its centroid is calculated, and the abscissa of the smoke column centroid is marked as the abscissa of the projectile's landing point. If the photoelectric theodolite does not detect the target in the preset landing point field of view, it means the projectile's landing point is not within that field of view. The search direction is first to the right, then to the left, until a smoke column target is detected in the narrow horizon region of the current frame. Subsequently, using the frame difference between the current frame and the background image corresponding to the current azimuth angle, the largest connected region is extracted, its centroid is calculated, and the abscissa of the smoke column centroid is marked as the abscissa of the projectile's landing point.

[0025] Step 4: Output the miss distance of the projectile during staring or sweeping based on the shooting distance and the x-coordinate of the projectile's impact point.

Claims

1. A method for extracting the miss distance of a large field of view sweeping impact point based on background stitching, characterized in that... Includes the following steps: 1) Capture high-speed sequence images, determine the search area with a large field of view, divide the large field of view area into multiple horizontally arranged single field of view areas based on the horizontal distance of the field of view, and each single field of view area has an equal horizontal distance of the single field of view area. Calculate the single field of view angle based on the horizontal distance of the single field of view and the shooting distance. 2) The photoelectric theodolite scans the background image of each single field of view area, and stitches the background images of the single field of view areas together to obtain a large field of view background image; 3) At the triggering moment, the photoelectric theodolite gazes at the preset landing point field of view and detects smoke columns in the narrow band area of ​​the horizon. The narrow band area of ​​the horizon is a horizontal strip area within a certain width range above the horizon of the preset landing point field of view. If a smoke column is detected in the narrow band of the horizon at the trigger time, the x-coordinate of the smoke column's centroid is marked as the x-coordinate of the projectile's impact point. If no smoke column is detected in the narrow band of the horizon at the trigger time, the oscillating smoke column detection algorithm is activated, and the oscillating method is used to find the smoke column until it is detected. The impact point of the smoke column is then extracted, and the x-coordinate of the smoke column's centroid is marked as the x-coordinate of the projectile's impact point. The above method for detecting smoke columns involves performing background frame difference detection on the background region corresponding to the preset impact point's field of view azimuth angle, setting a grayscale segmentation threshold T, repeatedly eroding the background difference image, extracting the largest connected region, and calculating the x-coordinate of the centroid of the connected region. 4) Output the miss distance of the projectile during staring or sweeping based on the shooting distance and the horizontal coordinate of the projectile's impact point.

2. The method for extracting the miss distance of a large field of view sweeping impact point based on background stitching according to claim 1, characterized in that, In step 1), the single field of view angle is calculated using the following formula. Theta = 2·arctan(D / 2L) In the above formula, Theta is the field of view angle of a single field of view; D represents the horizontal distance of a single field of view; L represents the shooting distance.

3. The method for extracting the miss distance of a large field of view sweeping impact point based on background stitching according to claim 1 or 2, characterized in that, In step 2), the photoelectric theodolite starts from the zero point of the field of view and rotates from left to right, with the azimuth angle gradually increasing. Each time it rotates through a field of view angle, it caches a single field of view background image until the entire large field of view area is scanned and multiple single field of view background images are obtained. The multiple single field of view background images are then stitched together to obtain the large field of view background image.

4. The method for extracting the miss distance of a large field of view sweeping impact point based on background stitching according to claim 1 or 2, characterized in that, In step 3), the sweeping search for the smoke column involves using the photoelectric theodolite with the current field of view as the center, searching first to the right and then to the left.

Images