Method and system for restoring and enhancing nuclear power station fuel assembly character low-quality image

Abstract

The invention discloses a restoration enhancement method and system for a nuclear power station fuel assembly character low-quality image, and relates to the technical field of underwater image enhancement, restoration is performed based on a structure forest, and the method specifically comprises the steps: carrying out the edge extraction of a spectrogram of an original image through the structure forest; estimating a point spread function based on the structure forest edge map by adopting a straight line detection and projection mapping algorithm; after estimating a point spread function, restoring the original image through Wiener filtering to obtain a restored clear image; the method specifically comprises the following steps of: obtaining a high-frequency enhanced image by adopting an improved Sobel operator; a low-frequency enhanced image is obtained based on a curve mapping method; and carrying out pixel-level weighted fusion on the high-frequency and low-frequency synchronously enhanced images based on the unsharpened mask to obtain a final enhanced image. The method has the capability of restoring and enhancing character low-quality imaging of the underwater fuel assembly, and solves the two problems of insufficient image character contrast and poor definition at the same time.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater image enhancement, and in particular to a method and system for restoring and enhancing low-quality images of characters of fuel assemblies of nuclear power plants. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] Nuclear reactor fuel assemblies are important core components of nuclear power plants, and their normal maintenance and management are one of the keys to ensuring the normal operation of nuclear power plants. Fuel assemblies are often distinguished and managed by character numbers engraved on their surfaces. The character numbers generally include information such as batch, model, and production date. Therefore, accurate identification of character information is required during normal operation and maintenance.

[0004] Fuel assemblies are installed underwater, which poses a great challenge to the visual-based character number recognition system. On the one hand, the underwater lighting conditions are insufficient, the imaging quality of the character numbers on the fuel assemblies is poor, and the image contrast is low. In this case, both the visual recognition system and manual operation recognition are difficult to accurately distinguish; on the other hand, the turbulence caused by the irregular flow of water will also cause image blur, making it more difficult to recognize the character numbers and making it difficult to accurately recognize them. Summary of the invention

[0005] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a method and system for restoring and enhancing low-quality images of nuclear power plant fuel assembly characters, which adopts fuzzy image restoration based on structured forest and unsharp masking technology with synchronous fusion of high and low frequencies, and solves the problem of preprocessing and enhancement of low-contrast and fuzzy optical images in the visual recognition system of nuclear power plant fuel assemblies.

[0006] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0007] In a first aspect, the present invention provides a method for restoring and enhancing a low-quality image of a character of a nuclear power plant fuel assembly, comprising:

[0008] Get the original image of the fuel assembly character;

[0009] The edge extraction based on the structure forest is applied to the estimation of the parameters of the point spread function, and the original image is restored to obtain a restored clear image; the image restoration based on the structure forest specifically includes: extracting the edge of the spectrum transform map of the original image based on the structure forest algorithm, and estimating the point spread function by using the line detection and projection mapping algorithm; after estimating the point spread function, the original image is restored by Wiener filtering to obtain a restored clear image;

[0010] An image enhancement algorithm based on an unsharp mask is used to perform contrast enhancement on the clear image to obtain a final enhanced image; the image enhancement algorithm based on an unsharp mask specifically includes: using an improved Sobel operator to enhance the high-frequency information of the clear image to obtain a high-frequency enhanced image; using a curve mapping method to enhance the low-frequency information of the clear image to obtain a low-frequency enhanced image; based on an unsharp mask, the high-frequency enhanced image and the low-frequency enhanced image are synchronously pixel-level weighted fusion enhanced to obtain a final enhanced image.

[0011] A further technical solution is to restore the original image, specifically comprising:

[0012] Performing spectrum transformation on the original image to obtain a spectrum transformation graph of the original image;

[0013] Performing edge extraction on the spectrum transformation graph based on a structured forest algorithm to obtain an edge graph;

[0014] Performing straight line detection on the edge map by a straight line detection algorithm to estimate the blur angle of the point spread function; rotating the edge map according to the blur angle, and then performing projection transformation to obtain a projection value map, and estimating the blur scale of the point spread function by the projection value map;

[0015] The original image is inversely transformed based on Wiener filtering to obtain a restored clear image.

[0016] A further technical solution is to perform straight line detection and extraction on the edge map through Radon transform to obtain a Radon transform map, and obtain the blur angle through the feature points in the Radon transform map.

[0017] A further technical solution is to rotate the edge map around the center by a blur angle and then estimate the blur scale based on a projection method.

[0018] According to a further technical solution, the improved Sobel operator is to improve the gradient weight of the Sobel convolution kernel in a set direction.

[0019] A further technical solution is that the mapping curve expression in the curve mapping method is:

[0020]

[0021] Among them, γ is the mapping coefficient.

[0022] In a further technical solution, the unsharp mask body is represented as:

[0023]

[0024] Among them, f(x,y) fusionTo enhance the final image, To enhance the image at high frequencies, is the low frequency enhanced image, and α is the weight parameter.

[0025] In a second aspect, the present invention provides a restoration and enhancement system for low-quality images of characters of nuclear power plant fuel assemblies, comprising:

[0026] An image acquisition module is configured to: acquire an original image of a fuel assembly character;

[0027] The image restoration module is configured to: apply the edge extraction based on the structure forest to the estimation of the point spread function parameters, restore the original image, and obtain a restored clear image; the image restoration based on the structure forest specifically includes: extracting the edge of the spectrum transform map of the original image based on the structure forest algorithm, estimating the point spread function by using the line detection and projection mapping algorithm; after estimating the point spread function, restore the original image by using the Wiener filter to obtain a restored clear image;

[0028] A contrast enhancement module is configured to: use an image enhancement algorithm based on an unsharp mask to perform contrast enhancement on the clear image to obtain a final enhanced image; the image enhancement algorithm based on an unsharp mask specifically includes: using an improved Sobel operator to enhance the high-frequency information of the clear image to obtain a high-frequency enhanced image; using a curve mapping method to enhance the low-frequency information of the clear image to obtain a low-frequency enhanced image; based on the unsharp mask, synchronously perform pixel-level weighted fusion enhancement on the high-frequency enhanced image and the low-frequency enhanced image to obtain a final enhanced image.

[0029] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for restoring and enhancing low-quality images of characters of a nuclear power plant fuel assembly as described in the first aspect.

[0030] In a fourth aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for restoring and enhancing low-quality images of characters of nuclear power plant fuel assemblies as described in the first aspect are implemented.

[0031] One or more of the above technical solutions have the following beneficial effects:

[0032] The present invention firstly restores the blurred image based on the edge extraction of structured forest and applies it to the point spread function parameter estimation, and then performs synchronous pixel-level weighted fusion enhancement on the high- and low-frequency enhanced images based on the unsharp mask, and performs enhancement preprocessing on the low-contrast, blurred image of the characters of the nuclear power plant fuel assembly. The image quality, especially the character contrast, is significantly improved, laying a solid foundation for the subsequent character detection and recognition.

[0033] The present invention has the ability to restore and enhance low-quality imaging of underwater fuel assembly characters. For underwater low-contrast and blurred images, it first proposes a technical solution of first restoring the image based on structured forests, and then performing synchronous pixel-level weighted enhancement of high- and low-frequency enhanced images based on unsharp masks. At the same time, it solves the two major problems of insufficient image contrast and poor clarity.

[0034] The structure forest-based edge extraction algorithm used in the present invention is more robust, less dependent on parameters, and the extracted edges are more complete, thereby ensuring the estimation accuracy of the point spread function.

[0035] The present invention improves and optimizes the traditional Soble convolution operator to adapt to the two character morphology distribution directions of 45° and 135°, thereby improving the enhancement effect of the high-frequency information of the image.

[0036] The present invention is based on the unsharp mask enhancement method. For the first time, the high-frequency and low-frequency information are simultaneously enhanced and then pixel-level weighted fusion is performed. Compared with the traditional method, the overall contrast of the image is significantly improved. After processing, the image-to-character ratio is increased by 155%, which greatly improves the subsequent detection and recognition success rate of the characters. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0038] Figure 1 is a flow chart of a restoration enhancement method according to an embodiment of the present invention;

[0039] Figure 2 It is a diagram showing the relationship before and after mapping of different γ values ​​in the curve mapping method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0041] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0042] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0043] Embodiment 1

[0044] like Figure 1 As shown, this embodiment discloses a method for restoring and enhancing a low-quality image of a nuclear power plant fuel assembly character, the method comprising the following steps:

[0045] S1: Get the original image of the fuel assembly characters.

[0046] In this embodiment, the fuel assembly is usually installed underwater. Due to insufficient underwater illumination, the character number imaging quality on the fuel assembly is poor, and the contrast of the acquired fuel assembly character image is low. In addition, the turbulence caused by the irregular flow of water causes the acquired fuel assembly character image to be blurred, which seriously affects the accurate recognition of the fuel assembly character number.

[0047] The original image of the underwater fuel assembly characters is captured by a visual recognition system, i.e., a visual-based character number recognition system, and then processed. The visual recognition system is an existing recognition system and will not be described in detail here.

[0048] S2: applying the edge extraction based on the structure forest to the estimation of the parameters of the point spread function, restoring the original image, and obtaining a restored clear image.

[0049] In this embodiment, the original image is subjected to spectrum transformation to obtain a spectrum transformation diagram of the original image;

[0050] The edge of the spectrum transformation graph is extracted based on the structure forest algorithm to obtain an edge graph;

[0051] Performing straight line detection on the edge map by a straight line detection algorithm to estimate the blur angle of the point spread function; rotating the edge map according to the blur angle, and then performing projection transformation to obtain a projection value map, and estimating the blur scale of the point spread function by the projection value map;

[0052] The original image is inversely transformed based on the Wiener filter to obtain a restored clear image.

[0053] Specifically:

[0054] (1) Perform a spectrum transformation based on Fourier transform on the original image (blurred image) to obtain a spectrum transformation diagram of the original image.

[0055] Image blur caused by motion can generally be restored by estimating the point spread function (PSF) of the optical imaging system. The PSF parameters of motion blur include blur angle (direction) and blur scale (length). The estimation of point spread parameters is to obtain blur angle and blur scale.

[0056] The blurred image degradation model can be expressed as:

[0057] g(x,y)=f(x,y)*h(x,y) (1)

[0058] Among them, g(x,y) is the blurred image (original image), f(x,y) is the clear image, h(x,y) is the point spread function (blur kernel), and * represents the convolution operation.

[0059] In fact, image restoration is to estimate the point spread function h(x,y) based on the known blurred image g(x,y), and then obtain the clear image f(x,y). The estimation of the point spread function is generally performed in the frequency domain, and the model of formula (1) is Fourier transformed into:

[0060] G(u,v)=F(u,v)·H(u,v). (2)

[0061] Assuming that within a short exposure time T, the displacement of the target object along the vertical direction is a, and the displacement along the vertical direction is b, we can further obtain:

[0062]

[0063] From the above formula, we can see that the point spread function spectrum amplitude |H(u,v)| value has the characteristics of the sinc function, and the blurred image spectrum transformation diagram also has the characteristics of the sinc function, which is manifested as parallel stripes of light and dark in the spectrum transformation diagram, such as Figure 1 (b) as shown.

[0064] (2) The edge of the spectrum transformation graph is extracted based on the structured forest algorithm to obtain an edge graph.

[0065] By detecting the inclination of the bright and dark parallel stripes in the blurred image spectrum transformation diagram, the point spread function parameter motion blur angle θ can be estimated, and the point spread function parameter blur scale L can be estimated by the blur angle and the spacing between adjacent dark stripes. Therefore, the key to the spread function estimation is to obtain the statistics of the geometric features of parallel stripes, and the statistics of the geometric features of parallel stripes depend on the extraction of image edges.

[0066] This embodiment uses the structural forest edge extraction algorithm. The structural forest edge detection algorithm was proposed by Piotr et al. This algorithm uses structured learning for edge detection and uses image blocks instead of pixels to detect the edge of the image. It not only utilizes the image spatial information, but also retains the computing performance of the random forest, and has both accuracy and efficiency. Compared with the traditional canny edge detection operator, the structural forest algorithm has less parameter dependence and the extracted edges are more complete. The extraction results are shown in Figure 1. Figure 1 (c) as shown.

[0067] (3) performing straight line detection on the edge map using a straight line detection algorithm to estimate the blur angle of the point spread function; rotating the edge map according to the blur angle, and then performing a projection transformation to obtain a projection value map, and estimating the blur scale of the point spread function using the projection value map.

[0068] After edge extraction is completed, it is necessary to perform straight line detection and extraction on the edge map to achieve statistics of stripe angles and intervals. Commonly used straight line detection algorithms include Radon transform and Hough transform. Hough transform requires the edge map to be binarized first, which involves the selection of threshold parameters, while Radon transform does not require this step and is more accurate and more robust. Therefore, this embodiment uses Radon transform for straight line detection and extraction. That is, in this embodiment, the edge map is detected and extracted by Radon transform to obtain a Radon transform map.

[0069] The basic principle of Radon transform line detection is to project the image in the direction of a specified angle ray. When the projection transformation is performed along the straight line direction, significant feature points will be formed in the transform domain image (i.e., Radon transform image), that is, the feature processing of the points on the transform domain image is converted, and the blur angle of the point spread function is obtained through the feature points.

[0070] Radon transform diagram Figure 1 As shown in (d), the maximum position of the Radon transform image is the blur angle of the point spread function.

[0071] In some other implementations, this step may also use Hough transform to perform line detection.

[0072] After rotating the obtained edge image around the center by a blur angle, the blur scale is estimated based on the projection method. Specifically:

[0073] After obtaining the straight line angle θ, the edge graph needs to be rotated around the center by θ. The transformation matrix is:

[0074]

[0075] Among them, x and y are the image coordinates before rotation, xt ,y t is the image coordinate after rotation.

[0076] The rotated image is directly projected along the vertical direction to obtain the projection value map. The projection result is as follows: Figure 1 As shown in (e), in order to facilitate the subsequent extreme value statistics, the projection value graph is inverted here, expressed as:

[0077] invSum pro =MAX(Sum pro )-Sum pro (5)

[0078] Among them, Sum pro is a direct projection structure, and MAX is a maximum value operation. In addition, in order to eliminate the interference caused by noise points, the projection result value Value is smoothed by moving average:

[0079]

[0080] according to Figure 1 In (e), the projection result is used to obtain the distance between the extreme points, which is the fuzzy scale L.

[0081] (4) Performing an inverse transformation on the original image based on Wiener filtering to obtain a restored clear image.

[0082] Through the above process, the point spread function parameters are obtained: blur angle θ and blur scale L, that is, H(u,v) in formula (2) is obtained. Further, to obtain the restored image, it is only necessary to perform inverse filtering on G(u,v) in formula (2), and F(u,v) refers to the clear image obtained after Wiener filtering. Wiener filtering is an improved inverse filtering, which solves the estimated value of the restored image:

[0083]

[0084] Where H(u,v) is the Fourier transform of the point spread function, H*(u,v) is its conjugate; S η (u,v)=|N(u,v) 2 is the power spectrum of the noise; S f (u,v)=|F(u,v) 2 is the power spectrum of the original image. The final image restoration result is obtained by The inverse Fourier transform of Figure 1 As shown in (f), compared with the original image ( Figure 1 (a)), the clarity of the restored image is significantly improved.

[0085] S3: Using an image enhancement algorithm based on an unsharp mask to perform contrast enhancement on the clear image to obtain a final enhanced image; the image enhancement algorithm based on an unsharp mask specifically includes: using an improved Sobel operator to enhance the high-frequency information of the clear image to obtain a high-frequency enhanced image; using a curve mapping method to enhance the low-frequency information of the clear image to obtain a low-frequency enhanced image; based on an unsharp mask, performing synchronous pixel-level weighted fusion enhancement on the high-frequency enhanced image and the low-frequency enhanced image to obtain a final enhanced image.

[0086] In this embodiment, the clarity of the original image is significantly improved through the above processing, and its contrast needs to be enhanced. The image information can be divided into low-frequency information and high-frequency information, where the high-frequency part contains the detailed information of the image, and the low-frequency part contains the overall shape and grayscale distribution information of the image. Contrast enhancement can be performed separately from the high-frequency and low-frequency information of the image.

[0087] Gradient-based high-frequency enhancement:

[0088] High-frequency image enhancement uses an improved Sobel operator to enhance the gradient weight of the Sobel convolution kernel in the set direction. This embodiment uses an improved Sobel operator to enhance the high-frequency information of the image. The Sobel operator calculates the gradients in the horizontal and vertical directions respectively. Because the character morphology is basically distributed in the 45° and 135° directions in the image, the above directions are set, so the Sobel convolution kernel is improved and optimized:

[0089]

[0090] Among them, G x is the horizontal gradient convolution kernel, G y is the vertical gradient convolution kernel, and the synthetic gradient is:

[0091]

[0092] Compared with the traditional Soble operator, the operator in this embodiment increases the gradient weights in the 45° and 135° directions. The high-frequency enhancement result based on the gradient is as follows: Figure 1 As shown in (g), the details are significantly enhanced.

[0093] Curve-based mapping for low frequency enhancement:

[0094] The low-frequency image enhancement is performed based on a curve mapping method. The goal of the low-frequency part is mainly to adjust and map the overall grayscale of the image. This embodiment uses a curve mapping method to enhance the low-frequency area of ​​the image. The mapping curve expression is:

[0095]

[0096] Among them, γ is the mapping coefficient, which is used to adjust the relationship between the front and back of the mapping. Figure 2 The figure shows the relationship before and after mapping when different γ values ​​are taken. In this embodiment, γ = 2, and the low-frequency enhancement result after mapping is as follows: Figure 1 As shown in (h), the overall contrast of the image is significantly improved.

[0097] In this embodiment, an image enhancement algorithm based on unsharp mask is used. The core of this method is to enhance the high-frequency information and low-frequency information of the image respectively, and then perform pixel-level weighted fusion on the high-frequency enhanced image and the low-frequency enhanced image. The enhancement of high-frequency and low-frequency image information has been obtained in the above steps. The unsharp mask can be expressed as:

[0098]

[0099] Among them, f(x,y) fusion To enhance the final image, is a high-frequency enhanced image of high-frequency detail information, is a low-frequency enhanced image of the overall low-frequency information, and α is a weight parameter, i.e., a fusion weight for adjusting high-frequency and low-frequency pixel information. In this embodiment, α=0.1.

[0100] Enhanced result based on unsharp mask Figure 1 (i) shows that compared with the result of only blur restoration ( Figure 1 (f)), the overall image contrast is significantly enhanced, and the image quality is further improved. Compared with the original image ( Figure 1 (a)), after image restoration and fusion enhancement, the clarity and saliency of the image are significantly improved.

[0101] Compared with the original image, the quantitative statistical results of the image quality improvement after enhancement processing are shown in Table 1 below. It can be seen that after image restoration and fusion enhancement, the standard deviation of the image increased by 14% and the average gradient increased by 46%, indicating that the image details are more significant, thanks to the fuzzy image restoration and high-frequency enhancement processing; and the contrast of the character string compared to the nearby background increased by 155%, which is consistent with the perceived effect. Therefore, after the preprocessing operation of image restoration based on structure forest and high- and low-frequency pixel-level weighted fusion enhancement based on unsharp mask, the image quality, especially the character contrast, has been significantly improved, laying a solid foundation for subsequent character detection and recognition.

[0102] Table 1

[0103] Original Image After enhancement Capacity Enhancement Standard Deviation 1.39 1.59 14％ Average Gradient 12.44 18.17 46％ Character contrast 0.16 0.4 155％

[0104] Embodiment 2

[0105] This embodiment discloses a restoration and enhancement system for low-quality images of characters of nuclear power plant fuel assemblies, including:

[0106] An image acquisition module is configured to: acquire an original image of a fuel assembly character;

[0107] The image restoration module is configured to: apply the edge extraction based on the structure forest to the estimation of the point spread function parameters, restore the original image, and obtain a restored clear image; the image restoration based on the structure forest specifically includes: extracting the edge of the spectrum transform map of the original image based on the structure forest algorithm, estimating the point spread function by using the line detection and projection mapping algorithm; after estimating the point spread function, restore the original image by using the Wiener filter to obtain a restored clear image;

[0108] A contrast enhancement module is configured to: use an image enhancement algorithm based on an unsharp mask to perform contrast enhancement on the clear image to obtain a final enhanced image; the image enhancement algorithm based on an unsharp mask specifically includes: using an improved Sobel operator to enhance the high-frequency information of the clear image to obtain a high-frequency enhanced image; using a curve mapping method to enhance the low-frequency information of the clear image to obtain a low-frequency enhanced image; based on the unsharp mask, synchronously perform pixel-level weighted fusion enhancement on the high-frequency enhanced image and the low-frequency enhanced image to obtain a final enhanced image.

[0109] Embodiment 3

[0110] The purpose of this embodiment is to provide a computing device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method of embodiment 1 when executing the program.

[0111] Embodiment 3

[0112] The purpose of this embodiment is to provide a computer-readable storage medium, a computer-readable storage medium having a computer program stored thereon, and when the program is executed by a processor, the steps of the method of embodiment 1 are performed.

[0113] The steps involved in the apparatus of the above embodiments 3 and 4 correspond to the method embodiment 1, and the specific implementation method can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0114] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0115] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0116] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A method for restoring and enhancing low-quality images of characters on nuclear power plant fuel assemblies, characterized in that: include: Get the original image of the fuel assembly character; The edge extraction based on the structure forest is applied to the estimation of the parameters of the point spread function, and the original image is restored to obtain a restored clear image; The image restoration based on the structure forest specifically includes: extracting edges from the spectrum transform graph of the original image based on the structure forest algorithm, estimating the point spread function using the line detection and projection mapping algorithm; after estimating the point spread function, restoring the original image through the Wiener filter to obtain a restored clear image; An image enhancement algorithm based on an unsharp mask is used to perform contrast enhancement on the clear image to obtain a final enhanced image; the image enhancement algorithm based on an unsharp mask specifically includes: using an improved Sobel operator to enhance the high-frequency information of the clear image to obtain a high-frequency enhanced image; using a curve mapping method to enhance the low-frequency information of the clear image to obtain a low-frequency enhanced image; based on an unsharp mask, the high-frequency enhanced image and the low-frequency enhanced image are synchronously pixel-level weighted fusion enhanced to obtain a final enhanced image.

2. A method for restoring and enhancing low-quality images of characters of nuclear power plant fuel assemblies as claimed in claim 1, characterized in that: Restoring the original image specifically includes: Performing spectrum transformation on the original image to obtain a spectrum transformation graph of the original image; Performing edge extraction on the spectrum transformation graph based on a structured forest algorithm to obtain an edge graph; Performing straight line detection on the edge map by a straight line detection algorithm to estimate the blur angle of the point spread function; rotating the edge map according to the blur angle, and then performing projection transformation to obtain a projection value map, and estimating the blur scale of the point spread function by the projection value map; The original image is inversely transformed based on Wiener filtering to obtain a restored clear image.

3. A method for restoring and enhancing low-quality images of characters of nuclear power plant fuel assemblies as claimed in claim 2, characterized in that: Straight line detection and extraction are performed on the edge image through Radon transformation to obtain a Radon transformation image, and the blur angle is obtained through the feature points in the Radon transformation image.

4. A method for restoring and enhancing low-quality images of characters of nuclear power plant fuel assemblies as claimed in claim 2, characterized in that: After rotating the edge map around the center by a blur angle, a blur scale is estimated based on a projection method.

5. A method for restoring and enhancing low-quality images of characters of nuclear power plant fuel assemblies as claimed in claim 1, characterized in that: The improved Sobel operator is used to improve the gradient weight of the Sobel convolution kernel in a set direction.

6. A method for restoring and enhancing low-quality images of characters of nuclear power plant fuel assemblies as claimed in claim 1, characterized in that: The mapping curve expression in the curve mapping method is: Among them, γ is the mapping coefficient.

7. A method for restoring and enhancing low-quality images of characters of nuclear power plant fuel assemblies as claimed in claim 1, characterized in that: The unsharp mask body is represented as: Among them, f(x,y) fusion To enhance the final image, To enhance the image at high frequencies, is the low frequency enhanced image, and α is the weight parameter.

8. A restoration and enhancement system for low-quality images of characters of nuclear power plant fuel assemblies, characterized in that: include: An image acquisition module is configured to: acquire an original image of a fuel assembly character; An image restoration module is configured to: apply edge extraction based on structure forest to estimate point spread function parameters, restore the original image, and obtain a restored clear image; The image restoration based on the structure forest specifically includes: extracting edges from the spectrum transform graph of the original image based on the structure forest algorithm, estimating the point spread function using the line detection and projection mapping algorithm; after estimating the point spread function, restoring the original image through the Wiener filter to obtain a restored clear image; A contrast enhancement module is configured to: use an image enhancement algorithm based on an unsharp mask to perform contrast enhancement on the clear image to obtain a final enhanced image; the image enhancement algorithm based on an unsharp mask specifically includes: using an improved Sobel operator to enhance the high-frequency information of the clear image to obtain a high-frequency enhanced image; using a curve mapping method to enhance the low-frequency information of the clear image to obtain a low-frequency enhanced image; based on the unsharp mask, synchronously perform pixel-level weighted fusion enhancement on the high-frequency enhanced image and the low-frequency enhanced image to obtain a final enhanced image.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in a method for restoring and enhancing a low-quality image of characters of a nuclear power plant fuel assembly as described in any one of claims 1 to 7 are implemented.

10. A computer 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 in the method for restoring and enhancing low-quality images of characters of nuclear power plant fuel assemblies as described in any one of claims 1-7 are implemented.

Images