Quantitative Detection Method for Image Sawtooth Distortion of Scanning Thermal Imager

Abstract

The present invention relates to the technical field of imaging quality evaluation of scanning infrared thermal imagers, and specifically discloses a method for quantitatively detecting image sawtooth distortion of a scanning thermal imager, including the following steps: turning on the sawtooth distortion detection device, replacing the blackbody with visible light, detecting the vertical line target angle using a total station, and detecting the horizontal field of view size of the thermal imager using a two-dimensional turntable; calculating the theoretical target width W according to the vertical line target angle, horizontal field of view size, and image resolution T ; adjusting the position of the thermal imager through the two-dimensional turntable to make the vertical line target image in the area to be measured of the image, and collecting the current image through a video capture card; writing an image recognition algorithm by a computer to calculate the target image width W of the area to be measured M ; through the formula D S =W M -W T calculate the size of the sawtooth distortion in the area to be measured. The result of detecting the sawtooth distortion by this method is stable and objective. According to the detection result, the size of the image distortion can be quickly and objectively evaluated. The present invention lays a method foundation for establishing an evaluation standard for image sawtooth distortion of a scanning thermal imager.

Description

Technical Field

[0001] The present invention relates to the technical field of imaging quality evaluation of scanning infrared thermal imagers, and particularly relates to a method for quantitatively detecting image sawtooth distortion of a scanning thermal imager. Background Art

[0002] The subjective detection method for image sawtooth distortion of a scanning thermal imager is as follows. After starting up the scanning thermal imager, aim it at the outdoor scene target to be observed. By adjusting the position of the thermal imager, make the target image on the area of the image to be evaluated. Observe the target image on the monitor with the human eye, and subjectively judge the degree of sawtooth distortion of the target image. Adjust the position of the thermal imager in the same way, and subjectively judge the degree of sawtooth distortion in other areas of the thermal imager image, so as to comprehensively evaluate the sawtooth distortion. Such an evaluation method is highly subjective and varies from person to person, often causing evaluation disputes. "Different people have different views." Such a method cannot objectively evaluate the size of sawtooth distortion, and it is also impossible to establish an evaluation standard for the evaluation of sawtooth distortion. Summary of the Invention

[0003] The main purpose of the present invention is to address the shortcoming that the current image sawtooth distortion evaluation method cannot objectively judge the size of image distortion. By using a collimator and a vertical line target to simulate an infinitely distant target, and adopting image acquisition technology and software programming technology to analyze the target image, an objective method for quantitatively detecting image sawtooth distortion is invented. The results of detecting sawtooth distortion by this method are stable and objective. According to the detection results, the size of image distortion can be quickly and objectively evaluated. The present invention lays a methodological foundation for establishing an evaluation standard for image sawtooth distortion of scanning thermal imagers.

[0004] To achieve the above objectives, the present invention provides the following technical solutions:

[0005] A method for quantitatively detecting image sawtooth distortion of a scanning thermal imager, comprising the following steps:

[0006] Step 1: Turn on the sawtooth distortion detection device, replace the blackbody with visible light, and use a total station to detect the vertical line target angle. The sawtooth distortion detection device includes a collimator, on the focal plane of the collimator, a vertical line target is installed, a blackbody is provided below, a two-dimensional turntable is arranged directly in front of the collimator optical path, and a total station is set up. The scanning thermal imager is fixed on the two-dimensional turntable, and the scanning thermal imager is connected to a computer through wired or wireless signals;

[0007] Step 2: Use the two-dimensional turntable to detect the horizontal field of view size of the thermal imager;

[0008] Step 3: Calculate the theoretical target width W according to the vertical line target angle, horizontal field of view size and image resolution T ;

[0009] Step 4: Adjust the position of the thermal imager through a two-dimensional turntable to make the vertical line target image in the area to be measured in the image, and collect the current image through a video capture card;

[0010] Step 5: Write an image recognition algorithm through calculation to calculate the width W of the target image in the area to be measured M ;

[0011] Step 6: Calculate the size of the sawtooth distortion in the area to be measured through Formula 1,

[0012] Formula 1: D S = W M - W T D S D represents the size of the sawtooth distortion in the area to be measured, with the unit of pixel;

[0013] The theoretical target width W in Step 3 T is calculated according to Formula 2 and Formula 3. Formula 2: β = θ / n, Formula 3: W T = α / β, where β represents the horizontal unit pixel, θ represents the horizontal field of view size, n represents the horizontal resolution of the thermal imager, and α represents the target angle size;

[0014] The calculation method of the target image width W in the area to be measured in Step 5 M includes the following steps:

[0015] S1: Image acquisition

[0016] Adjust the position of the thermal imager through a two-dimensional turntable to make the vertical line target image in the area to be measured in the image, and collect k images through a video

[0017] capture card, where k should be no less than 3;

[0018] S2: Measurement selection area setting

[0019] Take the center of the target image as the center of the detection selection area. The horizontal length of the selection area is slightly larger than the width of the target image, and the vertical

[0020] width of the selection area should not be greater than the length of the vertical line target, and can be adjusted appropriately according to the length of the target;

[0021] S3: Target image width measurement algorithm

[0022] Assume that the length (horizontal direction) of the measurement selection area is X pixels, and the width (vertical direction) of the selection area is Y pixels; take the upper left corner of the selection area as the relative coordinate origin, with the right side of the origin as the horizontal positive direction and the lower side of the origin as the vertical positive direction; the gray value of each pixel point in the selection area is represented by G (x,y) and (x, y) represents the coordinate value of the current pixel point relative to the

[0023] coordinate origin;

[0024] S3-1: Detect the boundary coordinates of each vertical line target in the selected area;

[0025] First, use Formula 4 to take the derivative of the grayscale values of each row to obtain the grayscale derivative;

[0026] Formula 4: V (x,y) = G (x+1,y) - G (x,y)

[0027] Use Formula 5 and Formula 6 to find the maximum and minimum values of the grayscale derivative of each row, and record the maximum derivative Vmax (y) The corresponding horizontal coordinate is Xmax (y) , record the minimum derivative Vmin (y) The corresponding horizontal coordinate is Xmin (y) , Xmax (y) is the X coordinate of the left boundary of the target in the y-th row, and Xmin (y) is the X coordinate of the right boundary of the target in the y-th row;

[0028] Formula 5: Vmax (y) = max(V (0,y) , V (1,y) , …… V (X-2,y) , V (X-1,y) )

[0029] Formula 6: Vmin (y) = min(V (0,y) , V (1,y) , …… V (X-2,y) , V (X-1,y) )

[0030] S3-2: Calculate the width of the target image in the selected area of the i-th frame;

[0031] The width of the vertical line target in the selected area can be obtained using Formula 7, Formula 8, and Formula 9; where X L represents the X coordinate of the leftmost contour of the target in the selected area, and X R represents the X coordinate of the rightmost contour of the target in the selected area, and W M(i) represents the width of the target image in the selected area of the i-th frame, in pixels;

[0032] Formula 7: X L = min(Xmax (0) , Xmax (2) , …… Xmax (y-1) , Xmax (y) )

[0033] Formula 8: X R = max(Xmin (0) , Xmin (2) , …… Xmin(y-1) , Xmin (y) )

[0034] Formula 9: W M(i) = |X R - X L |

[0035] S3-3: Calculate the target image width W of multiple images collected in S1 M :

[0036] Use the methods in Article S3-1 and S3-2 to calculate the target width of the k images collected. The target image width of the i-th image is represented by W M(i) , and calculate the vertical line target width according to Formula 10;

[0037] Formula 10:

[0038] The present invention uses a collimator and a vertical line target to simulate an infinitely distant target. The included angle of the vertical line target is measured by a total station or a theodolite, which is called the target included angle; the scanning thermal imager is aligned with the vertical line target of the collimator, and the horizontal field of view size of the thermal imager is detected through a two-dimensional turntable; according to the target included angle, the horizontal field of view size and the image resolution, the theoretical target image width (in pixels) of the vertical line target imaging is calculated; the image is collected to the computer by a video capture card, and then through the computer software programming algorithm, the contour of the vertical line target image is identified, and the size of the pixels occupied by the target image in the horizontal direction (hereinafter referred to as the target image width) is detected. By comparing the target image width value of the area to be measured with the theoretical target image width, the sawtooth distortion value of the target image in different areas can be accurately and objectively obtained, providing an objective judgment basis for the imaging quality evaluation. This method lays a method foundation for establishing the sawtooth distortion evaluation standard of the scanning thermal imager image. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Schematic diagram of the sawtooth distortion detection device;

[0040] Figure 2 Schematic diagram of the vertical line target;

[0041] Figure 3 Central target image display diagram;

[0042] Figure 4 Target image display diagram at the left 90% area;

[0043] Figure 5 Total station "Basic Measurement - Angle Measurement" interface diagram;

[0044] Figure 6 Schematic diagram of the measurement selection area;

[0045] Figure 7 Target contour unclear display diagram;

[0046] Figure 8 Target saturation phenomenon display diagram;

[0047] Figure 9 LX thermal imager sitf test result diagram;

[0048] Figure 10 D03 thermal imager sitf test result diagram;

[0049] Figure 11 D03 scanning thermal imager image center vertical line target image display diagram;

[0050] Figure 12 D03 scanning thermal imager image 90% area left edge vertical line target image display diagram;

[0051] Figure 13 D03 scanning thermal imager image 90% area right edge vertical line target image display diagram;

[0052] Figure 14 Sawtooth distortion detection result display diagram;

[0053] Figure 1 Middle: 1. Collimator; 2. Black body; 3. Two-dimensional turntable; 4. Thermal imager; 5. Computer. Specific implementation mode

[0054] The present invention will be further described below according to the attached drawings and specific embodiments.

[0055] Embodiment 1

[0056] The sawtooth distortion detection device is as Figure 1 shown, including a collimator 1, a vertical line target is installed on the focal plane of the collimator 1, a black body 2 is arranged below, a two-dimensional turntable 3 is arranged directly in front of the optical path of the collimator 1, and a total station is set up. A scanning thermal imager 4 is fixedly installed on the two-dimensional turntable 3. The scanning thermal imager 4 is connected to a computer 5 through a wired or wireless signal. The vertical line target is as Figure 2 shown. The black body provides an infrared heat source. The combination of the vertical line target and the collimator simulates an infinitely distant vertical target. The total station is used to detect the included angle of the vertical line target. The scanning thermal imager is fixedly placed on the electric two-dimensional turntable. The horizontal field of view angle of the thermal imager to be measured is detected through the two-dimensional turntable. The azimuth and pitch angles of the thermal imager relative to the parallel optical path are adjusted so that the vertical line target is imaged in the area to be measured. The image output by the thermal imager is collected into the computer through a video capture card. The computer writes a sawtooth distortion detection program algorithm to calculate the contour and width of the vertical line target; The specific steps include:

[0057] Step 1: Turn on the sawtooth distortion detection device, replace the black body with visible light, and use the total station to detect the included angle of the vertical line target;

[0058] Step 2: Fix the thermal imager on the two-dimensional turntable. After startup, switch the thermal imager to the field of view to be measured through operation, adjust the relevant parameters of the thermal imager, modulate the best imaging state, and use the two-dimensional turntable to detect the horizontal field of view angle of the thermal imager;

[0059] Step 3: Calculate the theoretical target width W according to the vertical line target angle, horizontal field of view angle, and image resolution T (in pixels);

[0060] Step 4: Adjust the position of the thermal imager through the two-dimensional turntable to make the vertical line target image in the area to be measured in the image (the central area, as shown in Figure 3 , at the 90% area of the left edge, as shown in Figure 4 ), and collect the current image through the video capture card;

[0061] Step 5: Write an image recognition algorithm through calculation to calculate the target image width WM in the area to be measured (in pixels);

[0062] Step 6: Calculate the size of the sawtooth distortion in the area to be measured (in pixels) through Formula 1

[0063] Formula 1: D S = W M - W T ; D S represents the size of the sawtooth distortion in the area to be measured.

[0064] Furthermore, the specific data measurement method is as follows:

[0065] I. Measure the vertical line target angle

[0066] The vertical line target is installed on the focal plane of the collimator. Replace the black body in Figure 1 with a visible light source. Set up a total station in front of the collimator optical path. Adjust the horizontal adjustment knob of the total station to make the circular and strip bubbles of the total station both in the center position; Start the total station and enter the basic test interface. Adjust the objective lens focal length and eyepiece adjustment knob of the total station to clearly observe the target contour and the crosshair target plate through the eyepiece; Adjust the azimuth and pitch knobs of the total station to make the vertical line of the total station crosshair coincide with the left (right) edge of the vertical line target. At this time, click the "Set Zero" button in the "Basic Measurement - Angle Measurement" interface (as shown in Figure 5 ), zero the horizontal angle, and then rotate the horizontal adjustment fine-tuning knob to make the vertical line of the total station crosshair coincide with the right (left) edge of the vertical line target. At this time, the horizontal angle displayed in the "Basic Measurement - Angle Measurement" interface is the size of the vertical line target angle. Repeat the measurement operation more than 3 times, and take the average of the 3 measurement results, represented by α;

[0067] II. Measure the field of view size of the thermal imager

[0068] The thermal imager is fixed on a two-dimensional turntable. After startup, the thermal imager is switched to the field of view to be measured by manipulation. The parameters of the thermal imager are adjusted to the optimal imaging state. The azimuth of the two-dimensional turntable is adjusted so that the vertical line target image coincides with the left (right) edge of the image, and the current horizontal angle θ1 of the turntable is recorded. The turntable is rotated so that the vertical line target image coincides with the right (left) edge of the image, and the horizontal angle θ2 of the turntable at this time is recorded; the field of view size θ of the thermal imager = |θ2 - θ1|; III. Calculation of the theoretical target width

[0069] The image resolution of the thermal imager is n*m (unit: pixel), where n represents the horizontal resolution and m represents the vertical resolution; according to formula 2, the angular spread β of a single pixel in the horizontal direction can be calculated, where θ is the field of view size of the thermal imager; according to formula 3, the theoretical target image width W is calculated T , where α is the angular spread of the target

[0070] Formula 2: β = θ / n Formula 3: W T = α / β

[0071] IV. Measurement of the target image width:

[0072] 4.1 Image acquisition

[0073] The position of the thermal imager is adjusted through the two-dimensional turntable so that the vertical line target is imaged in the area to be measured of the image (such as 90% of the left edge area, as shown in Figure 4 ), and k images (k should be no less than 3) are acquired through the video capture card;

[0074] 4.2 Measurement selection area setting

[0075] Taking the center of the target image as the center of the detection selection area, the horizontal length of the selection area is slightly larger than the width of the target image, and the vertical width of the selection area cannot be greater than the length of the vertical line target, which can be adjusted appropriately according to the length of the target, as shown in Figure 6 the red frame in is the measurement selection area;

[0076] 4.3 Target image width measurement algorithm

[0077] Assume that the length (horizontal direction) of the measurement selection area is X pixels and the width (vertical direction) of the selection area is Y pixels; taking the upper left corner of the selection area as the relative coordinate origin, the right side of the origin is the positive horizontal direction, and the lower side of the origin is the positive vertical direction; the gray value of each pixel point in the selection area is represented by G (x,y) , and (x, y) represents the coordinate value of the current pixel point relative to the coordinate origin;

[0078] 4.3.1 Detect the boundary coordinates of the vertical line target in each row within the selection area;

[0079] First, use formula 4 to take the derivative of the gray value of each row to obtain the gray derivative;

[0080] Formula 4: V (x,y) = G(x+1,y) -G (x,y)

[0081] Use formulas 5 and 6 to find the maximum and minimum values of the grayscale derivative for each row, and record the maximum derivative value Vmax (y) The corresponding horizontal coordinate is Xmax (y) , record the minimum derivative value Vmin (y) The corresponding horizontal coordinate is Xmin (y) , Xmax (y) is the X coordinate of the left boundary of the target in the y-th row, and Xmin (y) is the X coordinate of the right boundary of the target in the y-th row;

[0082] Formula 5: Vmax (y) = max(V (0,y) , V (1,y) , …… V (X-2,y) , V (X-1,y) )

[0083] Formula 6: Vmin (y) = min(V (0,y) , V (1,y) , …… V (X-2,y) , V (X-1,y) )

[0084] 4.3.2 Calculate the width of the target image in the selected area of the i-th frame image;

[0085] Use formulas 7, 8, and 9 to find the width of the vertical line target in the selected area; where X L represents the X coordinate of the leftmost contour of the target in the selected area, and X R represents the X coordinate of the rightmost contour of the target in the selected area, and W M(i) represents the width of the target image in the selected area of the i-th frame image (in pixels);

[0086] Formula 7: X L = min(Xmax (0) , Xmax (2) , …… Xmax (y-1) , Xmax (y) )

[0087] Formula 8: X R = max(Xmin (0) , Xmin (2) , …… Xmin (y-1) , Xmin (y) )

[0088] Formula 9: W M(i) = |X R - X L |

[0089] 4.3.2 Calculate the target image width of multiple images collected in Article 4.1:

[0090] Use the methods in Articles 4.3.1 and 4.3.2 to calculate the target width of the k images collected. The target image width of the i-th image is represented by W M(i) and calculate the vertical line target width according to Formula 10;

[0091] Formula 10:

[0092] 4.4 Principles for setting blackbody temperature difference and verification of the accuracy of the target image width measurement algorithm;

[0093] 4.4.1 Influence of blackbody temperature difference on target contour recognition

[0094] The imaging principle of the thermal imager is "temperature difference imaging". When the temperature difference is low, the target signal is weak, the target contour is not clear, and the software algorithm cannot accurately identify the target contour. As shown in Figure 7 , the red frame is the selected measurement area box, and the green frame is the contour detection result; when the temperature difference is too high, the target image will show a "saturation" phenomenon, and there will be a brightness afterglow at the edge of the image target, resulting in the width of the target contour recognized by the software being greater than the actual width, causing inaccurate measurement, as shown in Figure 8 ;

[0095] 4.4.2 Method for setting the "moderate" temperature difference of the blackbody

[0096] Use the SITF test software (the principle of the SITF test software is to detect the size of the target signal in the thermal imager image at different temperature differences) to detect the sitf curve of the thermal imager. As shown in Figure 9 and Figure 10 , it can be seen from Figure 9 that when the blackbody temperature difference is less than 2.108K, the target signal value of the LX thermal imager increases linearly with the temperature difference. At this time, the thermal imager is in the temperature linear region. When the temperature difference is greater than or equal to 2.108K, the target signal value of the LX thermal imager no longer increases linearly with the temperature difference. At this time, the thermal imager image enters the "saturation" state and enters the temperature saturation region. 2.108K is the turning point between the temperature linear region and the temperature saturation region, and 2.108K is called the inflection point temperature difference; it is not difficult to see from Figure 10 that the inflection point temperature differences of different types of thermal imagers are different.

[0097] Sawtooth distortion usually exists in scanning thermal imagers, while staring thermal imagers (the detector is a focal plane array and the imaging process does not require a scanner) do not have sawtooth distortion; use a certain LX staring thermal imager (hereinafter referred to as the LX thermal imager) to verify the target image width measurement algorithm; the parameters of the LX thermal imager are shown in Table 1;

[0098] Table 1 Parameters of LX Thermal Imager

[0099] Thermographic camera name Horizontal field of view (°) Horizontal resolution (pixels) Inflection point temperature difference (K) LX Thermographic camera 2.148 640 2.108

[0100] The black body temperature difference is set to the inflection point temperature difference (2.108 K). Using vertical line targets with different angular spreads and placing the vertical line targets in different areas of the thermal imager, the verification results are shown in Table 2;

[0101] Table 2 Verification Results Table

[0102]

[0103] As can be seen from Table 2, for vertical line targets with different angular spreads, the sawtooth distortion results calculated by the 4.3 target image width measurement algorithm are all less than 0.5 pixels, and the image algorithm recognition accuracy is 1 pixel. The test results are in line with the expected results and are consistent with the fact that there is no sawtooth distortion in this LX thermal imager;

[0104] V. Principle for Selecting the Angular Spread of the Vertical Line Target

[0105] If the angular spread of the vertical line target is too small, the target signal will be weak, the target contour will not be clear, and the contour software recognition algorithm cannot accurately identify the target width. As shown in Table 3, after experimental verification, when the theoretical target image width corresponding to the selected target is greater than or equal to 4, the software algorithm can accurately identify it;

[0106] The software algorithm can accurately identify when the theoretical target image width corresponding to the selected target is greater than or equal to 4;

[0107] Table 3

[0108]

[0109] If the target angular spread is too large, the target image occupies more pixels in the image and spans a larger area, making it impossible to accurately measure the sawtooth distortion size in a specific area. The evaluation area of sawtooth distortion is usually described by a percentage (%), for example, "evaluating the sawtooth distortion size in 90% of the area of XX thermal imager". To accurately detect the sawtooth distortion size in the percentage area, the theoretical target image width corresponding to the selected vertical line target should not exceed 1% of the horizontal resolution of the thermal imager image;

[0110] In summary, the selection of the angular spread of the vertical line target needs to be determined according to the resolution and field of view of the thermal imager to be measured. According to the fourth method above, calculate the theoretical target image width W T , and it should satisfy the relationship of Formula 11, where n is the horizontal resolution of the thermal imager, Formula 11: 4 ≤ W T ≤ n * 1%.

[0111] VI. Measuring the Sawtooth Distortion Sizes at the Center and the Left and Right Edges of the 90% Area of the D03 Scanning Thermal Imager

[0112] 6.1 The parameters of the thermal imager are shown in Table 4

[0113] Table 4 Information Parameters of Thermal Imager and Vertical Line Target

[0114]

[0115] 6.2 Determine the inflection point temperature difference of the thermal imager through the SITF test software, and the results are as Figure 10 shown. It can be obtained from the figure that the inflection point temperature difference is 8.574K.

[0116] 6.3 Collect the vertical line target images. Set the black body temperature difference to 8.574K. After the thermal imager is started, adjust the parameters of the thermal imager to make the imaging of the thermal imager in the best state. Rotate the two-dimensional turntable to make the vertical line target image at the image center, the left edge and the right edge of the 90% area respectively. Collect 10 images at each position. The images are shown in detail in Figure 11 、 Figure 12 and Figure 13 ;

[0117] 6.4 Use the software written according to the "Target Image Width Measurement Algorithm" in Article 4.3 to calculate the sawtooth distortion sizes at the image center, the left edge and the right edge of the 90% area. The software and the sawtooth distortion detection results are as Figure 14 shown.

[0118] From Figure 14 it can be seen that the sawtooth distortion at the center of the D03 thermal imager is very small, and the result is only 0.41 pixel. While the maximum sawtooth distortion at the left edge of the 90% area of the image is 2.41 pixels, and the sawtooth distortion result at the right edge of the 90% area is 1.91 pixels.

[0119] It has been experimentally verified that through this method, the size of the sawtooth distortion of the scanning thermal imager can be quantitatively and objectively evaluated.

[0120] The above embodiments are only for explaining the technical concept and characteristics of the present invention, and the purpose is to enable those who are familiar with this technology to understand the content of the present invention and implement it accordingly, and it cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit and essence of the present invention should be covered within the protection scope of the present invention.

Claims

1. A quantitative detection method for image sawtooth distortion of a scanning thermal imager, characterized in that, It includes the following steps: Step 1: Turn on the sawtooth distortion detection device, replace the black body with visible light, and use a total station to detect the vertical line target opening angle. The sawtooth distortion detection device includes a collimator. A vertical line target is installed on the focal plane of the collimator, and a black body is arranged below. A two-dimensional turntable is set in the front of the collimator optical path, and a total station is set up. A thermal imager is fixed on the two-dimensional turntable, and the thermal imager is connected to a computer through wired or wireless signals; Step 2: Use the two-dimensional turntable to detect the horizontal field of view size of the thermal imager; Step 3: Calculate the theoretical image width W of the vertical line target according to the vertical line target opening angle, horizontal field of view size, and image resolution T ; Step 4: Adjust the position of the thermal imager through the two-dimensional turntable to make the vertical line target image in the area to be measured in the image, and collect the current image through a video capture card; Step 5: Write an image recognition algorithm by computer to calculate the width W of the vertical line target image in the area to be measured M ; Step 6: Calculate the size of the sawtooth distortion in the area to be measured through Formula 1, Formula 1: D S = W M - W T ; D S represents the size of the sawtooth distortion of the area to be measured, in pixels; The theoretical vertical line target image width W in step 3 T Calculated according to Formula 2 and Formula 3. Formula 2: β = θ / n, Formula 3: W T = α / β, where β represents the horizontal unit pixel, θ represents the horizontal field of view size, n represents the horizontal resolution of the thermal imager, and α represents the vertical line target angle size; In step 5, the vertical line target image width W of the area to be tested M The calculation method includes the following steps: S1: Image acquisition Adjust the position of the thermal imager through the two-dimensional turntable to make the vertical line target image in the area to be measured in the image, and collect k images through a video capture card. k should be not less than 3; S2: Measurement selection area setting Take the center of the vertical line target image as the center of the detection selection area. The horizontal length of the selection area is greater than the width of the vertical line target image, and the vertical width of the selection area should not be greater than the length of the vertical line target, and adjust according to the length of the vertical line target; S3: Algorithm for measuring the width of the vertical line target image Suppose the horizontal length of the measured selection area is X pixels, and the vertical width of the selection area is Y pixels; taking the upper left corner of the selection area as the relative coordinate origin, the right side of the origin is the positive horizontal direction, and the lower side of the origin is the positive vertical direction; the grayscale value of each pixel point in the selection area is represented by G (x,y) is represented, and (x, y) represents the coordinate value of the current pixel point relative to the coordinate origin; S3-1: Detect the boundary coordinates of each row of the vertical line target in the selection area; First, use Formula 4 to take the derivative of the gray value of each row to obtain the gray derivative; Formula 4: V (x,y) = G (x+1,y) - G (x,y) Use Equation 5 and Equation 6 to find the maximum and minimum values of the gray-scale derivative for each row, and record the maximum derivative value Vmax (y) The corresponding horizontal coordinate is Xmax (y) , record the minimum derivative value Vmin (y) The corresponding horizontal coordinate is Xmin (y) , Xmax (y) is the x coordinate of the left boundary of the vertical line target in the y-th row, and Xmin (y) is the x coordinate of the right boundary of the vertical line target in the y-th row; Formula 5: Vmax (y) = max(V (0,y) , V (1,y) , …… V (X-2,y) , V (X-1,y) ) Formula 6: Vmin (y) = min(V (0,y) , V (1,y) , …… V (X-2,y) , V (X-1,y) ) S3-2: Calculate the width of the vertical line target image in the selection area of the i-th frame image; The width of the vertical line target image in the selected area is obtained by using Equation 7, Equation 8, and Equation 9; where X L represents the x coordinate of the leftmost contour of the vertical line target in the selected area, X R represents the x coordinate of the rightmost contour of the vertical line target in the selected area, W M(i) represents the width of the vertical line target image in the selected area of the i-th frame, in pixels; Formula 7: X L = min(Xmax (0) , Xmax (2) , …… Xmax (Y-1) , Xmax (Y) ) Formula 8: X R = max(Xmin (0) , Xmin (2) , …… Xmin (Y-1) , Xmin (Y) ) Formula 9: W M(i) = |X R - X L | S3-3: Calculate: the vertical target image width W of multiple images collected by S1 M : Calculate the vertical target image width of the k captured images using the methods of Articles S3-1 and S3-2. The vertical target image width of the i-th image is denoted by W M(i) and calculate the vertical target image width according to Formula 10; Formula 10:

2. The quantitative detection method for the sawtooth distortion of the scanning thermal imager image according to claim 1, wherein, During the detection process, the temperature difference of the black body is set to the inflection point temperature difference of the measured scanning thermal imager.

3. The quantitative detection method for the sawtooth distortion of the scanning thermal imager image according to claim 1, wherein The theoretical vertical line target image width W T should satisfy the relationship of Formula 11, where n is the horizontal resolution of the thermal imager, and Formula 11: 4 ≤ W T ≤ n * 1%.

Images