Method for Generating Rotational Motion Space-Variant Blurred Images

Abstract

The present invention belongs to the field of infrared imaging technology, and particularly relates to a method for generating a spatially variant blurred image of rotational motion, which is used to solve the problem of simulating and generating blurred images in the case of rotational motion in an imaging system, and provides support for the analysis and demonstration of various infrared imaging systems. The method proposes a method for generating a motion blurred image with spatially variant characteristics when there is rotational motion in an infrared imaging system. A blur kernel is generated for each pixel, and the template direction of the conventional motion blur model is optimized. The method specifically includes: establishing an image coordinate system, generating a rotational motion blur template, obtaining all blurred pixels based on the rotational motion blur template, and thus generating a final rotational motion blurred image. The present invention gives a specific description of the method for generating the blurred image of rotational motion, which can provide support for the analysis and demonstration of various infrared imaging systems.

Description

Technical Field

[0001] The present invention belongs to the technical field of infrared imaging, and particularly relates to a method for generating a spatially variant blurred image of rotational motion, which is used to solve the problem of simulating and generating a blurred image in the case of rotational motion of an imaging system, and provides support for the analysis and demonstration of various infrared imaging systems. Background Art

[0002] Infrared imaging has the technical advantages of all-day, quasi-all-weather, and passive thermal imaging, and has been widely used in both military and civilian fields. During actual use, relative motion occurs between the imaging system and the scene, causing image pixels to be sensitive to the moving scene information during the exposure time, smoothing the original image and resulting in motion blur. In an infrared imaging system, since the exposure time is longer than that of a visible light imaging system, the impact of motion blur on image degradation is more serious. Common degradation models are usually based on the principle of space-invariant, that is, the motion blur of all pixels in the image is the same. The most common problem of this type is linear motion blur. However, during actual use, the imaging system may have rotational motion. At this time, the motion blur kernels between pixels in the image are not the same, and common degradation models cannot be used. Summary of the Invention

[0003] (1) Technical Problems to be Solved

[0004] The technical problem to be solved by the present invention is: how to propose a method for generating a motion blurred image with spatially variant characteristics when the infrared imaging system has rotational motion.

[0005] (2) Technical Solutions

[0006] To solve the above technical problems, the present invention provides a method for generating a spatially variant blurred image of rotational motion, and the method includes the following steps:

[0007] Step 1: Establish an image coordinate system, with the coordinate origin o as the center of the image, the x-axis horizontally to the right, and the y-axis vertically upward;

[0008] Step 2: Assume that the angular velocity of the rotational motion of the imaging system is ω (rad / s), and it is positive when counterclockwise;

[0009] Step 3: Assume that the input image is I, and p ij is a pixel point in the image I, and i and j respectively represent the coordinate positions of the pixel point on the x-axis and y-axis; a is the pixel size; R ij is the distance of the pixel p ij from the origin;

[0010]

[0011] Step 4: Starting from the 0° position, considering the characteristics of rotational blur, the rotational blur phenomenon is very slight when the distance from the origin is very close. Therefore, calculations are started from the pixel position at a distance of R0 = 10a, that is, i = 10, j = 0, and this pixel is denoted as p 0 , and let it belong to set C0;

[0012] Step 5: Traverse each pixel in the eight-neighborhood of p ij , calculate the distance R of each pixel point to the origin respectively, and find the pixel point p uv where the difference between R and R0 is the smallest, and ensure that |R - R0| < a. If such a point does not exist within the eight-neighborhood range, the neighborhood range is expanded until it is found; this pixel is denoted as p 1 , and let it belong to set C0; where u and v respectively represent the coordinate positions of the pixel point on the x-axis and y-axis;

[0013] Step 6: According to pixel p 1 , calculate the angular increment Δθ, as shown in Equation (2);

[0014] Δθ = arg tan(u / v) (2)

[0015] Step 7: With R0 as the radius and Δθ as the angular increment, that is, (2Δθ, 3Δθ,... kΔθ), loop once around, and calculate the image coordinates at each angular position one by one; k is the loop sequence number, that is, the superposition times of the angular increment;

[0016]

[0017] Step 8: If the pixel has not been included in set C0, then include it in C0;

[0018] Step 9: Calculate the number n of motion-blurred pixels caused by rotation r ;

[0019] l = t int ωR0 (4)

[0020]

[0021] where t int (s) is the integration time of the imaging system;

[0022] Step 10: Generate the rotational motion blur template of set C0; the rotational motion blur template is a weight value sequence composed of 0 and 1;

[0023] If the angular velocity is positive, then in the motion blur template caused by rotation, the number of 1s is n r and the number of 0s is n r - 1; that is, this weight value sequence is composed of nr -1 zeros and n r ones in sequence;

[0024] If the angular velocity is negative, in the motion blur template caused by rotation, the number of ones is n r -1, and the number of zeros is n r ; that is, the weight value sequence consists of n r -1 ones and n r zeros in sequence;

[0025] Step 11: For the pixel p in set C0 k Perform convolution operation using the above rotation motion blur template to obtain the blurred pixel g after generating rotation blur ij ;

[0026] Step 12: Expand the radius, set R0 = R0 + a = 11a, and repeat steps (4) to (11) until all pixels are processed, and then generate the final rotation motion blur image G from all the blurred pixels g ij .

[0027] Among them, in step 2, the unit of the angular velocity ω is rad / s.

[0028] Among them, in step 3, a is the pixel size.

[0029] Among them, in step 5, u and v respectively represent the coordinate positions of the pixel point on the x-axis and y-axis.

[0030] Among them, in step 7, k is the loop sequence number, that is, the superposition times of the angle increment.

[0031] Among them, in step 9, t int (s) is the integration time of the imaging system.

[0032] Among them, in step 10, if the angular velocity is positive, in the motion blur template caused by rotation, the number of ones is n r ones, and the number of zeros is n r -1 ones.

[0033] Among them, in step 10, if the angular velocity is positive, in the motion blur template caused by rotation, its weight value sequence consists of n r -1 zeros and n r ones in sequence.

[0034] Among them, in step 10, if the angular velocity is negative, in the motion blur template caused by rotation, the number of ones is n r -1 ones, and the number of zeros is n r ones.

[0035] Among them, in step 10, if the angular velocity is negative, in the motion blur template caused by rotation, the weight value sequence is composed of n r -1 ones and n r zeros in sequence.

[0036] (III) Beneficial effects

[0037] The present invention proposes a method for generating a motion-blurred image with spatially variant characteristics when there is rotational motion in an infrared imaging system. A blur kernel is generated for each pixel, and the template direction of the conventional motion blur model is optimized.

[0038] The present invention gives a specific description of the method for generating a blurred image of the rotational motion, which can provide support for the analysis and demonstration of various infrared imaging systems. Description of the drawings

[0039] Figure 1 It is the motion blur template when rotating counterclockwise.

[0040] Figure 2 It is the motion blur template when rotating clockwise.

[0041] Figure 3 It is the original image.

[0042] Figure 4 It is the schematic diagram of set C0 at radius R0.

[0043] Figure 5 It is the finally generated rotational motion blurred image. Detailed implementation manners

[0044] To make the objectives, contents, and advantages of the present invention clearer, the following further describes in detail the specific implementation manners of the present invention with reference to the drawings and embodiments.

[0045] To solve the above technical problems, the present invention provides a method for generating a spatially variant blurred image of rotational motion, and the method includes the following steps:

[0046] Step 1: Establish an image coordinate system, with the coordinate origin o as the center of the image, the x-axis horizontally to the right, and the y-axis vertically upward;

[0047] Step 2: Let the angular velocity of the rotational motion of the imaging system be ω (rad / s), which is positive when rotating counterclockwise;

[0048] Step 3: Let the input image be I, and p ij be a pixel point in the image I, where i and j respectively represent the coordinate positions of the pixel point on the x-axis and y-axis; a is the pixel size; R ij is the distance of the pixel p ij from the origin;

[0049]

[0050] Step 4: Starting from the 0° position, considering the characteristics of rotational blur, the rotational blur phenomenon is very slight when the distance from the origin is very close; therefore, the calculation starts from the pixel position at a distance of R0 = 10a, that is, i = 10, j = 0, and this pixel is denoted as p 0 , and let it belong to set C0;

[0051] Step 5: Traverse p ij each pixel in its eight-neighborhood, calculate the distance R of each pixel point to the origin respectively, and find the pixel point p uv where the difference between R and R0 is the smallest, and ensure that |R - R0| < a. If such a point does not exist within the eight-neighborhood range, expand the neighborhood range until it is found; denote this pixel as p 1 , and let it belong to set C0; where u and v respectively represent the coordinate positions of the pixel point on the x-axis and y-axis;

[0052] Step 6: According to pixel p 1 , calculate the angle increment Δθ, as shown in Equation (2);

[0053] Δθ = arg tan(u / v) (2)

[0054] Step 7: Using R0 as the radius and Δθ as the angle increment, that is, (2Δθ, 3Δθ,... kΔθ), loop one week, and calculate the image coordinates of each angular position one by one; k is the loop sequence number, that is, the superposition times of the angle increment;

[0055]

[0056] Step 8: If the pixel has not been included in set C0, then include it in C0;

[0057] Step 9: Calculate the number n of motion-blurred pixels caused by rotation r ;

[0058] l = t int ωR0 (4)

[0059]

[0060] where t int (s) is the integration time of the imaging system;

[0061] Step 10: Generate a rotational motion blur template for set C0; the rotational motion blur template is a weight value sequence composed of 0 and 1;

[0062] If the angular velocity is positive, the motion blur template caused by rotation is as follows Figure 1 shown, where the number of 1s is n r and the number of 0s is n r - 1; that is, the weight value sequence is composed of n r - 1 0s and n r 1s in sequence;

[0063] If the angular velocity is negative, the motion blur template caused by rotation is as follows Figure 2 shown, where the number of 1s is n r - 1 and the number of 0s is n r ; that is, the weight value sequence is composed of n r - 1 1s and n r 0s in sequence;

[0064] Step 11: Perform convolution operation on the pixel p of the set C0 k using the above rotation motion blur template to obtain the blurred pixel g after generating rotation blur ij ;

[0065] Step 12: Expand the radius, set R0 = R0 + a = 11a, and repeat steps (4) to (11) until all pixels are processed, and then generate the final rotation motion blur image G from all the blurred pixels g ij .

[0066] Example 1

[0067] This example uses an infrared image obtained by a refrigerated infrared imager with a resolution of 640×512 to illustrate the application of the method of the present invention.

[0068] Step 1: Input the original image I as Figure 3 shown.

[0069] Step 2: Set the rotation angular velocity ω = 15 rad / s, rotate clockwise, t int = 0.004 s, and take the distance R0 = 200a for illustration. Its set C0 is as Figure 4 shown. The length of the blur kernel is 12.

[0070] Step 3: The finally generated rotation motion blur image is as Figure 5 shown.

[0071] The above is only the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and deformations can be made, and these improvements and deformations should also be regarded as the protection scope of the present invention.

Claims

1. A method for generating a spatially variant blurred image with rotational motion, characterized in that The method includes the following steps: Step 1: Establish an image coordinate system with the coordinate origin o as the center of the image, the x-axis horizontally to the right, and the y-axis vertically upward; Step 2: Assume that the angular velocity of the rotational motion of the imaging system is ω (unit: rad / s), and it is positive when rotating counterclockwise; Step 3: Let the input image be I, and p ij be a pixel point in the image I, where i and j represent the coordinate positions of the pixel point on the x-axis and y-axis respectively; a is the pixel size; R ij be the distance of the pixel p ij from the origin; Step 4: Starting from the 0° position, considering the characteristics of rotational blur, the rotational blur phenomenon is very slight when very close to the origin; therefore, start the calculation from the pixel position at a distance of R0 = 10a, that is, i = 10, j = 0, and denote this pixel as p 0 , and let it belong to set C0; Step 5: Traverse p ij For each pixel in the eight-neighborhood, calculate the distance R from each pixel point to the origin respectively, and find the pixel point p where the difference between R and R0 is the smallest uv , and ensure that |R - R0| < a. If such a point does not exist within the eight-neighborhood range, expand the neighborhood range until it is found; denote this pixel as p 1 , and let it belong to set C0; where u and v respectively represent the coordinate positions of the pixel point on the x-axis and y-axis Step 6: According to pixel p 1 , calculate the angle increment Δθ as shown in Equation (2); Δθ = argtan(u / v) (2) Step 7: With R0 as the radius and Δθ as the angular increment, that is, 2Δθ, 3Δθ,... kΔθ, cycle through one full circle and calculate the image coordinates at each angular position one by one; k is the cycle sequence number, that is, the superposition times of the angular increment; Step 8: If the pixel has not been included in set C0, then include it in C0; Step 9: Calculate the number n of motion-blurred pixels caused by rotation r ; l = t int ωR0 (4) where t int is the integration time of the imaging system; Step 10: Generate a rotational motion blur template for set C0; the rotational motion blur template is a weight value sequence composed of 0 and 1; If the angular velocity is positive, in the motion blur template caused by rotation, the number of 1s is n r and the number of 0s is n r −1; that is, this weight value sequence consists of n r −1 0s and n r 1s in sequence; If the angular velocity is negative, in the motion blur template caused by rotation, the number of 1s is n r -1, and the number of 0s is n r ; that is, this weight value sequence consists of n r -1 1s and n r 0s in sequence; Step 11: Pixel p of set C0 k Perform convolution operation using the above rotational motion blur template to obtain the blurred pixel g after generating rotational blur ij ; Step 12: Increase the radius, set R0 = R0 + a = 11a, and repeat steps (4) to (11) until all pixels are processed. Then, from all the blurred pixels g ij , generate the final rotation motion blurred image G.

Images