Method for estimating ship deflection angle based on local characteristic constraint of two-dimensional projection of infrared image

Abstract

The application discloses an infrared image two-dimensional projection ship deflection angle estimation method based on local characteristic constraint, which comprises the following steps: S1, extracting contour information of a ship in an infrared ship image to obtain a binary image of the ship; S2, performing multiple image rotations on the binary image of the ship according to a preset rotation degree to obtain a rotation image set; S3, performing horizontal projection on each rotation image in the rotation image set to obtain each row projection value in the rotation image, and performing local characteristic constraint based on the each row projection value to obtain an optimized projection image; S5, estimating a deflection angle of the ship according to a projection pixel value of the optimized projection image; and S6, performing estimation performance evaluation based on the estimated deflection angle of the ship and an actual measured deflection angle of the ship. The linear constraint on the two-dimensional projection of the infrared image can effectively detect the deflection angle of the ship.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ship deflection angle estimation, and particularly relates to an infrared image two-dimensional projection ship deflection angle estimation method based on local characteristic constraint. BACKGROUND

[0002] The deflection angle of a ship is the basis for ship detection and tracking, and it is crucial to effectively estimate the deflection angle of a ship.

[0003] Currently, remote sensing technology is commonly used to monitor ships sailing on the sea. Remote sensing images can generally be divided into two types: synthetic aperture radar (SAR) images and optical images. SAR images are not affected by weather conditions and can be observed at all times and in all weathers, but the images have a coherent spot, and image interpretation is difficult. Optical images are generated by an optical imaging system, have high image resolution, and can obtain rich gray scale and texture information under good imaging conditions, which facilitates target interpretation. Infrared images belong to a type of optical images, and compared with traditional optical images, infrared images can obtain temperature distribution information of an object surface, have high thermal sensitivity and penetration, and can obtain good imaging results at night or in poor lighting conditions. Effectively estimating the deflection angle of a ship in an infrared image can meet the needs of actual work such as sea rescue and combating illegal smuggling.

[0004] There are five commonly used methods for estimating the deflection angle of a target in an image: a Fourier transform-based method, a linear fitting-based method, a Hough transform-based method, a nearest neighbor connected domain-based method, and a projection-based method. The Fourier transform method analyzes the characteristic information of an image in the frequency domain by performing Fourier transform on the target image, thereby realizing detection of the deflection angle of the target. This method detects the deflection angle of an image from the frequency domain, is easily affected by noise, and has the characteristics of high computational complexity and low efficiency. The linear fitting method solves the problem of high computational complexity, determines the deflection angle of an image by fitting a straight line to the boundary pixels, but is easily affected by image noise in the fitting process. The Hough transform method has strong anti-interference ability, uses the transformation relationship between two coordinate spaces to map lines with the same shape in space to a point in the parameter space, but is time-consuming and requires a large storage space. The nearest neighbor connected domain method establishes a connected domain based on the spatial relationship between target edge points, calculates the straight line angle between all edge points and the starting point in the connected domain, and takes the average value to solve the deflection angle of the target, but the algorithm has a large amount of calculation. The projection method detects the deflection angle according to the maximum value of the sum of the four maximum projection values, but it excessively relies on the straight line feature of the target edge.

[0005] To address the aforementioned problems, this invention proposes a method for estimating the ship's deflection angle based on two-dimensional projection of infrared images using local characteristic constraints. Summary of the Invention

[0006] This invention provides a method for estimating the ship's deflection angle based on two-dimensional projection of infrared images using local characteristic constraints, in order to overcome the aforementioned technical problems.

[0007] To achieve the above objectives, the technical solution of the present invention is as follows:

[0008] A method for estimating the yaw angle of a ship based on two-dimensional projection of infrared images using local characteristic constraints includes the following steps:

[0009] S1: Extract the contour information of the ship from the infrared ship image to obtain the binary image of the ship;

[0010] The binary image is a two-dimensional matrix consisting of m rows and n columns of pixels;

[0011] S2: Rotate the binary image of the ship multiple times according to a preset rotation degree to obtain a rotated image set;

[0012] S3: Horizontally project each rotated image in the rotated image set to obtain the projection value of each row in the rotated image;

[0013] Based on the projection values ​​of each row, local characteristic constraints are applied to obtain an optimized projection image;

[0014] S4: Estimate the ship's yaw angle based on the projected pixel values ​​of the optimized projected image;

[0015] S5: Estimated performance evaluation is performed based on the estimated ship deflection angle and the actual measured ship deflection angle.

[0016] Furthermore, the binary image of the ship described in S2 is rotated multiple times according to a preset rotation degree, calculated using the following formula:

[0017]

[0018] In the formula: Represents any point in a binary image Located in the image OK; Represents any point in a binary image Located in the image List; Indicates the preset rotation degree counterclockwise; This indicates the row number of any point in the rotated image; This indicates the column number of any point in the rotated image;

[0019] Further, the optimized projection image is acquired in S3, specifically

[0020] S3.1: calculating the rotation of the preset rotation degree and acquiring the projection value of each row in the rotated image; the calculation formula of the row projection value is

[0021]

[0022] In the formula: represents the number of columns of the image; , represents the number of rows of the image; represents the pixel value of the rotated image at point ; represents the projection on the i-th row of the image;

[0023] S3.2: selecting the maximum value of the image row projection in each row projection value ; the expression of the maximum value of the image row projection is

[0024]

[0025] In the formula: represents the number of rows of the maximum value of the image row projection; represents the projection on the i-th row of the image;

[0026] S3.3: setting the first preset projection threshold and the second projection threshold according to the maximum value of the image row projection ; and performing local characteristic constraint on each row projection value based on the first preset projection threshold and the second projection threshold, and the first preset projection threshold is greater than the second projection threshold;

[0027] The local characteristic constraint includes acquiring the number of rows of the first upper limit projection and the number of rows of the first lower limit projection in the image according to the first preset projection threshold in the vertical direction, and the projection values of the number of rows of the first upper limit projection and the number of rows of the first lower limit projection are equal to the first preset projection threshold;

[0028] and setting the projection values of all rows between the number of rows of the first upper limit projection and the number of rows of the first lower limit projection as the first preset projection threshold;

[0029] acquiring the number of rows of the second upper limit projection and the number of rows of the second lower limit projection in the image according to the second preset projection threshold in the vertical direction, and the projection values of the number of rows of the second upper limit projection and the number of rows of the second lower limit projection are equal to the second preset projection threshold;

[0030] ​​​Set the projection value of the row number above the row number where the second upper bound projection is located to zero, and set the projection value of the row number below the row number where the second lower bound projection is located to zero.

[0031] The projection values ​​of the rows between the first upper bound projection row number and the second upper bound projection row number are retained, as well as the projection values ​​of the rows between the first lower bound projection row number and the second lower bound projection row number.

[0032] S3.4: Define a new image row projection based on the image row projection constrained by local characteristics. The new image line projection The expression is

[0033]

[0034] In the formula: Indicates the image at the 1st Projection onto the row; express The Middle The row number of the uppermost row whose projection value is closest to the second projection threshold; express The Middle The row number of the uppermost row whose projection value is closest to the first preset projection threshold; express The Middle The row number of the next lower row whose projection value is closest to the first preset projection threshold; express The Middle The row number of the next lower row whose projection value is closest to the second preset projection threshold.

[0035] Furthermore, in S4, the ship's deflection angle is estimated based on the projected pixel values ​​of the optimized projected image, specifically as follows:

[0036] S4.1: Determine the new image row projection from top to bottom. The row number where the projection value equals the second projection threshold is located. And select the row number Number of consecutive rows with row number ;

[0037] S4.2: Select the rows from bottom to top. , number of rows and number of rows The number of rows containing the same projection value across all image rows; and the selected row number and the total row number. , number of rows and number of rows the first row number average difference value ;

[0038] the first row number average difference value The calculation formula of the first row number average difference value is

[0039]

[0040] In the formula, the row number t represents any one of the row number , the row number or the row number ;

[0041] S4.3: Determine the new image row projection from top to bottom The row number where the projection value equal to the first projection threshold value is , and the row number arranged continuously with the row number and the row number ;

[0042] S4.4: Select the row number where the projection value of each image row projection is the same from bottom to top respectively, the row number , the row number and the row number ; and calculate the second row number average difference value between the selected row number and the row number

[0043] The calculation formula of the second row number average difference value is

[0044]

[0045] In the formula, the row number t represents any one of the row number , the row number and the row number ;

[0046] S4.5: Calculate the difference value between the first row number average difference value and the second row number average difference value , and select the rotation image with the minimum difference value between the first row number average difference value and the second row number average difference value ;

[0047] According to the rotation image with the minimum difference value between the first row number average difference value and the second row number average difference value , an estimated value of the deflection angle of the ship in the image is obtained ;​​​

[0048] and the estimated value of the deflection angle of the ship The calculation formula is

[0049] .

[0050] Further, the calculation formula for estimating the performance evaluation in S5 is

[0051]

[0052] In the formula: represents the absolute value error between the estimated deflection angle of the ship from the image and the actual measured deflection angle of the ship . represents the angle estimation accuracy under the condition that the absolute value error between the estimated deflection angle of the ship and the actual measured deflection angle of the ship is less than a certain angle error tolerance factor . represents the number of ship images that meet the conditions in the brackets . represents the m-th ship image . represents the angle error tolerance factor . represents the number of ship images in the entire set of rotated images

[0053] Beneficial effects: The present application is an infrared image two-dimensional projection ship deflection angle estimation method based on local feature constraint. By extracting the contour information of the ship in the infrared ship image, the binary image of the ship is obtained. The binary image of the ship is rotated multiple times according to the preset rotation degree, and a set of rotated images is obtained. The horizontal projection values of each row in the rotated image are obtained by performing horizontal projection on each rotated image in the set of rotated images. Based on the horizontal projection values, the local feature constraint is obtained, and the optimized projection image is obtained. The deflection angle of the ship is estimated according to the projection pixel value of the optimized projection image. The estimated deflection angle of the ship and the actual measured deflection angle of the ship are used to evaluate the estimation performance. By making local feature constraint on the two-dimensional projection of the infrared image, the deflection angle of the ship can be effectively detected. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0055] Figure 1A flow chart of a ship deflection angle estimation method based on local characteristic constraint two-dimensional projection of an infrared image according to the present application;

[0056] Figure 2 A method flow chart in the embodiment;

[0057] Figure 3 A schematic diagram of estimating a ship deflection angle according to a projected pixel value in the embodiment;

[0058] Figure 4 A comparison diagram of deflection angle estimation results of the present application and a conventional algorithm in the embodiment. DETAILED DESCRIPTION

[0059] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0060] The embodiment provides a ship deflection angle estimation method based on local characteristic constraint two-dimensional projection of an infrared image, as shown in the figure, comprising the following steps: Figures 1 to 2

[0061] S1: Extracting contour information of a ship in an infrared ship image, obtaining a binary image of the ship;

[0062] The binary image is a two-dimensional matrix composed of pixels, m rows and n columns;

[0063] In order to calculate the deflection angle by using the contour information of the ship, the existing Chan-Vese level set model is used here to extract the contour information of the ship from the original image. The output result of the Chan-Vese level set model for ship contour extraction is a binary image, the pixel value of the ship area in the image is 1, and the pixel value of the non-ship area is 0. In this way, the contour of the ship is obtained by detecting the edge;

[0064] An image can be regarded as a two-dimensional matrix composed of pixels, the number of rows of the image represents the height of the matrix or the number of pixels in the vertical direction. If the size of an image is , the number of rows of the image is , and the number of columns of the image is ;

[0065] S2: Rotating the binary image of the ship multiple times according to a preset rotation degree, obtaining a set of rotated images;

[0066] The binary image​ , there is , the image is rotated counterclockwise; if the preset rotation degree is , and the preset rotation degree Specifically, 1°, that is, the binary image is rotated by 180°, and one rotated image is obtained every 1° rotation; the then rotates Any point position can be obtained by the following formula:

[0067] Specifically, the binary image of the ship in S2 is rotated multiple times according to the preset rotation degree, and the calculation formula is

[0068]

[0069] In the formula: Indicates that any point in the binary image is located in the row of the image; Indicates that any point in the binary image is located in the column of the image; Indicates the preset rotation degree of counterclockwise rotation; Indicates the row number of any point in the rotated image; Indicates the column number of any point in the rotated image;

[0070] S3: Perform horizontal projection on each rotated image in the set of rotated images to obtain the projection value of each row in the rotated image;

[0071] And based on the row projection value, a local feature constraint is performed to obtain an optimized projection image;

[0072] S4: Estimate the deflection angle of the ship according to the projection pixel value of the optimized projection image;

[0073] S5: Perform estimation performance evaluation based on the estimated ship deflection angle and the actual measured deflection angle of the ship.

[0074] The infrared image two-dimensional projection ship deflection angle estimation method based on local feature constraint mainly includes two parts: one part is ship contour extraction, and the Chan-Vese model is used to extract the contour information of the ship in the infrared image; the other part is ship deflection angle estimation, which first performs horizontal projection on the rotated image, and then performs local feature constraint on the two-dimensional projection of the infrared image; then the projection local mean difference minimum method is used to estimate the deflection angle of the ship, which can effectively detect the deflection angle of the ship.

[0075] In specific embodiments, the optimized projection image is obtained in S3, specifically

[0076] S3.1: Calculate the rotation preset rotation degree Obtain the projection value of each row in the rotated image; the calculation formula of the row projection value is

[0077]

[0078] In the formula: represents the number of columns of the image; , represents the number of rows of the image; represents the pixel value of the rotated image at point ; represents the projection of the image on the th row;

[0079] S3.2: Select the maximum image row projection value in each row projection value ; the expression of the maximum image row projection value is

[0080]

[0081] In the formula: represents the number of rows of the maximum image row projection value; represents the projection of the image on the th row;

[0082] S3.3: In order to remove the influence of noise in the image, set a first preset projection threshold and a second projection threshold according to the maximum image row projection value ; and perform local characteristic constraint on each row projection value based on the first preset projection threshold and the second projection threshold; the first preset projection threshold is greater than the second projection threshold, the first preset projection threshold is , and the second projection threshold is ;

[0083] The local characteristic constraint includes obtaining the number of rows of the first upper limit projection and the number of rows of the first lower limit projection in the image according to the first preset projection threshold in the vertical direction, and the projection values of the number of rows of the first upper limit projection and the number of rows of the first lower limit projection are equal to the first preset projection threshold;

[0084] and setting the projection values of all rows between the number of rows of the first upper limit projection and the number of rows of the first lower limit projection to the first preset projection threshold;

[0085] According to the second preset projection threshold, the number of rows where the second upper boundary projection is located and the number of rows where the second lower boundary projection is located are obtained in the vertical direction, and the projection values ​​of the number of rows where the second upper boundary projection is located and the number of rows where the second lower boundary projection is located are equal to the second preset projection threshold.

[0086] Set the projection value of the row number above the row number where the second upper bound projection is located to zero, and set the projection value of the row number below the row number where the second lower bound projection is located to zero.

[0087] The projection values ​​of the rows between the first upper bound projection row number and the second upper bound projection row number are retained, as well as the projection values ​​of the rows between the first lower bound projection row number and the second lower bound projection row number.

[0088] Specifically, due to the shape characteristics of ships, when the entire image contains only ships, the projection value of the middle row of the image's row projection is large, while the projection values ​​of the rows on both sides of the image are small; when the projection value of a certain row of the image is less than the second projection threshold value... When a row is defined as a noisy part of the image, its row number is determined. If the projection value of a row in the image exceeds a first preset projection threshold... When the value is greater than this, the rows that are considered to be the projections of the rows containing the ship's deck structure are identified.

[0089] S3.4: Based on the image row projection constrained by local characteristics, define the new row projection as follows: , ;

[0090] Define a new image row projection The new image line projection The expression is

[0091]

[0092] In the formula: Indicates the image at the 1st Projection onto the row;

[0093]

[0094] express The Middle The row number of the uppermost row whose projection value is closest to the second projection threshold; express The Middle The row number of the uppermost row whose projection value is closest to the first preset projection threshold; express The Middle The row number of the next lower row whose projection value is closest to the first preset projection threshold; express The Middle The row number where the projection value of the next lower row is closest to the second preset projection threshold;

[0095] In a specific embodiment, to ensure the accuracy and universality of the estimation, local characteristic constraints are applied to the ship's profile, and different local features of the ship are selected for estimating the ship's yaw angle. In S4, the ship's yaw angle is estimated based on the projected pixel values ​​of the optimized projected image. Specifically...

[0096] S4.1: Determine the new image row projection from top to bottom. The row number where the projection value equals the second projection threshold is located. And select the row number Number of consecutive rows with row number ;

[0097] S4.2: Select the rows from bottom to top. , number of rows and number of rows The number of rows containing the same or closest projection values ​​in each image row;

[0098] The selection rule is to select the number of rows respectively. , number of rows The row number containing the projection value with the smallest absolute value of the numerical difference between the projection values ​​of each row in the image;

[0099] And calculate the number of selected rows and the total number of rows. , number of rows and number of rows First row mean difference ;

[0100] The first row mean difference The calculation formula is:

[0101]

[0102] In the formula: t Indicates the number of rows , number of rows or number of lines Any row number in the array;

[0103] like Figure 3 As shown, the horizontal axis represents the row number of the image, and the vertical axis represents the projection value of a certain row in the image. The horizontal coordinates of the three leftmost points (numbers 1, 2, and 3 from the left) in the image are respectively located at the 1st, 2nd, and 3rd positions of the rotated image. The row; the vertical axis is the first row. the horizontal direction away from the coordinate origin, the row number of the projection value of the image row of the point (4, 5, 6) can be obtained the horizontal direction away from the coordinate origin, the row number of the projection value of the image row of the point (4, 5, 6) can be obtained The three points can be found, which are respectively the same as or closest to the image row projection values of the three leftmost points (1, 2, 3), that is, the three rightmost points (the 10th, 11th and 12th blue points from the left) in the figure. The horizontal coordinates of the three points are the row numbers of the projection values of the image rows of the three points respectively selected from the bottom to the top which are respectively the same as or closest to the image row projection values of the row numbers , row number and row number ;

[0104] S4.3: Determine the new image row projection from top to bottom The row number of the projection value of the image row of the point (4, 5, 6) is equal to the first projection threshold value , and the row number of the projection value of the image row of the point (4, 5, 6) is equal to the first projection threshold value ; ;

[0105] S4.4: Select the row number of the projection value of the image row of the point (4, 5, 6) from the bottom to the top which is respectively the same as or closest to the image row projection value of the row number , row number and row number ; the selection rule is to select the row number of the projection value of the image row of the point (4, 5, 6) which is respectively the same as or closest to the image row projection value of the row number , row number and row number ;

[0106] Calculate the second row number average difference , row number and row number between the selected row number and the row number ;

[0107] The calculation formula of the second row number average difference is

[0108]

[0109] As shown in FIG. 6, Figure 3 is the horizontal coordinate of the 6th point in the figure, is the row number of the projection value of the image row of the point (4, 5, 6), and is the row number of the projection value of the image row of the point (4, 5, 6), and is the row number of the projection value of the image row of the point (4, 5, 6), and the horizontal direction away from the coordinate origin, that is ​The row can find three points, which are respectively the three points of (4, 5, 6) and have the same or the closest image row projection value as the three points, that is, the (7th, 8th and 9th points) marked in the figure; the horizontal coordinates of the three points are respectively the row numbers selected from bottom to top , the row numbers and the row numbers of the projection values of the image row projection values;

[0110] S4.5: Calculate the difference between the first row mean difference value and the second row mean difference value in each rotated image, and select the rotated image with the minimum difference between the first row mean difference value and the second row mean difference value ;

[0111] According to the rotated image with the minimum difference between the first row mean difference value and the second row mean difference value , obtain the estimated value of the deflection angle of the ship in the image ; the difference between the first row mean difference value and the second row mean difference value of the obtained rotated image at different rotation angles is also not the same; considering the projection characteristics of the ship, when the ship is located at the horizontal position, the obtained difference is the smallest, and at this time, the rotation angle of the image is the deflection angle of the ship in the image;

[0112] and the deflection angle of the ship is calculated by the formula

[0113] .

[0114] In specific embodiments, in order to quantitatively evaluate the estimation performance of the deflection angle of the ship, the angle estimation accuracy is used to evaluate the estimation performance, and the calculation formula for evaluating the estimation performance in S5 is

[0115]

[0116] In the formula: represents the absolute value error between the estimated deflection angle of the ship and the actual measured deflection angle of the ship ; represents the angle estimation accuracy under a certain angle error tolerance factor, which satisfies that the absolute value error between the estimated deflection angle of the ship and the actual measured deflection angle of the ship is less than the angle error tolerance factor; represents the number of ship images that satisfy the condition in the brackets; represents the a plurality of ship images; an angle error tolerance factor is represented; representing the number of ship images in the entire set of rotated images, and the The greater the value, the higher the accuracy of the estimated ship deflection angle.

[0117] Specifically, as Figure 4 shown, the experiment uses 52 infrared sea ship images for testing, and four groups of data are selected for comparison and display. Figure 4 (a) is an infrared image of a ship, Figure 4 (b), (c) and (d) are the estimated results of the ship deflection angle by the algorithm of the patent, the randon transformation algorithm and the minimum circumscribed rectangle algorithm respectively. After the four groups of data are processed by the patent algorithm, the ship in the image is basically located in the horizontal position, indicating that the patent algorithm can effectively estimate the tilt angle of the ship. The randon transformation algorithm detects the deflection angle by extracting the projection line information of the target object at different angles. The minimum circumscribed rectangle algorithm estimates the deflection angle of the ship by finding the minimum circumscribed rectangle of the ship at different rotation angles. For complex-shaped ships, the accuracy of the angle estimation results of the two comparison algorithms cannot be guaranteed, as shown in Table 1;

[0118]

[0119] In order to intuitively compare the performance of the three methods, the angle evaluation standard is used to quantitatively analyze them, as shown in Table 1. The patent analyzes the angle evaluation standard under three different angle error tolerance factors, and the results show that the angle estimation accuracy of the patent algorithm is significantly higher than that of the other two algorithms. In the case of a larger angle error tolerance factor, the angle evaluation standards of the patent algorithm and the other two algorithms have little difference, but as the angle error tolerance factor decreases, the angle estimation accuracy of the patent algorithm is significantly higher than that of the other two comparison algorithms, which is because the two comparison algorithms do not fully consider the shape information of the ship.

[0120] The present application proposes an infrared image two-dimensional projection ship deflection angle estimation method based on local characteristic constraint. In order to effectively estimate the deflection angle of the ship, a ship deflection angle estimation method with the minimum local mean difference is given according to the horizontal projection characteristics of the ship, and the angle estimation results of the present application, the randon transformation algorithm and the minimum circumscribed rectangle algorithm are compared through experiments. The experiment shows that compared with the randon transformation algorithm and the minimum circumscribed rectangle algorithm, the patent algorithm has at least a 10% advantage.

[0121] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for estimating the yaw angle of a ship based on two-dimensional projection of infrared images using local characteristic constraints, characterized in that, Includes the following steps: S1: Extract the contour information of the ship from the infrared ship image to obtain the binary image of the ship; The binary image is a two-dimensional matrix consisting of m rows and n columns of pixels; S2: Rotate the binary image of the ship multiple times according to a preset rotation degree to obtain a rotated image set; S3: Horizontally project each rotated image in the rotated image set to obtain the projection values ​​for each row of the rotated image; and based on the projection values ​​for each row, apply local characteristic constraints to obtain an optimized projected image, specifically... S3.1: Calculate the preset rotation degree Then, the projection values ​​of each row in the rotated image are obtained; the formula for calculating the row projection values ​​is as follows: In the formula: Indicates the number of columns in the image; , Indicates the number of rows in the image; Represents a rotated image at a point Pixel value at; Indicates the image at the 1st Projection onto the row; S3.2: Select the maximum value of the image row projection from each row projection value. The maximum value of the image row projection The expression is In the formula: This indicates the row number containing the maximum value of the image's row projection; Indicates the image at the 1st Projection onto the row; S3.3: Based on the maximum value of the image row projection Set a first preset projection threshold and a second preset projection threshold; and impose local characteristic constraints on the projection values ​​of each row based on the first preset projection threshold and the second preset projection threshold, wherein the first preset projection threshold is greater than the second preset projection threshold; The local feature constraint includes obtaining the number of rows where the first upper bound projection is located and the number of rows where the first lower bound projection is located in the image in the vertical direction according to the first preset projection threshold, and the projection value of the number of rows where the first upper bound projection is located and the number of rows where the first lower bound projection is located is equal to the first preset projection threshold. And set the projection values ​​of all rows between the row number where the first upper bound projection is located and the row number where the first lower bound projection is located as the first preset projection threshold; According to the second preset projection threshold, the number of rows where the second upper boundary projection is located and the number of rows where the second lower boundary projection is located are obtained in the vertical direction, and the projection values ​​of the number of rows where the second upper boundary projection is located and the number of rows where the second lower boundary projection is located are equal to the second preset projection threshold. Set the projection value of the row number above the row number where the second upper bound projection is located to zero, and set the projection value of the row number below the row number where the second lower bound projection is located to zero. The projection values ​​of the rows between the first upper bound projection row number and the second upper bound projection row number are retained, as well as the projection values ​​of the rows between the first lower bound projection row number and the second lower bound projection row number. S3.4: Define a new image row projection based on the image row projection constrained by local characteristics. The new image line projection The expression is In the formula: Indicates the image at the 1st Projection onto the row; express The Middle The row number of the uppermost row whose projection value is closest to the second preset projection threshold; express The Middle The row number of the uppermost row whose projection value is closest to the first preset projection threshold; express The Middle The row number of the next lower row whose projection value is closest to the first preset projection threshold; express The Middle The row number of the next lower row whose projection value is closest to the second preset projection threshold; S4: Estimate the ship's yaw angle based on the projected pixel values ​​of the optimized projected image; S5: Estimated performance evaluation is performed based on the estimated ship deflection angle and the actual measured ship deflection angle.

2. The method for estimating the ship's deflection angle based on two-dimensional projection of infrared images using local characteristic constraints, as described in claim 1, is characterized in that... The binary image of the ship described in S2 is rotated multiple times according to a preset rotation degree, and the calculation formula is as follows: In the formula: Represents any point in a binary image Located in the image OK; Represents any point in a binary image Located in the image List; Indicates the preset rotation degree counterclockwise; This indicates the row number of any point in the rotated image; This indicates the column number of any point in the rotated image.

3. The method for estimating the ship deflection angle based on two-dimensional projection of infrared images using local characteristic constraints as described in claim 1, characterized in that, In S4, the ship's deflection angle is estimated based on the projected pixel values ​​of the optimized projected image, specifically as follows: S4.1: Determine the new image row projection from top to bottom. The row number where the projection value equals the second preset projection threshold is located. And select the row number Number of consecutive rows with row number ; S4.2: Select the rows from bottom to top. , number of rows and number of rows The number of rows containing the same projection value across all image rows; and the selected row number and the total row number. , number of rows and number of rows First row mean difference The first row mean difference The calculation formula is In the formula: t Indicates the number of rows , number of rows or number of lines Any row number in the array; S4.3: Determine the new image row projection from top to bottom. The row number where the projection value equals the first projection threshold is located. And select the row number Number of consecutive rows with row number ; S4.4: Select the rows from bottom to top. , number of rows and number of rows The number of rows containing the same projection value across all image rows; and the selected row number and the total row number. , number of rows and number of rows The second row of mean difference ; The second row mean difference The calculation formula is In the formula: t Indicates the number of rows , number of rows and number of rows Any row number in the array; S4.5: Calculate the mean difference of the first row in each rotated image. Difference from the average of the second row The difference, and select the mean difference of the first row. Difference from the average of the second row The rotated image with the smallest difference; Based on the first row mean difference Difference from the average of the second row The rotation image with the smallest difference is used to obtain an estimate of the ship's deflection angle in the image. ; And the estimated value of the ship's deflection angle The formula is 。 4. The method for estimating the ship deflection angle based on two-dimensional projection of infrared images using local characteristic constraints as described in claim 1, characterized in that, The calculation formula for performance estimation described in S5 is as follows: In the formula: This represents the ship deflection angle estimated from the image. and the actual measured deflection angle of the ship The absolute value error between them; This indicates the estimated ship deflection angle. and the actual measured deflection angle of the ship The absolute error between them is less than the angle estimation accuracy under a certain angle error tolerance factor; This represents the number of ship images that satisfy the conditions within the parentheses; Representing the Images of ships; Indicates the tolerance factor for angle error; This represents the number of ship images in the entire rotating image set.

Images