A method for dynamic parameter detection of horizontal stabilizers in aerial videos

Abstract

This invention relates to the field of aerial video image detection technology, and discloses a method for dynamic parameter detection of a horizontal stabilizer in aerial video, comprising the following steps: Step 1: Creating a horizontal stabilizer parameter matching template; Step 2: Acquiring a horizontal stabilizer display image from an aerial video, binarizing the horizontal stabilizer display image, and storing it; Step 3: Performing dynamic decimal point detection in the binarized horizontal stabilizer display image to identify the decimal point position; Step 4: Determining the position indicated by the arrow based on the decimal point position, thereby obtaining the horizontal stabilizer's movement unit; Step 5: Determining the range of the digit cropping area based on the decimal point position, and performing sliding matching on the digit cropping area based on the horizontal stabilizer parameter matching template to identify the range of horizontal stabilizer movement angles.

Description

Technical Field

[0001] This invention relates to the field of aerial video image detection technology, and in particular to a method for dynamic detection of parameters of the horizontal stabilizer in aerial videos. Background Technology

[0002] In aerospace design, the horizontal stabilizer is a crucial component of an aircraft, ensuring appropriate static stability in the pitch direction. The horizontal stabilizer is the fixed surface portion of the horizontal tail. When the aircraft is flying level, the horizontal stabilizer does not generate additional torque on the aircraft. However, when the aircraft is disturbed and pitches up, the aerodynamic forces acting on the horizontal stabilizer generate a torque that causes the aircraft to pitch down, restoring it to a level flight attitude. Similarly, if the aircraft pitches down, the torque generated by the horizontal stabilizer will cause it to pitch up until it returns to level flight. To improve the reliability of safe flight, the pilot's actual control results on the horizontal stabilizer need to be displayed accurately and in real-time on an interactive display unit (IDU) that allows direct interaction with the pilot. Based on this, the pilot can adjust the horizontal stabilizer parameters to achieve optimal flight conditions.

[0003] Because flight display screens may contain errors that could mislead pilots into making incorrect decisions and causing flight accidents, the horizontal stabilizer display parameters are monitored in real time to ensure flight safety and reliability, thereby preventing pilots from making incorrect decisions due to display errors.

[0004] The parameters of the horizontal stabilizer are displayed on the aircraft's IDU, including the angle range (HSTAB_position) and the unit of movement (HSTAB_position_value) for the upward and downward movement of the stabilizer's front edge. A schematic diagram is shown below. Figure 1 As shown, the number inside the arrow represents the HSTAB_position; the position the arrow points to, between NU and ND, corresponds to a value between -100 and +100, which is the HSTAB_position_value. As the pilot manipulates the horizontal stabilizer, the parameters change, with the arrow moving up and down corresponding to different positions and position_values.

[0005] Monitoring the horizontal stabilizer essentially involves dynamically identifying these two parameters in the image and comparing the results with information transmitted through other communication channels to ensure flight safety. This paper presents a dynamic detection method for the horizontal stabilizer in aerial video images, employing a Field-Programmable Gate Array (FPGA) to dynamically identify the values ​​of these two parameters. Summary of the Invention

[0006] Objective of the Invention: To address the aforementioned need for dynamic parameter detection of horizontal stabilizers, the objective of this invention is to provide a method for dynamic parameter detection of horizontal stabilizers in aerial video. This invention can detect the parameters of the horizontal stabilizer even when the image is dynamic, and the method is easily implemented using FPGA design.

[0007] The technical solution of this invention is:

[0008] A method for dynamic parameter detection of horizontal stabilizers in aerial videos includes the following steps:

[0009] Step 1: Create a template for matching the parameters of the horizontal stabilizer;

[0010] Step 2: Acquire the horizontal stabilizer display image from the aerial video, binarize the horizontal stabilizer display image, and then store it;

[0011] Step 3: Perform dynamic decimal point detection in the binarized horizontal stabilized surface display image to identify the decimal point position;

[0012] Step 4: Determine the position indicated by the arrow based on the decimal point position to obtain the unit of movement of the horizontal stabilizer;

[0013] Step 5: Determine the range of the digit interception area based on the decimal point position, and perform sliding matching on the range of the digit interception area based on the horizontal stabilizer parameter matching template to identify the range of horizontal stabilizer movement angle.

[0014] Furthermore, in step one, the horizontal stabilizer parameter matching template includes fourteen vertically arranged characters, namely the numbers 0-9, a minus sign, a left arrow, a blank space, and a left arrow plus a minus sign.

[0015] Furthermore, in step two, the binarization process of the horizontal stabilized surface display image includes: first converting the RGB image into a grayscale image, and then converting the grayscale image into a binary black and white image.

[0016] Furthermore, in step three, the dynamic detection of the decimal point includes the following process:

[0017] Determine the decimal point truncation region X1*X2, where X1 and X2 are the number of pixels in the row and column directions of the decimal point truncation region;

[0018] Based on the size of the decimal point, perform sliding matching from bottom to top within the decimal point truncation area;

[0019] Choose the row coordinate corresponding to the smallest total matching error value from all the total matching error values ​​as the row coordinate corresponding to the decimal point position.

[0020] Furthermore, in step four, the row coordinate of the position indicated by the arrow is equal to the decimal row coordinate + H;

[0021] H is the fixed difference between the position indicated by the arrow and the decimal point row coordinates.

[0022] Furthermore, in step four, the position_value of the horizontal stabilized surface is obtained as follows:

[0023] Define the position indicated by the arrow as pos_arrow, with its origin above NU and 12 pixels away from the NU line. Define pixel_num = |pos_arrow - 68|, with its origin being the origin of position_value.

[0024] When pos_arrow > 68, then position_value = pixel_num * 7, and the position_value error is 7 pixels.

[0025] When pos_arrow <= 68, then position_value = pixel_num * (-2) + offset, and the position_value error is 2 pixels.

[0026] The relationship between offset and pos_arrow is shown in the table below:

[0027] pos_arrow offset value (62,68] 0 (51,62] 2 (40,51] 4 (29,40] 6 (18,29] 8 (12,18] 10 .

[0028] Furthermore, in step three, no numbers are displayed within the decimal point truncation area; only the decimal point area and the black area are visible.

[0029] Furthermore, in step five, the range of digit truncation area is determined on both sides of the decimal point, and the digit truncation area contains only digits.

[0030] The present invention has the following beneficial effects:

[0031] This invention fully analyzes the display characteristics of horizontal stabilizer parameters in aerial video images and, utilizing the relative positional relationships of graphics and template matching, proposes a dynamic parameter detection method for horizontal stabilizers in aerial videos. This method can achieve high-confidence parameter identification with minimal FPGA resource usage and is easily implemented using FPGA programming. Attached Figure Description

[0032] Figure 1 This is a schematic diagram showing the parameters of the horizontal stabilizer.

[0033] Figure 2This is a flowchart of a method for dynamic parameter detection of horizontal stabilizers in aerial videos;

[0034] Figure 3 A diagram illustrating the decimal point truncation area;

[0035] Figure 4 A diagram illustrating the identification of the position_value parameter;

[0036] Figure 5 This is a template illustration;

[0037] Figure 6 This is a schematic diagram of the sliding matching method;

[0038] Figure 7 Display images of horizontal stabilizer parameters;

[0039] Figure 8 This image displays the horizontal stabilizer parameters after binarization. Detailed Implementation

[0040] The present invention will now be described in detail with reference to specific implementation processes. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0041] A method for dynamic parameter detection of horizontal stabilizers in aerial videos includes three steps: image binarization, dynamic decimal point detection, and parameter recognition. Figure 2 As shown.

[0042] Image binarization: This involves converting an RGB color image into a binary image for subsequent dynamic detection. The method involves inputting image information into an FPGA, converting the RGB values ​​of the image pixels to grayscale according to certain criteria, and then converting the grayscale images back to binary images. Image pixel color information is divided into three dimensions (R, G, B), each described by 8 bits of binary data. Therefore, the RGB value range is 0-255. The RGB values ​​of binarized image pixels are only (0,0,0) or (255,255,255), i.e., pure black or pure white. This allows for the description of image contours and other information with a very small amount of data. Image binarization reduces both FPGA resource overhead and algorithm complexity.

[0043] Decimal point dynamic detection: Because the decimal point occupies a small pixel area, detecting it during dynamic image movement is relatively simple, requires fewer FPGA hardware resources, and is easier to implement in terms of timing. After detecting the decimal point, the position indicated by the arrow can be determined by the relative positional relationship between graphics, thus obtaining the position_value. Simultaneously, the range of the digit truncation area can be determined, and after matching it with a template, the position value can be identified. A schematic diagram of the decimal point truncation area is shown below. Figure 3 As shown:

[0044] The size of the truncated region is 11*96 ([599:609,679:774]), and the size of the decimal point region is 3*4 ([603:605,a:a+3]), where a∈[679,771]. The truncated region information is stored in the FPGA's RAM. This truncated region does not display any numbers, only the decimal point region and the black region. At the top of the truncated region (row_cnt∈[679,682]), the matching error is calculated for the 11*4 region. The calculation method is as follows: if the col_cnt (column coordinate) of the region is in [603,605], it is a black block, and the error value is increased by 1; if the col_cnt is outside [603,605], it is a white block, and the error value is unchanged; if the col_cnt is outside [603,605], it is a white block, and the error value is increased by 1; if the col_cnt is outside [603,605], it is a white block, and the error value is unchanged. The total matching error value is obtained. Then move the 11*4 region down by 1 pixel and obtain the total matching error value at this time using the method described above. A total of 93 total matching error values ​​are obtained. The row_cnt (row coordinate) range corresponding to the smallest total matching error value in the 11*4 region is the row coordinate range corresponding to the position of the decimal point.

[0045] In actual display, due to limitations in graphics card rendering, there may be white gaps in the decimal point area. The impact of these white gaps on the algorithm implementation is analyzed below. According to the dynamic detection algorithm, the total matching error is calculated for each of the four lines, and the total matching error value ε is:

[0046] ε=ε0+ε1+ε2+ε3=12-ε w0 -ε w1 -ε w2 -ε w3

[0047] In the formula ε w0 ε represents the number of white pixels in row 0. w1 a is the number of white pixels in the first row. w2 ε represents the number of white pixels in the second row. w3Let x0 be the number of white pixels in the third row. From rows 90, 91, 92, and 93 to rows 93, 94, 95, and 96, four scanning areas contain white pixels, and the total error value is less than 12. Let these four scanning areas be x0, x1, x2, and x3. Then the matching error value is:

[0048] ε x0 =12-ε 93

[0049] ε x1 =12-ε 93 -ε 94

[0050] ε x2 =12-ε 93 -ε 94 -ε 95

[0051] ε x3 =12-ε 93 -ε 94 -ε 95 -ε 96

[0052] In the formula, ε 93 ε 94 ε 95 ε 96 These represent the number of white pixels in rows 93, 94, 95, and 96, respectively. The algorithm aims to minimize the matching error ε. x3 It is distinct from the other matching error values; otherwise, multiple decimal point regions would be identified based on the minimum matching error. Therefore, only ε... 96 When ε is 0, x2 With ε x3 Since they are equal, it is necessary to ensure that there are white pixels in row 96.

[0053] In summary, this algorithm allows for the display of white pixels as long as the last row (row 96) of the decimal area is displayed even if the decimal area is not fully displayed during image binarization. However, no white pixels were found in the 96th row of the graphics card rendering result.

[0054] Parameter recognition: The position indicated by the arrow can be obtained based on the location of the decimal point. The difference between the row coordinate of the arrow's position and the row coordinate of the last row after the decimal point is 9 pixels. An illustration of obtaining the position_value based on the arrow's position is shown below. Figure 4 As shown:

[0055] Define the position indicated by the arrow as pos_arrow, with its origin above NU, 12 pixels away from the NU line. Define pixel_num = |pos_arrow - 68|, and its origin is the origin of position_value. When pos_arrow > 68, then position_value = pixel_num * 7, with a position_value error of 7 pixels; when pos_arrow <= 68, then position_value = pixel_num * (-2) + offset, with a position_value error of 2 pixels. The relationship between offset and pos_arrow is shown in Table 1.

[0056] Table 1. Correspondence between offset and pos_arrow

[0057] pos_arrow offset value (62,68] 0 (51,62] 2 (40,51] 4 (29,40] 6 (18,29] 8 (12,18] 10

[0058] The position value needs to be obtained through a template matching method. The template is obtained by comprehensively training on characters from thousands of screenshots on the monitor and is stored in the FPGA's ROM. After decimal point matching, the correct character for position in the video image is obtained. A template diagram is shown below. Figure 5 As shown, it includes characters such as numbers 0-9, minus signs, arrows, arrows with minus signs, and empty areas:

[0059] The template size is 16*295, the cropping area size is 12*18, and the sliding matching method is used for parameter recognition, as shown in the diagram. Figure 6 As shown:

[0060] The region to be matched is extracted from the overall image with a resolution of 1680*1050. This region should ideally contain only characters from the template to reduce RAM storage space usage. Data in RAM is compared with data in the template ROM based on row and column pointers, simulating the sliding of the extracted region on the template. The number of pixels that differ between the extracted region and its corresponding template region during each slide is recorded as the matching error. The sliding process continues until the very end of the template, yielding a set of matching errors. The character corresponding to the template region with the smallest matching error is the matching result. The matching result includes numbers, negative signs, and empty regions.

[0061] like Figure 7 This is the image showing the horizontal stabilizer parameters in this example (cropped from a 1680*1050 image). At this time, HSTAB_position_value is -55 and HSTAB_position is 1.5.

[0062] Step 1 (Image Binarization): Convert the image to grayscale according to a certain color range, then binarize it. The resulting image is shown below. Figure 8 As shown.

[0063] Step 2 (Dynamic Decimal Point Detection): Dynamic decimal point detection is performed using the binarized image. The decimal point region in this image is not 3*4, resulting in pixel loss, but this does not affect the dynamic detection. Error values ​​are calculated, and the relationship between the error values ​​at each position is analyzed. The position_value is obtained by matching the position with the smallest error value. At this point, the row_cnt range for the identified decimal point is [722:725].

[0064] Step 3 (Parameter Recognition): Based on the recognized decimal point range, pos_arrow = 37, so position_value = -56. The recognition result is within the error limit. Then, the truncation range of position is determined. According to the template matching method, the value of position is 1.5.

[0065] A dynamic detection method for horizontal stabilizer parameters in aerial video images has been developed by fully analyzing the display characteristics of horizontal stabilizer parameters in aerial video images and utilizing the relative positional relationship of graphics and template matching. This method can achieve high confidence and high real-time parameter identification with minimal FPGA resource usage, and it is easy to program and implement on FPGA and other devices.

[0066] The embodiments disclosed herein are merely preferred embodiments of the present invention. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, and are not intended to limit the present invention. Any modifications and variations made by those skilled in the art within the scope of this specification should fall within the protection scope of the present invention.

Claims

1. A method for dynamic parameter detection of a horizontal stabilizer in aerial video, characterized in that: The method includes the following steps: Step 1: Create a horizontal stabilizer parameter matching template, which includes fourteen vertically arranged characters: numbers 0-9, a minus sign, a left arrow, a space, and a left arrow with a minus sign. Step 2: Acquire the horizontal stabilizer display image from the aerial video, binarize the horizontal stabilizer display image, and then store it; Step 3: Perform dynamic decimal point detection in the binarized horizontal stabilized surface display image to identify the decimal point position. Dynamic decimal point detection includes the following steps: determine the decimal point truncation region X1*X2, where X1 and X2 represent the number of pixels in the row and column directions of the truncation region; perform sliding matching from bottom to top within the truncation region based on the decimal point size; select the row coordinates corresponding to the smallest total matching error value from all total matching error values ​​as the row coordinates corresponding to the decimal point position. Step 4: Determine the position indicated by the arrow based on the decimal point position, thus obtaining the unit of movement of the horizontal stabilizer; the row coordinate of the position indicated by the arrow is equal to the row coordinate of the decimal point + H; H is a fixed difference between the position indicated by the arrow and the row coordinate of the decimal point. Step 5: Determine the range of the digit interception area based on the decimal point position, and perform sliding matching on the range of the digit interception area based on the horizontal stabilizer parameter matching template to identify the range of horizontal stabilizer movement angle.

2. The method according to claim 1, characterized in that: In step two, the binarization process of the horizontal stabilized surface display image includes: first, converting the RGB image into a grayscale image, and then converting the grayscale image into a binary black and white image.

3. The method according to claim 2, characterized in that: In step four, the position_value of the horizontal stabilizer is obtained as follows: Define the position indicated by the arrow as pos_arrow, with its origin above NU and 12 pixels away from the NU line. Define pixel_num = |pos_arrow-68|, with its origin being the origin of position_value. When pos_arrow > 68, then position_value = pixel_num * 7, and the position_value error is 7 pixels. When pos_arrow <= 68, then position_value = pixel_num * (-2) + offset, and the position_value error is 2 pixels. The relationship between offset and pos_arrow is shown in the table below: 。 4. The method according to claim 3, characterized in that: In step three, no numbers are displayed within the decimal point truncation area; only the decimal point area and the black area are visible.

5. The method according to claim 4, characterized in that: In step five, the range of digit truncation area is determined on both sides of the decimal point, and the digit truncation area contains only digits.

Images