Real-time two-point non-uniformity infrared image correction method and device based on prior information

Abstract

The invention discloses a real-time two-point non-uniformity infrared image correction method and device based on prior information, and belongs to the technical field of infrared image correction, and the method comprises the steps: unfolding a plurality of frames of non-uniformity infrared images into one-dimensional vectors, converting all one-dimensional vectors into a two-dimensional matrix, and carrying out the one-dimensional matrix correction; pixel values of each row of the two-dimensional matrix are averagely divided into a plurality of uniform areas after being arranged in an ascending order, two optimal uniform areas for correction are selected and substituted into a two-point correction formula to obtain a correction coefficient vector, the correction coefficient vector is restored to a two-dimensional form, and an original non-uniform scene image is correspondingly corrected. According to the method, the generation of transverse stripes on the image is avoided, the problem of discontinuous and non-real-time image display caused by the time drift problem is solved, the corrected image is balanced, and a good visual effect is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of infrared image correction, and particularly relates to a real-time two-point non-uniformity infrared image correction method and device based on prior information. Background Art

[0002] Infrared imaging systems are widely used in many fields. However, due to the influence of the materials and manufacturing processes of infrared detectors, etc., the problem of imaging non-uniformity will occur, that is, when uniformly radiated light is incident, blocky or stripy patterns will appear on the image, affecting the image quality and measurement accuracy. Therefore, the non-uniformity correction of infrared images has always been an important research direction.

[0003] To solve the problem of non-uniformity, some effective non-uniformity correction methods have been proposed and verified. The two-point correction method is simple and has high accuracy, and is convenient to apply in engineering. It is one of the most widely used correction methods. However, the problem of time drift when using this method makes it often necessary to re-correct at intervals in actual use, which is not conducive to the continuous real-time display of images. Existing technologies have directly applied the two-point correction method to the scene image to solve the time drift problem when solving the non-uniformity problem of push-broom infrared sensors. Subsequently, a similar method was extended to solve the non-uniformity problem of staring detectors, but row-by-row correction will cause new horizontal stripes to appear on the corrected image. It can be seen that the existing non-uniformity infrared image correction methods have technical problems such as non-real-time and discontinuous image display, or new horizontal stripes will appear on the corrected image. Summary of the Invention

[0004] To solve the above technical problems, the present invention adopts the following technical solutions:

[0005] A real-time two-point non-uniformity infrared image correction method based on prior information, comprising:

[0006] Step 1: Unfold F frames of non-uniformity infrared images with size (M, N) into one-dimensional vectors respectively;

[0007] Step 2: Transform the F vectors into a two-dimensional matrix with size (M×N, F);

[0008] Step 3: Arrange the pixel values of each row of the two-dimensional matrix in ascending order;

[0009] Step 4: Divide the two-dimensional matrix into n uniform regions along the vertical direction on average, and the width of each uniform region is L, where L = F / n;

[0010] Step 5: Select two best uniform regions for correction;

[0011] Step 6: Bring the two best uniform regions for calibration into the two-point calibration formula to obtain the gain calibration coefficient vector and the offset calibration coefficient vector;

[0012] Step 7: Restore the gain calibration coefficient vector and the offset calibration coefficient vector to the two-dimensional matrix form to obtain the calibration coefficient matrix;

[0013] Step 8: Calibrate the non-uniform scene image with the calibration coefficient matrix.

[0014] A real-time two-point non-uniformity infrared image calibration device based on prior information, comprising:

[0015] Vector generation module: Expand the F-frame images with size (M, N) into one-dimensional vectors respectively;

[0016] Matrix transformation module: Transform the F vectors into a two-dimensional matrix with size (M×N, F);

[0017] Arrangement module: Arrange the pixel values of each row of the two-dimensional matrix in ascending order;

[0018] Region generation module: Divide the two-dimensional matrix into n uniform regions along the vertical direction, and the width of each uniform region is L, where L = F / n;

[0019] Selection module: Select two best uniform regions for calibration;

[0020] Calibration coefficient vector acquisition module: Bring the two best uniform regions for calibration into the two-point calibration formula to obtain the gain calibration coefficient vector and the offset calibration coefficient vector;

[0021] Calibration coefficient matrix acquisition module: Restore the gain calibration coefficient vector and the offset calibration coefficient vector to the two-dimensional matrix form to obtain the calibration coefficient matrix;

[0022] Calibration module: Calibrate the non-uniform scene image with the calibration coefficient matrix.

[0023] An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, it implements the steps of the real-time two-point non-uniformity infrared image calibration method based on prior information.

[0024] A non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the real-time two-point non-uniformity infrared image calibration method based on prior information.

[0025] The present invention has the following beneficial effects:

[0026] The real-time two-point non-uniformity infrared image correction method based on prior information proposed by the present invention unfolds each of multiple non-uniform scene images into a vector in the same way and rearranges the pixel values in the frame direction. By leveraging the prior information of the laboratory blackbody radiation image, two optimal uniform regions for correction are selected and substituted into the two-point correction formula to obtain the correction coefficient vector. Finally, the correction coefficient vector is restored to the two-dimensional form corresponding to the image pixels, enabling real-time correction of non-uniform infrared scene images. Compared with the existing technologies, the present invention has the following advantages:

[0027] (1) By calculating after expanding the pixels into one-dimensional vectors, the present invention can generate the correction coefficients of all pixels at once. Compared with the row-by-row processing method of the existing technology, it avoids the generation of horizontal stripes, and calculating all pixels simultaneously makes the corrected image more balanced and has a better visual effect.

[0028] (2) By leveraging the prior information of the laboratory blackbody radiation image, the present invention selects two uniform regions closest to the blackbody radiation image under the same radiation for calculation, and can achieve a high-precision correction effect.

[0029] (3) The present invention directly applies two-point correction to the scene image, retaining the advantages of two-point correction while overcoming the problem of discontinuous and non-real-time image display caused by the time drift problem. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flowchart of the real-time two-point non-uniformity infrared image correction method based on prior information of the present invention;

[0031] Figure 2 is the experimental effect diagram of the real-time two-point non-uniformity infrared image correction method based on prior information of the present invention; among them, (a) is the original infrared non-uniform scene image, and (b) is the correction effect diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0033] The present invention proposes a real-time two-point non-uniformity infrared image correction method based on prior information. First, multiple frames of scene images are collected. Each image is unfolded into a vector in the same way and then spliced together to form a matrix. The matrix is sorted in ascending order in the frame direction and evenly divided into multiple uniform regions along the vertical direction. The two uniform regions closest to the blackbody radiation image under the same radiation are selected as the best uniform regions for correction. Substitute them into the two-point correction formula to obtain the correction gain vector and the correction bias vector. Restore the correction coefficient vector to the two-dimensional matrix form, and then the corresponding image can be corrected. Specifically:

[0034] Taking the F-frame images with the size of (M, N) as an example, where M is set to 256, N is set to 320, and F is set to 200, the real-time two-point non-uniformity infrared image correction method based on prior information of the present invention includes the following steps:

[0035] Step 1: Unfold the 200 frames of non-uniformity infrared images with the size of (256, 320) into one-dimensional vectors respectively, and the method is as follows:

[0036] As Figure 1 shown, each of the 200 frames of images with the size of (256, 320) is unfolded into a one-dimensional vector according to the same arrangement method. This step will form 200 vectors with 256×320 rows and 1 column.

[0037] Step 2: Transform the 200 vectors into a matrix with the size of (256×320, 200), and the method is as follows:

[0038] Splice the 200 column vectors with the size of (256×320, 1) together along the row direction to form a two-dimensional matrix with the size of (256×320, 200).

[0039] Step 3: Sort the pixel values of each row of the current two-dimensional matrix in ascending order.

[0040] Step 4: Evenly divide the two-dimensional matrix into 20 uniform regions along the vertical direction (that is, divide the 200 columns into regions, with 10 columns in each region). The width of each uniform region is L = 10 (L = F / n, where F is the number of image frames and n is the number of regions).

[0041] Step 5: Select the two best uniform regions for correction, and the method is as follows:

[0042] First, collect the blackbody radiation images of the detector at different blackbody temperatures in the laboratory, calculate the mean (Mean_blackbody) and standard deviation (Std_blackbody) of the blackbody radiation images at each blackbody temperature. Using the mean (Mean_blackbody) as the independent variable and the standard deviation (Std_blackbody) as the dependent variable, perform a continuous linear function fitting to obtain the fitting continuous linear function relationship F1. At the same time, for the 20 uniform regions obtained from the scene image in step 4, calculate the mean (Mean_scene) and standard deviation (Std_scene) for each uniform region, and establish a discrete function relationship F2 between the two, where the mean (Mean_scene) is the independent variable and the standard deviation (Std_scene) is the dependent variable. On this basis, establish the discrete function F3 = |F2 - F1|. The independent variable of F3 is the mean (Mean_scene), and the dependent variable of F3 is the absolute value of the difference between the standard deviation (Std_scene) and the dependent variable obtained by substituting the mean (Mean_scene) as the independent variable into F1.

[0043] It is considered that the blackbody radiation images with the same mean and the uniform regions have the same incident radiation. Therefore, it is considered that the two uniform regions that minimize F3 are the two uniform regions closest to the blackbody radiation images under the same radiation. In addition, since the means of the two uniform regions being too close will affect the effect of two-point correction, a threshold g is set. When the difference between the means of the uniform regions is greater than the threshold g, the above calculation of the minimum of F3 is performed. The two uniform regions that minimize F3 obtained under the threshold condition are the two best uniform regions for correction.

[0044] In this embodiment, 14-bit images are used. According to experience, the initial value of the threshold g is set to 2000. At the same time, a stepped selection rule for the threshold g is set. If the gray value range of the actual image is less than 2000, the threshold is changed to 1500. If the gray value range of the actual image is smaller, the threshold can be changed to 1000, 500, 200. If it is smaller, the existence of this threshold is not considered.

[0045] Step 6: Substitute the two best uniform regions for correction into the two-point correction formula to obtain the gain correction coefficient vector and the bias correction coefficient vector. The method is as follows:

[0046] Set the region numbers of the two best uniform regions selected in step 5 to be and , and the means of these two regions are and , respectively serving as the ideal responses of each detector element when the two different radiation uniform scenes are incident. At the and The mean value in each horizontal direction within the region and are respectively used as the initial responses to be calibrated, where = 1, 2, ……, 256×320.

[0047] The calibration coefficients of each detector element can be calculated using Equations (1) and (2):

[0048] (1)

[0049] (2)

[0050] Where and can be obtained from Equations (3) and (4), where = 1, 2, ……, 200:

[0051] (3)

[0052] (4)

[0053] Where and are respectively the number of rows and columns of the two-dimensional matrix of (256×320, 200), is the gray value of the th row and th column of the current matrix.

[0054] and can be obtained from Equations (5) and (6):

[0055] (5)

[0056] (6)

[0057] The obtained and are the gain calibration coefficient vector and bias calibration coefficient vector of each detector element, and their sizes are both (256×320, 1).

[0058] Step 7: Restore the gain calibration coefficient vector and bias calibration coefficient vector back to the two-dimensional matrix form to obtain the calibration coefficient matrix, and the method is as follows:

[0059] Restore the gain calibration coefficient vector and bias calibration coefficient vector back to the two-dimensional matrix form of 256 rows and 320 columns. The way of this one-dimensional to two-dimensional transformation is the reverse process of the two-dimensional image to one-dimensional vector in Step 1.

[0060] Step 8: Use the correction coefficient matrix to correct the corresponding non-uniform scene image.

[0061] The experimental effect of the real-time two-point non-uniformity infrared image correction method based on prior information of the present invention is as Figure 2 shown. Figure 2 In (a) is the original infrared non-uniform scene image, Figure 2 and in (b) is the correction effect diagram of the present invention. The image corrected by the present invention can effectively remove the stripe non-uniformity, shot noise and low-frequency non-uniformity with bright in the middle and dark around in the original non-uniform image. The overall image is balanced, the details are well retained, and no horizontal stripes are introduced. Since the present invention is a correction method based on the scene image, while ensuring a good correction effect, it can achieve real-time and continuous display of the image.

[0062] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented in various computer languages. For example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript.

[0063] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0064] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that realizes the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0065] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus, such that a series of operational steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or boxes Figure 1 steps of the functions specified in one box or multiple boxes.

[0066] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic creative concept. Therefore, the appended claims are intended to be construed to cover the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0067] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A real-time two-point non-uniform infrared image correction method based on prior information, characterized in that: include: Step 1: Expand the F frames of non-uniform infrared images with a size of (M, N) into one-dimensional vectors respectively; Step 2: Transform the F vectors into a two-dimensional matrix of size (M×N, F); Step 3: Arrange the pixel values ​​of each row of the two-dimensional matrix in ascending order; Step 4: Divide the two-dimensional matrix into n uniform regions along the vertical direction, each of which has a width of L, where L=F / n; Step 5: Select two optimal uniform regions for correction; Step 6: Substitute the two best uniform regions for correction into the two-point correction formula to obtain the gain correction coefficient vector and the offset correction coefficient vector; Step 7: Restore the gain correction coefficient vector and the offset correction coefficient vector back to a two-dimensional matrix form to obtain a correction coefficient matrix; Step 8: Use the correction coefficient matrix to correct the non-uniform scene image.

2. The real-time two-point non-uniform infrared image correction method based on prior information according to claim 1, characterized in that: In step 1, each of the F frame images of size (M, N) is expanded into a one-dimensional vector in the same arrangement to form F vectors of M×N rows and 1 column.

3. The real-time two-point non-uniform infrared image correction method based on prior information according to claim 1, characterized in that: In step 2, F column vectors of size (M×N, 1) are pieced together along the row direction to form a two-dimensional matrix of size (M×N, F).

4. The real-time two-point non-uniform infrared image correction method based on prior information according to claim 1, characterized in that: In step 5, two optimal uniform areas for correction are selected. The specific method is as follows: first, blackbody radiation images of the detector at different blackbody temperatures are collected in the laboratory, the mean and standard deviation of the blackbody radiation image at each blackbody temperature are calculated, and a continuous linear function fitting is performed with the mean as the independent variable and the standard deviation as the dependent variable to obtain a fitted continuous linear function relationship F1; at the same time, for the n uniform areas obtained from the scene image in step 4, the mean and standard deviation are calculated for each uniform area, and a discrete function relationship F2 between the two is established, wherein the mean is the independent variable and the standard deviation is the dependent variable; on the basis of the above, a discrete function F3=|F2-F1| is established, and it is considered that the blackbody radiation image with the same mean has the same incident radiation as the uniform area, so the two uniform areas that satisfy the minimum F3 are the two uniform areas closest to the blackbody radiation image under the same radiation; when the difference between the means of the two uniform areas is greater than the set threshold, the minimum F3 is calculated, so that the two uniform areas with the minimum F3 are the two optimal uniform areas for correction.

5. The real-time two-point non-uniform infrared image correction method based on prior information according to claim 4, characterized in that: The initial value of the threshold is set to g0, and a step-by-step selection rule for the threshold is set: the threshold is reduced step by step according to the gray value range of the actual image, and if the gray value is too small, the threshold is not set.

6. The real-time two-point non-uniform infrared image correction method based on prior information according to claim 1, characterized in that: Step 6 includes: setting the area numbers of the two best uniform areas selected in step 5 for correction to be and The means of these two regions are and , as the ideal response of each detector element when two uniform scenes with different radiation are incident. and Within the region, the mean value of each horizontal direction and are respectively used as the initial responses to be corrected; The correction coefficient of each detector is calculated using equations (1) and (2): （1） （2） in, and From equation (3) and equation (4), we can get: （3） （4） in, , are the number of rows and columns of the two-dimensional matrix (M, N), The current matrix Line Gray value of the column; and From equation (5) and equation (6), we can obtain: （5） （6） Requested and The gain correction coefficient vector and the bias correction coefficient vector of each detector have sizes of (M×N, 1).

7. The real-time two-point non-uniform infrared image correction method based on prior information according to claim 1, characterized in that: Step 7 includes: restoring the gain correction coefficient vector and the offset correction coefficient vector back to a two-dimensional matrix form of M rows and N columns. This restoration process is the reverse process of step 1.

8. A real-time two-point non-uniform infrared image correction device based on prior information, characterized in that: include: Vector generation module: expand the F frame images of size (M, N) into one-dimensional vectors respectively; Matrix transformation module: transforms F vectors into a two-dimensional matrix of size (M×N, F); Arrangement module: arrange the pixel values ​​of each row of the two-dimensional matrix in ascending order; Region generation module: divide the two-dimensional matrix into n uniform regions along the vertical direction, each uniform region has a width of L, L=F / n; Selection module: selects two best uniform areas for correction; Correction coefficient vector acquisition module: the two best uniform regions for correction are brought into the two-point correction formula to obtain the gain correction coefficient vector and the offset correction coefficient vector; Correction coefficient matrix acquisition module: restore the gain correction coefficient vector and the offset correction coefficient vector back to a two-dimensional matrix form to obtain a correction coefficient matrix; Correction module: Use the correction coefficient matrix to correct the non-uniform scene image.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the real-time two-point non-uniformity infrared image correction method based on prior information as described in any one of claims 1 to 7 are implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the real-time two-point non-uniform infrared image correction method based on prior information as described in any one of claims 1 to 7 are implemented.

Images