Image processing method and device, program product and storage medium

Abstract

According to the image processing method and device provided by the invention, aiming at the characteristics of source intensity fluctuation of an accelerator neutron source, instantaneity and randomness of neutron photography white spot noise and the like, a detection-positioning-replacement mechanism is adopted to process a multi-frame image sequence so as to realize image denoising. The method comprises the following steps: firstly, for each pixel position, judging whether each frame at the pixel position contains noise or not by using a noise reference value obtained based on time domain median filtering, and replacing the gray value of the noise-containing frame with the average value of the gray values of the noise-free frames at the pixel position; fusing the replaced frame sequences and traversing all pixel positions to obtain a primary denoising fusion image; and then, judging whether each fused pixel contains residual noise or not by using a spatial filtering method, and for each pixel containing the residual noise, carrying out gray value replacement of a noisy frame again by using another noise reference value which is not based on time domain filtering and traversing all pixel positions to obtain a secondary de-noised fused image.

Description

Technical Field

[0001] The present invention relates to an image processing method and device capable of removing white spot noise caused by radiation particles or waves, and in particular to a low-noise image processing method and device suitable for neutron photography, as well as a corresponding program product and storage medium, and further to a neutron photography system. Background Art

[0002] Neutron photography is a high-performance non-destructive testing technology that uses the difference in the number of neutrons projected after the interaction between neutrons and samples to obtain information such as the internal structure and defects of the irradiated object. Neutron photography is divided into different types such as cold neutron photography, thermal neutron photography and fast neutron photography according to the different neutron energies used. Among them, thermal neutron photography is widely used in industrial fields such as aerospace, national defense security, and precision manufacturing due to its excellent imaging quality, low cost of use, and high detection sensitivity to specific nuclides.

[0003] In a commonly used neutron radiography system, neutrons from a neutron source penetrate a sample and interact with a conversion screen, converting the irradiated neutrons into, for example, visible light, and a neutron image is obtained by collecting the light using a camera or the like.

[0004] However, neutron images are always affected by various noises, resulting in image degradation. Among these noises, white spot noise is the most prominent. It is mainly caused by the interaction between gamma rays or charged particles and the camera's imaging devices such as CCD chips. Although the camera itself is not directly irradiated by the neutron beam, there are still many radiation particles (such as scattered neutrons and secondary gamma) that may reach the CCD chip and deposit energy, generating prominent random white spot noise in the image. It will lead to a decrease in image resolution and loss of image details, and may affect quantitative analysis and 3D reconstruction. Therefore, the industry often uses image denoising methods to eliminate image noise.

[0005] However, the white spot noise on neutron images is often large in size, high in magnitude, and unevenly distributed, which may make the commonly used denoising algorithms (image domain and frequency domain denoising methods) unsuitable for real neutron images containing a large amount of white spot noise.

[0006] Therefore, how to ensure the imaging resolution while avoiding the influence of white spot noise caused by radiation on the imaging quality of neutron images has become an urgent problem to be solved. Summary of the invention

[0007] The present application aims to solve the above technical problem, namely, the problem of degradation of neutron image quality caused by white spot noise caused by radiation.

[0008] In view of the problems existing in the prior art, the present invention provides an image processing method and device, which can effectively remove the white spot noise caused by radiation during neutron radiography, thereby eliminating the white spot noise of the neutron image while ensuring the imaging resolution, and improving the resolution capability of the imaging system.

[0009] Furthermore, the present invention is not limited to neutron radiography, but can also be applied to removing white spot noise caused by radiation particles or waves in other situations, such as noise from cosmic rays in remote sensing images.

[0010] Specifically, the present invention provides an image processing method, which is applied to a neutron radiography system, and is characterized in that it comprises: an image acquisition step, wherein, for a photography task, multiple frames of detection images with the same imaging time are acquired through multiple exposures to obtain a detection image frame sequence; a denoising step, wherein the detection image frame sequence is subjected to time domain denoising processing and fusion processing according to each pixel position, and in the time domain denoising processing, the mean or median of the pixel grayscale of each frame in the detection image frame sequence at the processed pixel position is used as the pixel frame sequence grayscale reference value, and the pixel grayscale of each frame at the processed pixel position is sequentially subtracted from the pixel frame sequence grayscale. a grayscale reference value, if the result is not greater than the first threshold, the grayscale of the pixel of the frame is retained, if the result is greater than the first threshold, the original grayscale value is replaced by the average value of all the retained pixel grayscales of each frame at the processed pixel position, thereby obtaining a time-domain denoised pixel grayscale sequence of the processed pixel position as a primary denoised pixel grayscale sequence, in the fusion process, calculating the cumulative value of all pixel grayscales of the primary denoised pixel grayscale sequence of the processed pixel position as the grayscale value of the pixel position of the primary denoised fused image, thereby obtaining a primary denoised fused image by traversing all pixel positions; and a secondary denoising step, wherein the primary denoised fused image is denoised by traversing all pixel positions; The combined image is subjected to residual noise judgment processing and denoising and restoration processing according to each pixel position. In the residual noise judgment processing, a spatial pixel grayscale reference value is obtained by calculating a pixel group closest to the processed pixel position through mean filtering or Gaussian filtering, and an absolute value of a difference between the pixel grayscale of the processed pixel position of the primary denoised fused image and the spatial pixel grayscale reference value is calculated. By comparing the absolute value with a second threshold, it is judged whether there is residual noise in the processed pixel position of the primary denoised fused image. In the denoising and restoration processing, if the processed pixel position is judged to have no residual noise, the primary denoised fused image is retained. The pixel grayscale of the pixel position of the denoised fused image is used as the pixel grayscale of the corresponding pixel position of the secondary denoised fused image. If the processed pixel position is judged to have residual noise, the ratio of the spatial pixel grayscale reference value to the number of image frames is used as the new pixel frame sequence grayscale reference value, and the processed pixel position is re-performed with the time domain denoising process to obtain a new time domain denoised pixel grayscale sequence as the secondary denoised pixel grayscale sequence, and the accumulated value of the obtained pixel grayscale sequence is used as the corresponding pixel grayscale to replace the grayscale value of the corresponding pixel position of the primary denoised fused image. Thus, the secondary denoised fused image is obtained by traversing all pixel positions.

[0011] In addition, the present invention provides an image processing device, which is applied to a neutron radiography system, and is characterized in that it includes: an image acquisition module, which, for a photography task, acquires multiple frames of detection images with the same imaging time through multiple exposures to obtain a detection image frame sequence; a primary denoising module, wherein the detection image frame sequence is subjected to time domain denoising processing and fusion processing according to each pixel position, and in the time domain denoising processing, the mean or median of the pixel grayscale of each frame in the detection image frame sequence at the processed pixel position is used as the pixel frame sequence grayscale reference value, and the pixel grayscale of each frame at the processed pixel position is sequentially subtracted from the pixel frame sequence grayscale reference value. , if the result is not greater than the first threshold, the pixel grayscale of the frame is retained; if the result is greater than the first threshold, the original grayscale value is replaced by the average value of all retained pixel grayscales of each frame at the processed pixel position, thereby obtaining a time-domain denoised pixel grayscale sequence of the processed pixel position as a primary denoised pixel grayscale sequence, and in the fusion process, calculating the cumulative value of all pixel grayscales of the primary denoised pixel grayscale sequence of the processed pixel position as the grayscale value of the pixel position of the primary denoised fused image, thereby obtaining a primary denoised fused image by traversing all pixel positions; and a secondary denoising module, wherein the primary denoised fused image is The image is subjected to residual noise judgment processing and denoising and restoration processing according to each pixel position. In the residual noise judgment processing, a spatial pixel grayscale reference value is obtained by calculating a pixel group closest to the processed pixel position through mean filtering or Gaussian filtering, and an absolute value of a difference between the pixel grayscale of the processed pixel position of the primary denoising and fusion image and the spatial pixel grayscale reference value is calculated. By comparing the absolute value with a second threshold, it is judged whether there is residual noise in the processed pixel position of the primary denoising and fusion image. In the denoising and restoration processing, if the processed pixel position is judged to have no residual noise, the primary denoising and fusion image is retained. The pixel grayscale of the pixel position in the de-noised fused image is used as the pixel grayscale of the corresponding pixel position in the secondary denoised fused image. If the processed pixel position is judged to have residual noise, the ratio of the spatial pixel grayscale reference value to the number of image frames is used as the new pixel frame sequence grayscale reference value, and the processed pixel position is re-performed with the time domain denoising process to obtain a new time domain denoised pixel grayscale sequence as the secondary denoised pixel grayscale sequence, and the accumulated value of the obtained pixel grayscale sequence is used as the corresponding pixel grayscale to replace the grayscale value of the corresponding pixel position in the primary denoised fused image. Thus, the secondary denoised fused image is obtained by traversing all pixel positions.

[0012] In addition, the present invention provides a program product, which includes a program, and when the program is executed by a processor, the steps of the above-mentioned image processing method are implemented.

[0013] In addition, the present invention provides a storage medium on which a program is stored, and when the program is executed by a processor, the steps of the above-mentioned image processing method are implemented.

[0014] In addition, the present invention provides a neutron radiography system, comprising: a neutron source, which provides a collimated neutron beam; a conversion screen, which is arranged in front of the neutron emission direction of the neutron source across a sample and converts the irradiated neutrons into visible light; an imaging detector, which detects the visible light converted by the conversion screen in a multiple short-time exposure mode; and the above-mentioned image processing device.

[0015] By using the image processing method and device of the present invention as described above, the detection image frame sequence is subjected to time domain denoising and fusion processing with the pixel grayscale sequence at each pixel position (i.e., the sequence of grayscale values ​​at the corresponding pixel position of each frame) as the object, and a denoised fusion image (equivalent to a first denoising) is obtained by traversing all pixel positions, and then, for the first denoised fusion image, the pixel grayscale of the pixels that do not contain residual noise is retained, and for the pixels that still contain residual noise, the pixel grayscale sequence at its pixel position is changed with a reference value and the time domain denoising and fusion processing is performed again, and a second denoised fusion image (equivalent to a second denoising) is obtained by traversing all pixel positions. Thus, the white spot noise caused by radiation in the neutron radiography process can be effectively removed, thereby eliminating the white spot noise of the neutron image while ensuring the imaging resolution, and improving the resolution capability of the imaging system.

[0016] In addition, the secondary denoising can be repeatedly performed iteratively until there is no residual noise or the number of iterations reaches a preset upper limit, thereby further eliminating noise. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Schematic diagram showing the structure of a neutron radiography system 10 according to the present invention.

[0018] Figure 2 : is a flowchart showing the image processing method according to the present embodiment.

[0019] Figure 3 is with Figure 2 The illustrated image processing method corresponds to a schematic processing procedure.

[0020] Figure 4 It shows the specific process of the normalization step S202.

[0021] Figure 5 It shows the specific process of the denoising step S203.

[0022] Figure 6 It shows the specific process of the secondary denoising step S204.

[0023] Figure 7 It shows the specific process of the time domain repairing step S204b.

[0024] Figure 8 It is a diagram showing the effect of the image processing method according to this embodiment.

[0025] Fig. 9 1 is a functional block diagram showing the image processing device 900 according to the present embodiment.

[0026] Fig.10 This figure shows the hardware configuration when the image processing device is configured by a processor executing a computer program. DETAILED DESCRIPTION

[0027] The specific implementation of the present invention will be described below in conjunction with the accompanying drawings.

[0028] It should be understood that all the drawings are merely representative schematic diagrams, which are used to assist in understanding the technical ideas of the invention and do not limit specific shapes, sizes, and positional relationships.

[0029] In the following embodiments, when numbers of elements (including numbers, values, amounts, ranges, etc.) are mentioned, they are not limited to specific numbers except for cases where they are specifically stated or are obviously limited to specific numbers in principle, and may be above or below the specific numbers.

[0030] In addition, in the following embodiments, the structural elements (including step elements, etc.) are not necessarily essential unless otherwise specified or clearly understood to be essential in principle, and elements not explicitly mentioned in the description may be included.

[0031] The implementation method described in this specification is only an example of a complete description and does not limit the scope of protection of the present invention. Based on the implementation method of the present invention, all other implementation methods that can be obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0032] The image processing method of the present invention can be implemented by a computer executing a program stored in a storage medium, or by implementing the various steps of the method through dedicated hardware. The storage medium can be any readable medium capable of storing data and programs, such as a hard disk, an optical disk, a flash disk, a chip, etc. The program can be any form of computer program, such as binary code, source code, instruction set, etc. The program can be executed by a processor on a computer, a server, a cloud computing device, etc., to implement the image processing method of the present invention.

[0033] [Implementation Method]

[0034] The following embodiments are specifically described for neutron radiography, but the present invention is not limited to neutron radiography, and can be applied to other situations of removing white spot noise caused by radiation particles or waves, etc. However, considering the characteristics of source intensity fluctuation of accelerator neutron source, conventional time domain median filtering cannot remove noise well, so the present invention is particularly suitable for neutron radiography using accelerator neutron source.

[0035] First, the brief structure of the neutron radiography system is described. Figure 1 Schematic diagram showing the structure of a neutron radiography system 10 according to the present invention.

[0036] like Figure 1 As shown, the neutron radiography system 10 includes: a neutron source 1, a sample 2, a neutron conversion screen 3, a reflector 4, an imaging detector 5, a data acquisition and processing unit 6, and a dark box 7. The neutron conversion screen 3, the reflector 4, and the imaging detector 5 are placed in the dark box 7, and the neutron source 1, the sample 2, and the data acquisition and processing unit 6 are placed outside the dark box 7.

[0037] The neutron source 1 is, for example, a radioisotope neutron source, a reactor neutron source, or an accelerator neutron source, and the present invention is not limited thereto. However, an accelerator neutron source is preferred, in which case the denoising effect of the present invention is particularly significant.

[0038] The neutron source 1 emits a collimated neutron beam 8, and the sample 2 and the neutron conversion screen 3 are arranged in front of the neutron beam 8 in the emission direction relative to the neutron source 1. The sample 2 and the neutron conversion screen 3 are arranged perpendicular to the direction of the neutron beam 8. The neutron conversion screen 3 contains neutron conversion material, fluorescent material, etc. The neutron conversion material interacts with the neutrons penetrating the sample 2 to generate α, β, γ rays or recoil material, etc. These charged particles or secondary rays make the fluorescent material emit light.

[0039] The neutron conversion screen 3 is installed in front of the dark box 7 whose cross section is in the shape of an inverted "L". The light emitted by the fluorescent substance travels along the emission direction of the neutron beam 8. The reflector 4 is placed at a certain angle at the corner of the dark box 7 to ensure that the light emitted by the neutron conversion screen 3 can be received by the imaging detector 5. The imaging detector 5 is a device that can convert the received light into image data output, including but not limited to a CCD camera, a CMOS camera, etc.

[0040] The imaging detector 5 transmits the image data to the data acquisition and processing unit 6 via the data line, and the image data is processed in the data acquisition and processing unit 6. Here, the imaging detector 5 can adjust the exposure time of each image (hereinafter referred to as "detection image") output, for example, the exposure mode of the imaging detector 5 can be set by the data acquisition and processing unit 6, and multiple short-time exposure modes (the definition of "short-time" will be described later) are used for imaging.

[0041] Thus, compared with the neutron radiography system commonly used in the prior art, the difference of the neutron radiography system of the present invention can be considered to be only in the data acquisition and processing unit 6, and this difference can be realized by hardware as well as by software. For example, it can be considered that the difference only involves the hardware structure of the data acquisition and processing unit 6, or it can be considered that the difference only involves the program stored and run on the computer in the data acquisition and processing unit 6.

[0042] Therefore, during the operation of the neutron radiography system 10, the neutron source 1 irradiates the neutron beam 8 to the sample 2 to be tested, and an imaging detector 5 (sometimes directly referred to as a CCD camera hereinafter) is used to detect the light generated from the neutron conversion screen 3. Under the control of the data acquisition and processing unit 6, the CCD camera outputs a sequence of detection images at fixed time intervals.

[0043] As mentioned above, the detection image output by the CCD camera contains white spot noise caused by radiation particles. Especially when an accelerator neutron source is used, the existing technology cannot effectively remove the white spot noise.

[0044] In this regard, the image processing method of the present invention can effectively remove the white spot noise and improve the visual effect of the neutron image. In the image processing method, at least two denoising steps are applied in stages to a sequence of multiple detection images obtained from an imaging detector.

[0045] Specifically, in view of the fact that the source intensity of the neutron source of the accelerator fluctuates and the white spot noise of neutron photography is instantaneous and random, a detection-positioning-replacement mechanism is used to perform time domain denoising and fusion processing on the pixel grayscale sequence at each pixel position as the object for the sequence of denoising object images obtained based on the detection image sequence, so as to obtain the denoised pixels at each pixel position, wherein, for the pixel grayscale sequence at a pixel position, a prescribed noise reference value for the pixel position is used to judge whether each frame of the frame sequence contains noise, and the average value of the grayscale values ​​at the pixel position of all frames judged to contain no noise is used to replace the grayscale value at the pixel position of the frame judged to contain noise, and all frames of the replaced frame sequence at the pixel position are fused into one fused pixel. Specifically, first, the first noise reference value based on the adaptive time domain median filter is used to perform time domain denoising and fusion processing according to the frame sequence of each pixel position, thereby obtaining a denoised fused image (corresponding to a denoising step) that eliminates the white spot noise in a large range. On this basis, each pixel of the first denoising and fusion image is identified based on spatial domain filtering to determine whether it is a pixel to be repaired containing residual noise, and the frame sequence at the pixel position of the pixel to be repaired is re-processed with time domain denoising and fusion using a second noise reference value to perform time domain restoration (corresponding to the secondary denoising step).

[0046] Among them, the two time domain denoising processes are aimed at the different noises to be processed (large-scale white spot noise, residual noise), and different standards for judging whether there is noise (noise reference value) are adopted according to their characteristics, thereby effectively removing the white spot noise caused by radiation during neutron radiography.

[0047] Furthermore, the secondary denoising step may be iteratively performed repeatedly until there is no residual noise in the obtained denoised fused image or the number of iterations reaches a preset upper limit, thereby further removing noise.

[0048] The following describes the details of the image processing method of this embodiment with reference to the drawings.

[0049] Figure 2 is a flowchart showing the image processing method of this embodiment. Figure 3 is a description of an exemplary processing procedure corresponding thereto. Figures 4 to 7 express Figure 2 The specific process of the corresponding steps in the image processing method shown.

[0050] like Figure 2 As shown, the image processing method of this embodiment generally includes an image acquisition step S201, a normalization step S202, a primary denoising step S203 and a secondary denoising step S204. Among them, the main processing of the primary denoising step S203 and the secondary denoising step S204 is the above-mentioned time domain denoising processing and fusion processing, but the object pixels processed may be different, and the noise reference values ​​used are different. Therefore, they can also be collectively referred to as "denoising step S205". As described in detail later, the purpose of the normalization step S202 is to eliminate the fluctuation of the image grayscale value caused by the change of the neutron source intensity, which is particularly suitable for the case of using an accelerator neutron source. However, when the source intensity of the neutron source is stable or the source intensity fluctuation is negligible, the normalization step S202 can be omitted.

[0051] (Step S201)

[0052] First, in the image acquisition step S201, a plurality of continuous detection images with the same imaging duration are acquired from the above-mentioned imaging detector 5 to obtain a detection image sequence.

[0053] Specifically, unlike conventional single long exposure imaging, this embodiment sets the exposure time of the imaging detector 5, and uses short-time multiple exposures to obtain multiple equally spaced detection images, each of which is, for example, a grayscale image of M×N pixels. For example, raw data can be obtained through a CCD camera, and the CCD camera can be controlled to output process images at fixed time intervals to obtain a sequence of images with grayscale changes at the same time interval. In this way, multiple equally spaced neutron images can be obtained, thereby achieving more sophisticated control and processing of neutron image data.

[0054] For example, compared to the traditional exposure time of about 3 to 5 minutes per image, in one embodiment, the CCD camera outputs a detection image every 10 seconds (exposure time 10 seconds), while maintaining the same exposure time and aperture size, thereby obtaining a sequence of detection images at the same time interval. The image sequence is output to the data acquisition and processing unit 6 through the data transmission interface of the CCD camera for subsequent processing.

[0055] It should be noted that the "short time" mentioned in this application requires that the thermal noise of the neutron radiography itself is negligible compared with the imaging signal of the detection image. For example, if the grayscale value of the imaging signal is 500, the grayscale value fluctuation caused by other noises (such as thermal noise, dark current noise, readout noise, etc.) in the image is 5, and the fluctuating signal accounts for less than 1% of the imaging signal, then it can be considered that the conditions for short-time imaging are met.

[0056] The reason why such an imaging mode is adopted in the present invention is that, in the case of long exposure, the amount of white spot noise will increase with the increase of imaging time. Assuming that the image noise of one minute of imaging time accounts for 10%, and the white spot noise of a five-minute image accounts for 30%, and the number of imaging times is 5, the probability that the median of the pixel sequence of a one-minute image is noise is 0.9%, and the probability that the median of the pixel sequence of a five-minute image is noise is 16.3%. Such a long exposure mode is not particularly suitable for the image processing described later in the present invention.

[0057] The above-mentioned image acquisition step S201 corresponds to Figure 3 In the process "①", Figure 3 As an example, 5 detection images are obtained, each of which has approximately the same exposure time. As shown in the figure, each detection image has a large amount of white spot noise.

[0058] (Step S202)

[0059] In the normalization step S202, a weight factor based on the source intensity of the neutron source corresponding to each detection image obtained in the image acquisition step S201 is obtained, and each detection image is normalized using the weight factor.

[0060] Here, the weight factor can be calculated according to the specific requirements and methods of the application. For example, it can be calculated based on factors such as the grayscale distribution and contrast of each detection image, or it can be calculated by directly measuring the source intensity or based on the accelerator current, etc. As long as the impact of source intensity fluctuations between the detection images can be eliminated, the present invention is not limited to this.

[0061] In addition, since the present invention can also be applied to fields other than neutron radiography, the "source intensity of the neutron source" herein can also be appropriately expressed as "source intensity of the detection signal source" or the like.

[0062] Figure 4 The specific process of the normalization step S202 is shown, which for example includes the following weight factor obtaining step S202a and normalization processing step S202b.

[0063] (Step S202a)

[0064] In the weight factor obtaining step S202a, each detection image obtained in the image obtaining step S201 is firstly pre-smoothed, and then the average gray value of each pre-smoothed detection image is obtained as the source intensity weight factor.

[0065] The pre-smoothing process may adopt any smoothing process commonly used in the art, or may be omitted, which does not impair the technical effect of the present invention. In one example, the smoothing process may be performed by a 5×5 spatial median filter, but the size of the sliding window is not limited and will not be described in detail here.

[0066] Let the detection image sequence have a total of m images, and the weight factor of the nth image is , can be calculated by the following formula (1).

[0067]

[0068] Among them, M and N are the image size, that is, the number of horizontal and vertical pixels of each detection image. represents the nth detection image, Represents the grayscale value of the (i, j) pixel of the nth detection image after the above pre-smoothing process (spatial median filtering).

[0069] (Step S202b)

[0070] Next, in the normalization processing step S202 b , each detection image (the original image obtained in step S201 ) is normalized using the weight factor obtained in step S202 a .

[0071] The normalization process is calculated using the following formula (2).

[0072]

[0073] in, is the actual gray value of the pixel at pixel position (i, j) in the nth detection image, is the corresponding normalized grayscale value of the pixel, 1≤i≤M, 1≤j≤N.

[0074] In one embodiment, the CCD camera takes an image every 10 seconds, and obtains a sequence of 10 equally spaced neutron images with an exposure time of 10 seconds. Each neutron image is pre-smoothed using a 5×5 spatial median filter to obtain the average grayscale value of each image after filtering. For example, assuming that the average grayscale value after median filtering is 50 for the first image according to formula (1), the weight factor based on the source strength is is 50. Then each detection image is normalized, and the normalized grayscale value of each pixel in each image is calculated using formula (2). For example, for the pixel (i, j) in the first image, if its actual grayscale value is 100, its normalized grayscale value is 100 / 50=2.

[0075] Thus, through the normalization step S202, the multiple detection images acquired in the image acquisition step S201 are each normalized to obtain a sequence consisting of multiple normalized images (normalized frame sequence).

[0076] The above normalization step S202 corresponds to Figure 3 The process “②” in FIG. 2 schematically illustrates the process of calculating the weight factor after the detection image is subjected to median filtering (pre-smoothing) and further normalization.

[0077] (Step S203)

[0078] In the primary denoising step S203, a detection-location-replacement mechanism is used to perform time domain denoising and fusion processing on each pixel position of the normalized frame sequence output by the normalization step S202 to eliminate white spot noise, and a primary denoising fusion image described later is obtained by traversing all pixel positions.

[0079] It should be noted that this embodiment is based on the premise that the normalization step S202 is performed, so the normalized image is used as the denoising object in the first denoising step S203, and the normalized restoration process needs to be performed in the fusion process. However, in the case where the normalization step S202 is omitted, the multiple detection images obtained in the image acquisition step S201 can also be directly used as the denoising object for the first denoising, and the restoration process is of course omitted in this case. That is, in the present invention, the processing object of the first denoising step S203 can be either the detection image itself or the image obtained after the detection image is processed by the above-mentioned normalization step and / or other image processing commonly used in the art.

[0080] In addition, before the primary denoising step S203, pixel alignment can be performed on each image as needed. The alignment process can be performed on each normalized image after the normalization step S202, or can be performed on each acquired image in the image acquisition step S201, and the aligned image is used as the above-mentioned detection image. Of course, the alignment step can also be omitted according to the situation.

[0081] Figure 5 The specific process of a denoising step S203 is represented, for example, including the following first noise reference value calculation step S203a, first pixel noisy frame judgment step S203b, first pixel noisy frame replacement step S203c and first fusion step S203d. In the denoising step S203, the above-mentioned denoising and fusion processing is performed using the first noise reference value at each pixel position of the M×N pixels of the normalized image.

[0082] (Step S203a)

[0083] In the first noise reference value calculation step S203a, for each pixel position of M×N pixels of the normalized image, the median or mean of the grayscale values ​​of each frame in the normalized frame sequence at that position is calculated, and the median or mean is used as the first noise reference value at that pixel position, that is, the grayscale reference value of the pixel frame sequence.

[0084] For example, the first noise reference value at pixel position (i, j) can be expressed as .

[0085] (Step S203b)

[0086] In the first pixel noise frame determination step S203b, for each pixel position, based on the relationship between the normalized grayscale value of each frame of the normalized frame sequence at the pixel position and the first noise reference value at the corresponding pixel position calculated in the first noise reference value calculation step S203a, it is determined whether the frame contains white spot noise at the pixel position. If there is no noise, its pixel grayscale is retained; if there is noise, its grayscale value is replaced in the following steps. For example, it can be determined based on whether the difference between the normalized grayscale value and the first noise reference value at the corresponding pixel position is greater than a specified threshold (first threshold).

[0087] For example, the following formula (3) can be used for determination.

[0088]

[0089] in, It is the first threshold value, and its value is related to the degree of signal fluctuation of the background noise of the CCD camera, and can be appropriately set by those skilled in the art according to actual conditions. is a noisy Boolean value, where a value of 1 indicates that the frame contains noise at pixel position (i, j), and a value of 0 indicates no noise.

[0090] (Step S203c)

[0091] In the first pixel noisy frame replacement step S203c, according to each pixel position, the average value of all retained pixel grayscales at the pixel position is used to replace all noisy pixel grayscales at the pixel position.

[0092] In one embodiment, for example, a sequence of 10 equally spaced neutron images is obtained, which are 320×240 pixels in size. After normalization, the grayscale value of each image is 8 bits. For each pixel position, the median of the corresponding normalized grayscale value in each image (frame) is calculated as the corresponding first noise reference value. Then, for each pixel position, the corresponding normalized grayscale value of each frame is compared to see whether it is higher than the first noise reference value (for simplicity, it is assumed that is 0). If it is higher, z(i,j) of the pixel position in the corresponding frame is set to 1, otherwise it is set to 0. Then use time-domain adaptive median filtering to eliminate white spot noise. Specifically, for each pixel position (i,j), the value of z(i,j) of each frame is detected. If z(i,j)=0, it is considered that there is no white spot noise and no replacement is required; if z(i,j)=1, it is considered that there is white spot noise and needs to be replaced. In the case of being judged as noisy, the average grayscale value of all retained pixels at the pixel position is used to replace the pixel grayscale of the noisy frame. For example, if the 2nd, 4th, and 7th frames at (i,j) contain white spot noise, the average grayscale value of the normalized grayscale values ​​of the 1st, 3rd, 5th, 6th, 8th, 9th, and 10th frames at (i,j) can be used to replace the normalized grayscale values ​​of the 2nd, 4th, and 7th frames.

[0093] (Step S203d)

[0094] In the first fusion step S204, at each pixel position, the corresponding pixel grayscale sequence after the first pixel noisy frame replacement step S203c is fused (accumulated value is calculated). For example, weighted fusion is performed using the weight factor based on the source intensity calculated in the normalization step S202 (equivalent to accumulation after restoration processing). Specifically, at each pixel position, the grayscale value of the nth frame after the first pixel noisy frame replacement step S203c is multiplied by the corresponding weight factor Then, the image is superimposed from n=1 to m, and a denoised fused image is obtained by traversing all pixel positions. Of course, if the detection image is not normalized, no restoration processing is required during fusion.

[0095] As described in the normalization step S202, the weight factor can be calculated according to the specific requirements and methods of the application. For example, it can be calculated based on factors such as the grayscale distribution and contrast of each detection image, or it can be calculated by directly measuring the source intensity or based on an external detector, accelerator current, etc. As long as the influence of source intensity fluctuations between the detection images can be eliminated, the present invention is not limited to this.

[0096] In addition, the superposition here can be a simple weighted average or other complex image fusion algorithms, such as a fusion algorithm based on wavelet transform, optical flow estimation, etc., and the present invention is not limited to this.

[0097] The above-mentioned first denoising step S203 corresponds to Figure 3 In the process "③" and process "④", for each pixel position, the first noise reference value at each position is used to perform time domain denoising and fusion processing. Figure 3 As shown in the figure, white spot noise exists in each normalized image. If a certain pixel position of a frame is included in the noise, the frame contains noise at the pixel position and is marked with "×". Otherwise, it is marked with "·". For each pixel position, the frames passing through the dotted line at the position are combined based on the grayscale median and The sum is classified into noisy and noise-free. The grayscale of the noise-free pixel is averaged, and the grayscale of each noisy pixel is replaced with the average value. Then, all frames at the same pixel position after replacement are fused by restoring the grayscale value and superimposing. By traversing all pixel positions, the above-mentioned denoised fused image is obtained.

[0098] (Step S204)

[0099] After the first denoising, noise may still remain in the image. Therefore, the present embodiment performs a second denoising on this basis.

[0100] In the secondary denoising step S204, a second denoising is applied to the primary denoising fused image outputted from the primary denoising step S203 to obtain a secondary denoising fused image.

[0101] As described in detail below, during the second denoising, each pixel of the first denoised and fused image is identified as a pixel to be repaired that contains residual noise that could not be completely removed by the first denoising. The normalized frame sequence at the pixel position of each identified pixel to be repaired is taken as the object, and the second noise reference value that is not based on time domain filtering is used instead of the first noise reference value to perform time domain denoising and fusion processing again, thereby repairing the pixel to be repaired.

[0102] Figure 6 The specific process of the secondary denoising step S204 is shown, for example, including the following residual noise identification step S204a and time domain restoration step S204b.

[0103] (Step S204a)

[0104] In the residual noise identification step S204a, for the first denoised fusion image outputted from the first fusion step S203d, in order to determine whether it contains residual noise that cannot be completely removed by the first denoising, the residual noise identification value (i.e., spatial pixel grayscale reference value) for each pixel position is calculated by spatial adaptive mean filtering of a specified size, and whether the pixel is a pixel containing residual noise, i.e., a pixel to be repaired, is identified based on the relationship between the grayscale value of the pixel at each position and the corresponding residual noise identification value. As mentioned above, Gaussian filtering can also be used for residual noise identification, but the description is omitted here.

[0105] For example, for each pixel position, the mean value of the spatially filtered pixels around it (within the filter size) is used as the residual noise discrimination value for noise discrimination, which can be performed using the following formula (4).

[0106]

[0107] in, is the gray value of the pixel at position (i, j) of the denoised fused image. And a and b are not 0 at the same time, and 2f+1 is the filter size. =1 indicates that the pixel is to be repaired. =0 means that the pixel does not need to be repaired. is the second threshold, which is consistent with the above This is also a pre-set value. This is because, although the corresponding mean result is obtained through spatial mean filtering, in some cases (possibly due to fluctuations or specific imaging conditions), the actual pixel data obtained is slightly larger than the mean result, so it can be used Correct this situation. The same can be appropriately set by those skilled in the art according to actual conditions.

[0108] (Step S204b)

[0109] In the time domain restoration step S204b, for the pixels to be restored identified by the residual noise identification step S204a in the primary denoised and fused image, the normalized frame sequence at each pixel position is re-performed with time domain denoising and fusion processing according to their pixel positions to replace the pixels to be restored, and for the pixel positions other than the pixels to be restored, the pixel grayscale of the pixel position of the primary denoised and fused image is retained, thereby obtaining a secondary denoised and fused image.

[0110] In the temporal denoising process here, the noise reference value used is a second noise reference value (i.e., a new pixel frame sequence grayscale reference value) different from the first noise reference value, and the second noise reference value at each pixel position is obtained based on the above-mentioned residual noise discrimination value at the pixel position. More specifically, the second noise reference value is obtained by assigning the residual noise discrimination value to each normalized image.

[0111] Figure 7 The specific process of the time domain restoration step S204b is represented, for example, including the following second noise reference value calculation step S204b1, second pixel noisy frame judgment step S204b2, second pixel noisy frame replacement step S204b3 and second fusion step S204b4. In the time domain restoration step S204b, the second noise reference value is used to perform the above-mentioned time domain denoising and fusion processing on the pixel position of the pixel to be restored in the M×N pixels of the denoised and fused image.

[0112] These steps are respectively Figure 5 The first noise reference value calculation step S203a, the first pixel noisy frame determination step S203b, the first pixel noisy frame replacement step S203c and the first fusion step S203d shown correspond to each other, so only a brief description is given here.

[0113] (Step S204b1)

[0114] In the second noise reference value calculation step S204b1, a second noise reference value for each pixel position to be repaired is calculated according to the pixel position of the pixel.

[0115] The second noise reference value is different from the first noise reference value in that it is not obtained based on time domain filtering but based on spatial domain mean filtering or Gaussian filtering. Specifically, it is obtained based on the above-mentioned residual noise discrimination value for the pixel position.

[0116] As described above, the residual noise discrimination value is used to judge the residual noise of each pixel in the first denoised fused image, and the second noise reference value is used to re-judge whether each frame of the normalized image (multiple images before fusion) contains noise at the position of each pixel judged to have residual noise. Therefore, the residual noise discrimination value can be allocated to a single normalized image based on the weight factor of each normalized image to obtain the second noise reference value. For example, it can be obtained by dividing the residual noise discrimination value by the sum of each weight factor and then calculating the ratio with the number of image frames. Of course, if the detection image is not normalized, the ratio of the residual noise discrimination value to the number of image frames can be directly used.

[0117] (Step S204b2)

[0118] In the second pixel noisy frame judgment step S204b2, according to each pixel position of the pixel to be repaired, based on the relationship between the normalized grayscale value of each frame in the normalized frame sequence at this position and the second noise reference value, it is re-judged whether the frame contains white spot noise at this pixel position; if there is no noise, its pixel grayscale is retained; if there is noise, its grayscale value is replaced in the subsequent steps.

[0119] For example, the difference between the normalized gray value at a certain pixel position of a certain frame and the second noise reference value at the corresponding pixel position may be greater than a specified threshold. (Third threshold) to judge. Similarly, it can be appropriately set by those skilled in the art according to actual conditions, or the first threshold can be directly used.

[0120] (Step S204b3)

[0121] In the second pixel noisy frame replacement step S204b3, according to each pixel position of the pixel to be repaired, the average value of all retained pixel grayscales at the pixel position is used to replace all noisy pixel grayscales at the pixel position.

[0122] (Step S204b4)

[0123] In the second fusion step S204b4, according to each pixel position of the pixel to be repaired, the corresponding pixel grayscale sequence replaced in the second pixel noisy frame replacement step S204b3 is fused (accumulated value is calculated). For example, the weighted fusion is performed using the source intensity-based weight factor calculated in the normalization step S202 (equivalent to accumulation after restoration processing). Specifically, according to each pixel position of the pixel to be repaired, the grayscale value of the nth frame is multiplied by the corresponding weight factor Then, the images are superimposed from n = 1 to m. Of course, if the detection image is not normalized, no restoration processing is required during fusion.

[0124] Therefore, the pixels to be repaired in the primary denoised fused image are replaced, and the remaining pixels are retained, and the obtained image is the secondary denoised fused image.

[0125] The above-mentioned secondary denoising step S204 corresponds to Figure 3 The process "⑤" in Figure 3 As shown in the first denoised fused image, there is residual single-point white spot noise. Therefore, for each pixel position, the second noise reference value at each position (different from the first noise reference value, which is not obtained based on time domain filtering) is used to perform time domain denoising and fusion processing again, and the processing is traversed according to the pixel position of each pixel to be repaired. Thus, the second denoising fused image is obtained through the second denoising.

[0126] Furthermore, the secondary denoising step can be performed iteratively. For example, the secondary denoising step S204 can be performed again for the secondary denoising fusion image, wherein all pixel positions are traversed again, and a new residual noise discrimination value obtained based on the secondary denoising fusion image is used to determine whether each pixel position still has residual noise. If it is determined that there is no residual noise, the pixel grayscale is retained. If it is determined that there is still residual noise, a new second noise reference value is obtained based on the new residual noise discrimination value, and the repair step of step S204b is performed again.

[0127] Here, an upper limit may be preset for the number of iterations, or no upper limit may be set, and iterations are performed continuously until no residual noise is left in the output denoised fused image.

[0128] At this point, the image processing method of this embodiment ends, and the secondary denoising fusion image output after at least two denoising steps is the final neutron imaging image. Of course, the neutron imaging image can also be further subjected to other image processing known in the art as needed, which will not be described in detail here.

[0129] The image processing method of this embodiment is described above, which has the following effects:

[0130] (1) The imaging detector adopts an imaging method based on short-time multiple exposures to obtain the relationship between imaging data and imaging time while ensuring the accuracy of the neutron radiography detection image, which provides a basis for the denoising described later.

[0131] (2) In view of the instantaneous, random and source intensity fluctuation characteristics of white spot noise in neutron photography, especially accelerator neutron source neutron photography, the time domain adaptive median denoising of the source intensity normalized image sequence is adopted. Compared with the traditional time domain method, it can effectively solve the gray value fluctuation of each detection image caused by the change of neutron source intensity and has a good effect under the imaging condition of unstable source intensity.

[0132] (3) According to the imaging characteristics of neutron radiography, a secondary denoising method based on spatial domain discrimination and temporal domain data restoration is used to remove the residual noise after the first denoising. In this way, a high degree of image restoration can be achieved while removing the residual noise.

[0133] It can be understood that in the present invention, by performing primary denoising and secondary denoising, at least two time domain denoising processes (and fusion processes) are selectively performed. The difference between them is that the reference value used for the first time is based on the time domain median filtering; and the reference value used for the second time is based on the spatial domain mean filtering or Gaussian filtering.

[0134] The reason for this distinction is due to the imaging characteristics of neutron photography. One of the core ideas of the first denoising is to identify the noise within the pixel based on the low proportion of noise. Since the grayscale value of white spot noise is much larger than the grayscale value of conventional imaging, the result calculated by the mean filter exceeds the actual grayscale value by a lot, and this value cannot be used as a reference. Therefore, only the median filter can be used here. The detection of residual noise is based on the imaging characteristics of neutron photography-the signal will cause loss after passing through the system. Generally, this loss can be simulated by Gaussian filtering or mean filtering. Therefore, these two methods can be used to detect whether the pixel still contains residual noise through the grayscale value of the pixels around the pixel.

[0135] Figure 8 It is a diagram showing the effect of the image processing method according to this embodiment. Figure 8 (a) shows the neutron image obtained by directly synthesizing the detection image output by the imaging detector when using an accelerator neutron source for neutron radiography. Figure 8 (b) shows a neutron image obtained by using the image processing method of this embodiment.

[0136] according to Figure 8 It can be seen that the image processing method of this embodiment can effectively eliminate the white spot noise in the neutron image.

[0137] Refer to the following Fig. 9 , Fig.10 The image processing device of this embodiment is described below. The image processing device of this embodiment can be configured as the data acquisition and processing unit 6 in the above-mentioned neutron radiography system, and can be implemented by hardware or software.

[0138] Fig. 9 1 is a functional block diagram showing the image processing device 900 according to the present embodiment.

[0139] like Fig. 9 As shown, the image processing device 900 includes an image acquisition module 901, a normalization module 902, a primary denoising module 903 and a secondary denoising module 904 corresponding to each step of the image processing method. Similarly, the primary denoising module 903 and the secondary denoising module can be collectively referred to as a denoising module 905. In addition, the normalization module 902 can also be omitted, and the processing in the first and second fusion modules and the normalization module are omitted and the restoration processing is correspondingly omitted.

[0140] The following briefly describes each module. Of course, the above description of each step can also be directly applied to the following description of the features of each module.

[0141] The image acquisition module 901 is connected to the imaging detector 5 to acquire a plurality of continuous detection images with the same imaging duration to obtain a detection image sequence.

[0142] The normalization module 902 obtains a weight factor based on the source intensity of the neutron source corresponding to each detection image obtained by the image acquisition module 901, and performs normalization processing on each detection image using the weight factor.

[0143] The normalization module 902 specifically includes a weight factor obtaining module 902a and a normalization processing module 902b.

[0144] In the weight factor obtaining module 902a, each detection image obtained by the image obtaining module 901 is firstly pre-smoothed, and then the average gray value of each pre-smoothed detection image is obtained as the source intensity weight factor.

[0145] In the normalization processing module 902b, each detection image is normalized using the weight factor obtained by the weight factor obtaining module 902a.

[0146] The primary denoising module 903 performs time domain denoising and fusion processing on the normalized frame sequence output by the normalization module 902 at each pixel position to eliminate white spot noise, and obtains a primary denoising fusion image by traversing all pixel positions.

[0147] The primary denoising module 903 specifically includes a first noise reference value calculation module 903a, a first pixel noisy frame determination module 903b, a first pixel noisy frame replacement module 903c and a first fusion module 903d.

[0148] In the first noise reference value calculation module 903a, for each pixel position of M×N pixels of the normalized image, the median or mean of the grayscale values ​​of each frame in the normalized frame sequence at that position is calculated, and the median or mean is used as the first noise reference value at that pixel position, that is, the grayscale reference value of the pixel frame sequence.

[0149] In the first pixel noisy frame judgment module 903b, for each pixel position, based on the relationship between the normalized grayscale value of each frame in the normalized frame sequence at the pixel position and the first noise reference value at the corresponding pixel position calculated by the first noise reference value calculation module 903a, it is judged whether the frame contains white spot noise at the pixel position. If there is no noise, its pixel grayscale is retained; if there is noise, its grayscale value is replaced in the module described later.

[0150] In the first pixel noisy frame replacement module 903c, according to each pixel position, the average value of all retained pixel grayscales at the pixel position is used to replace all noisy pixel grayscales at the pixel position.

[0151] In the first fusion module 903d, the corresponding pixel grayscale sequence replaced by the first pixel noisy frame replacement module 903c is fused (accumulated value is calculated) at each pixel position. For example, weighted fusion is performed using the weight factor based on the source intensity calculated in the normalization module 902 (equivalent to accumulation after restoration), and a denoised fused image is obtained by traversing all pixel positions. Of course, if the normalization module is omitted, there is no need to perform restoration processing here.

[0152] In the secondary denoising module 904, the secondary denoising is applied to the primary denoising fused image output by the primary denoising module 903 to obtain a secondary denoising fused image. Specifically, for each pixel of the primary denoising fused image, it is identified whether it is a pixel to be repaired that contains residual noise that cannot be completely removed by the first denoising, and the normalized frame sequence at each pixel position of the identified pixel to be repaired is used as the object, and the second noise reference value that is not based on time domain filtering is used instead of the first noise reference value to perform time domain denoising and fusion processing again, thereby repairing the pixel to be repaired.

[0153] The secondary denoising module 904 specifically includes a residual noise identification module 904a and a time domain restoration module 904b.

[0154] In the residual noise identification module 904a, for the denoised fused image output by the first fusion module 903d, in order to determine whether it contains residual noise that cannot be completely removed by the first denoising, the residual noise identification value (i.e., spatial pixel grayscale reference value) for each pixel position is calculated by spatial adaptive mean filtering or Gaussian filtering of a specified size. Based on the relationship between the grayscale value of the pixel at each position and the corresponding residual noise identification value, it is determined whether the pixel is a pixel containing residual noise, i.e., a pixel to be repaired.

[0155] In the time domain restoration module 904b, for the pixels to be repaired identified by the residual noise identification module 904a in the primary denoised and fused image, the normalized frame sequence at each pixel position is re-processed in the time domain for denoising and fusion according to their pixel positions to replace the pixels to be repaired, and for the pixel positions other than the pixels to be repaired, the pixel grayscale of the pixel position of the primary denoised and fused image is retained, thereby obtaining a secondary denoised and fused image.

[0156] The time domain restoration module 904b includes the following second noise reference value calculation module 904b1, second pixel noisy frame judgment module 904b2, second pixel noisy frame replacement module 904b3 and second fusion module 904b4.

[0157] In the second noise reference value calculation module 904b1, the second noise reference value for each pixel position to be repaired is calculated according to the pixel position of the pixel. Specifically, it can be obtained by dividing the residual noise discrimination value by the sum of the weight factors and then calculating the ratio with the number of image frames. Of course, if the detection image is not normalized, the ratio of the residual noise discrimination value to the number of image frames can be directly used.

[0158] In the second pixel noisy frame judgment module 904b2, according to each pixel position of the pixel to be repaired, based on the relationship between the normalized grayscale value of each frame in the normalized frame sequence at this position and the second noise reference value, it is re-judged whether the frame contains white spot noise at this pixel position; if there is no noise, its pixel grayscale is retained; if there is noise, its grayscale value is replaced in the subsequent steps.

[0159] In the second pixel noisy frame replacement module 904b3, according to each pixel position of the pixel to be repaired, the average value of all retained pixel grayscales at the pixel position is used to replace all noisy pixel grayscales at the pixel position.

[0160] In the second fusion module 904b4, the corresponding pixel grayscale sequences replaced by the second pixel noisy frame replacement module 904b3 are fused according to the pixel positions of the pixels to be repaired. Of course, if the detection image is not normalized, no restoration processing is required during fusion.

[0161] Therefore, the pixels to be repaired in the primary denoised fused image are replaced, and the remaining pixels are retained, and the obtained image is the secondary denoised fused image.

[0162] The secondary denoising fusion image is input into the iteration module, and the iteration module inputs the secondary denoising fusion image back into the secondary denoising module. Then, all pixel positions are traversed again, and a new residual noise discrimination value based on the secondary denoising fusion image is used to judge whether there is still residual noise at each pixel position. If it is judged that there is no residual noise, the pixel grayscale is retained. If it is judged that there is still residual noise, a new second noise reference value is obtained based on the new residual noise discrimination value, and the time domain restoration module 904b performs restoration again.

[0163] Here, an upper limit may be preset for the number of iterations, or no upper limit may be set, and iterations are performed continuously until no residual noise is left in the output denoised fused image.

[0164] Fig.10 This figure shows the hardware configuration when the image processing device is configured by a processor executing a computer program.

[0165] like Fig.10As shown, the hardware structure of the image processing device may include at least one processor 501 and a memory such as RAM 503 and ROM 504 that is connected to the processor 501 through a bus 502. ROM 504 may include various storage media, such as a hard disk, an optical disk, etc. ROM 504 stores a program 505 that can be executed by the processor, and the processor executes the program 505 to implement the above-mentioned image processing method. The final neutron image after image processing and various thresholds, setting values, etc. used in the above-mentioned image processing method are stored in ROM 504, and ROM 504 also stores the original detection image input from the imaging detector 5. Various data generated in the intermediate process of executing the program 505 (i.e., during the execution of the image processing method) such as the above-mentioned various noise reference values, frame sequences, fused images, etc. can be temporarily stored in RAM 503, or appropriately stored in ROM 504.

[0166] The input unit 506 is used to input various parameters required for the execution of the program 505, such as set values, thresholds, and set values ​​for the imaging detector 5, such as exposure time, etc. The input unit 506 may include various input devices, such as a keyboard, a mouse, a touch screen, etc. The output unit 507 may be used to output the results of image processing, including various output devices, such as a display, a speaker, etc. The communication unit 508 performs, for example, two-way communication with the imaging detector 5, such as exchanging data via the Internet or other telecommunication networks.

[0167] Therefore, by making the CPU 501 execute the program 505 stored in the ROM 504, the detection image data is obtained from the imaging detector 5 via the communication unit 508, and the above-mentioned image processing method is applied thereto to obtain a neutron image with white spot noise effectively eliminated.

[0168] The present invention can be applied to the field of neutron photography, but is not limited thereto. For example, the present invention can also be applied to the field of remote sensing photography, and can effectively remove white spot noise caused by radiation particles or waves.

Claims

1. An image processing method, applied to a neutron radiography system, characterized in that: include: An image acquisition step, wherein, for a photography task, multiple detection image frames with the same imaging duration are acquired through multiple exposures to obtain a detection image frame sequence; a primary denoising step, wherein the detection image frame sequence is subjected to time domain denoising and fusion processing at each pixel position, wherein in the time domain denoising processing, the mean or median of the pixel grayscale of each frame in the detection image frame sequence at the processed pixel position is used as the pixel frame sequence grayscale reference value, and the pixel grayscale of each frame at the processed pixel position is sequentially subtracted from the pixel frame sequence grayscale reference value, and if the result is not greater than a first threshold, the pixel grayscale of the frame is retained, and if the result is greater than the first threshold, the original grayscale value is replaced by the average value of all retained pixel grayscales of each frame at the processed pixel position, thereby obtaining a time domain denoised pixel grayscale sequence at the processed pixel position as a primary denoised pixel grayscale sequence, and in the fusion processing, the cumulative value of all pixel grayscales of the primary denoised pixel grayscale sequence at the processed pixel position is calculated as the grayscale value of the pixel position of the primary denoised fused image, thereby obtaining a primary denoised fused image by traversing all pixel positions; and The secondary denoising step includes performing residual noise determination processing and denoising and repair processing on the first denoising fusion image at each pixel position. In the residual noise determination processing, a spatial pixel grayscale reference value is obtained by calculating a mean filter or a Gaussian filter for a pixel group closest to the processed pixel position. The absolute value of the difference between the pixel grayscale of the processed pixel position of the first denoising fusion image and the spatial pixel grayscale reference value is calculated. By comparing the absolute value with a second threshold, it is determined whether there is residual noise in the first denoising fusion image at the processed pixel position. In the denoising and repair processing, if the processed pixel position is determined to have no residual noise, the pixel grayscale reference value is calculated. If there is residual noise, the pixel grayscale of the pixel position of the first denoised fused image is retained as the pixel grayscale of the corresponding pixel position of the second denoised fused image. If the processed pixel position is judged to have residual noise, the ratio of the spatial pixel grayscale reference value to the number of image frames is used as the new pixel frame sequence grayscale reference value, the processed pixel position is re-performed with the time domain denoising process to obtain a new time domain denoised pixel grayscale sequence as the second denoised pixel grayscale sequence, and the accumulated value of the obtained pixel grayscale sequence is used as the corresponding pixel grayscale to replace the grayscale value of the corresponding pixel position of the first denoised fused image. Thus, the second denoised fused image is obtained by traversing all pixel positions.

2. The image processing method according to claim 1, characterized in that: Also includes: an iterative step, wherein all pixel positions of the secondary denoising fusion image are traversed, a new spatial pixel grayscale reference value is calculated using a pixel group closest to the processed pixel position, and the residual noise judgment process is re-executed; if residual noise is still found during the traversal process, the residual noise pixel position of the secondary denoising fusion image is treated as a new pixel frame sequence grayscale reference value by using the ratio of the new spatial pixel grayscale reference value to the number of image frames, and the time domain denoising process is re-executed to obtain a new secondary denoising pixel grayscale sequence, and the accumulated value of the newly obtained pixel grayscale sequence is used as the corresponding pixel grayscale to replace the pixel grayscale of the residual noise, thereby obtaining a new secondary denoising fusion image. The iterative steps are repeated until the obtained secondary denoising fusion image has no residual noise or the number of iterations reaches a preset upper limit, and the last obtained secondary denoising fusion image is used as the final denoising fusion image.

3. The image processing method according to claim 1 or 2, characterized in that: Also includes: a normalization step, wherein after acquiring the detection image frame sequence, according to each detection image frame or according to the source intensity signal size of the external detector within the imaging time of each detection image frame, a source intensity weight factor based on the detection signal source corresponding to the detection image frame is obtained, and the detection image frame sequence is normalized using the source intensity weight factor to obtain a normalized detection image frame sequence, Furthermore, in the first denoising step, the normalized detection image frame sequence is processed instead of the detection image frame sequence, and in the fusion process, when calculating the accumulated value of the pixel grayscale, all pixel grayscales are restored according to the source intensity weight factor of the corresponding frame image and then accumulated. In the denoising and restoration processing of the secondary denoising step, the spatial pixel grayscale reference value of the processed pixel position is divided by the sum of the weight factors, and then the ratio thereof to the number of image frames is taken as the new pixel frame sequence grayscale reference value, and when calculating the accumulated value of the pixel grayscale, all pixel grayscales are restored according to the source strength weight factor of the corresponding frame image and then accumulated.

4. An image processing device, applied to a neutron radiography system, characterized in that: include: An image acquisition module, for a photography task, acquires multiple detection image frames of the same imaging duration through multiple exposures to obtain a detection image frame sequence; A primary denoising module, wherein the detection image frame sequence is subjected to time domain denoising and fusion processing according to each pixel position. In the time domain denoising processing, the mean or median of the pixel grayscale of each frame in the detection image frame sequence at the processed pixel position is used as the pixel frame sequence grayscale reference value, and the pixel grayscale of each frame at the processed pixel position is sequentially subtracted from the pixel frame sequence grayscale reference value. If the result is not greater than a first threshold, the pixel grayscale of the frame is retained. If the result is greater than the first threshold, the original grayscale value is replaced by the average value of all retained pixel grayscales of each frame at the processed pixel position, thereby obtaining a time domain denoised pixel grayscale sequence at the processed pixel position as a primary denoised pixel grayscale sequence. In the fusion processing, the accumulated value of all pixel grayscales of the primary denoised pixel grayscale sequence at the processed pixel position is calculated as the grayscale value of the pixel position of the primary denoised fused image, thereby obtaining a primary denoised fused image by traversing all pixel positions; and A secondary denoising module, wherein the residual noise judgment processing and denoising and repair processing are performed on the first denoising and fusion image according to each pixel position. In the residual noise judgment processing, a spatial pixel grayscale reference value is obtained by calculating the distance between the pixel group closest to the processed pixel position through mean filtering or Gaussian filtering, and the absolute value of the difference between the pixel grayscale of the processed pixel position of the first denoising and fusion image and the spatial pixel grayscale reference value is calculated. By comparing the absolute value with the second threshold, it is judged whether there is residual noise in the first denoising and fusion image at the processed pixel position. In the denoising and repair processing, if the processed pixel position is judged to have no residual noise If there is residual noise, the pixel grayscale of the pixel position of the first denoised fused image is retained as the pixel grayscale of the corresponding pixel position of the second denoised fused image. If the processed pixel position is judged to have residual noise, the ratio of the spatial pixel grayscale reference value to the number of image frames is used as the new pixel frame sequence grayscale reference value, the processed pixel position is re-performed with the time domain denoising process to obtain a new time domain denoised pixel grayscale sequence as the second denoised pixel grayscale sequence, and the accumulated value of the obtained pixel grayscale sequence is used as the corresponding pixel grayscale to replace the grayscale value of the corresponding pixel position of the first denoised fused image. Thus, the second denoised fused image is obtained by traversing all pixel positions.

5. The image processing device according to claim 4, characterized in that Also includes: an iterative module, which traverses to obtain all pixel positions of the secondary denoising fusion image, uses the pixel group closest to the processed pixel position to calculate a new spatial pixel grayscale reference value, and re-executes the residual noise judgment process. If it is found that there is still residual noise during the traversal process, the residual noise pixel position of the secondary denoising fusion image is used. The ratio of the new spatial pixel grayscale reference value to the image frame number is used as a new pixel frame sequence grayscale reference value, and the time domain denoising process is re-performed to obtain a new secondary denoising pixel grayscale sequence, and the accumulated value of the newly obtained pixel grayscale sequence is used as the corresponding pixel grayscale to replace the pixel grayscale of the residual noise, thereby obtaining a new secondary denoising fusion image. The obtained new secondary denoising fusion image is input into the iteration module until the obtained secondary denoising fusion image has no residual noise or the number of iterations reaches a preset upper limit, and the last obtained secondary denoising fusion image is used as the final denoising fusion image.

6. The image processing apparatus according to claim 4 or 5, further comprising: A normalization module, for the detection image frame sequence obtained by the image acquisition module, obtains a source strength weight factor based on the detection signal source corresponding to each detection image frame or according to the source strength signal size of the external detector within the imaging time of each detection image frame, and uses the source strength weight factor to normalize the detection image frame sequence to obtain a normalized detection image frame sequence, Furthermore, in the primary denoising module, the normalized detection image frame sequence is processed instead of the detection image frame sequence, and in the fusion process, when calculating the accumulated value of the pixel grayscale, all pixel grayscales are restored according to the source intensity weight factor of the corresponding frame image and then accumulated. In the denoising and restoration processing of the secondary denoising module, the spatial pixel grayscale reference value of the processed pixel position is divided by the sum of the weight factors, and then the ratio thereof to the number of image frames is taken as the new pixel frame sequence grayscale reference value, and when calculating the accumulated value of the pixel grayscale, all pixel grayscales are restored according to the source strength weight factor of the corresponding frame image and then accumulated.

7. A program product comprising a program, characterized in that When the program is executed by a processor, the image processing method according to any one of claims 1 to 3 is implemented.

8. A storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the image processing method according to any one of claims 1 to 3 is implemented.

9. A neutron radiography system, characterized in that: include: a neutron source providing a collimated outgoing neutron beam; A conversion screen is arranged in front of the neutron emission direction of the neutron source across the sample to convert the irradiated neutrons into visible light; an imaging detector for detecting the visible light converted by the conversion screen in a multiple short-time exposure mode; and An image processing device as claimed in any one of claims 4 to 6.

Images