Visual field and radiation field combined nuclear environment monitoring system and method
Through the combination of CMOS sensors and FPGA modules, synchronous monitoring of visual fields and radiation fields is achieved, solving the problems of high cost, severe noise and lack of real-time performance in traditional methods, and providing efficient and synchronous nuclear leakage monitoring capabilities.
Patent Information
- Application Number
- CN202510621845.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-14
- Publication Date
- 2025-09-12
AI Technical Summary
Existing technologies make it difficult to simultaneously acquire images and radiation distribution at a nuclear leakage site under conditions of low cost, high real-time performance, and strong environmental adaptability. Traditional gamma cameras are expensive, susceptible to environmental interference, and produce severe image noise. Deep learning models are computationally complex and difficult to meet real-time requirements.
A CMOS sensor is combined with an FPGA module. Through the noise suppression submodule and the wavelet transform submodule, parallel suppression of image noise and synchronous calculation of radiation dose rate are achieved. Haar wavelet decomposition is used to extract radiation noise characteristics, and the radiation dose rate is calculated in combination with a priori functions to achieve synchronous monitoring of the visual field and radiation field.
It achieves low-cost, high-efficiency simultaneous acquisition of visual images and radiation information, reduces hardware costs and deployment complexity, ensures the temporal and spatial synchronization of data, and meets the rapid response requirements of nuclear leakage emergency scenarios.
Smart Images
Figure CN120630284A_ABST
Abstract
Description
[0001] Technology Neighborhood
[0002] The present invention relates to the technical field of nuclear radiation monitoring, and in particular to a nuclear environment monitoring system and method combining a visual field with a radiation field. Background Art
[0003] With the widespread adoption of nuclear energy technology, the demand for nuclear leak monitoring is increasing. Traditional nuclear leak monitoring methods rely on non-visual sensor data, such as temperature, humidity, vibration, or radiation dose rate measurements. While these methods can provide basic radiation information, they cannot simultaneously obtain intuitive on-site images and the temporal and spatial correlation of radiation distribution. This makes it difficult to quickly locate the leak source and assess the scope of contamination during emergency response.
[0004] Gamma cameras, as devices that fuse visual and radiation fields, generate radiation distribution images through a combination of scintillators and cameras. However, practical applications of gamma cameras have the following limitations: First, the devices are expensive and bulky, making them difficult to deploy flexibly in complex nuclear environments. Second, their detection limit is limited by the properties of the scintillator material, making them susceptible to signal saturation and even device damage in high-intensity radiation. Third, they have poor environmental adaptability, with performance significantly degrading in high-temperature, high-humidity, or strong electromagnetic interference environments. Fourth, the generated images are subject to radiation noise interference. Especially in high-energy radiation environments, CMOS sensors are prone to generating white speckling and other noise, which can obscure image details and cause errors in radiation dose rate measurements.
[0005] Current image noise suppression technologies (such as deep learning models) also face challenges: first, the inference time of complex models is usually tens to hundreds of milliseconds, which is difficult to meet the real-time requirements of nuclear leakage emergency scenarios; second, the model parameters are huge, which places strict requirements on the computing power and storage resources of embedded devices (such as FPGA modules or ARM chips), making it difficult to support real-time processing of high-resolution images.
[0006] Therefore, the existing technology has not yet effectively solved the problem of synchronous monitoring of visual field and radiation field under the conditions of low cost, high real-time performance and strong environmental adaptability. Summary of the Invention
[0007] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a nuclear environment monitoring system and method that combines visual field and radiation field, which solves the problem that traditional nuclear leakage monitoring methods are difficult to synchronously obtain on-site images and radiation distribution.
[0008] The technical solution of the present invention is: a nuclear environment monitoring system combining visual field and radiation field, comprising a camera, an FPGA module and a host computer;
[0009] The camera includes a CMOS sensor, an optical lens module, a circuit board, and a chip board. The CMOS sensor is fixedly mounted on a circuit board with a power supply interface. The optical lens module is fixedly mounted on the circuit board and coupled to the photosensitive surface of the CMOS sensor, which is used to guide external light to enter the photosensitive surface of the CMOS sensor. The chip board is connected to the circuit board and has a SOC chip installed on it, which is used to output continuous frame images containing radiation response signals.
[0010] The FPGA module includes a noise suppression submodule and a wavelet transform submodule; the FPGA module is communicatively connected to the chip board, and the noise suppression submodule and the wavelet transform submodule synchronously receive and parallelly process the image output by the camera; the noise suppression submodule is used to perform three-channel independent and parallel noise suppression on the image and output the noise-suppressed image; the wavelet transform submodule is used to perform Haar wavelet decomposition on the image and extract the diagonal high-frequency component, count the number of pixels contained in the specific pixel value segment therein, and then substitute the counted number of pixels into the "correspondence function between the number of pixels contained in the specific pixel value segment and the radiation dose rate" in the diagonal high-frequency component fitted based on prior experimental data to solve and obtain the radiation dose rate;
[0011] The host computer is connected to the FPGA module for communication. The host computer synchronously receives and displays on the same screen the noise suppression image output by the noise suppression submodule and the radiation dose rate output by the wavelet transform submodule.
[0012] The technical solution of the present invention is: a nuclear environment monitoring method combining visual field and radiation field, which is applied to a nuclear environment monitoring system combining visual field and radiation field, wherein the nuclear environment is limited to a single radiation field;
[0013] The method is as follows: the environmental image is collected by a camera, and the environmental image is a continuous frame image containing only a single type of radiation response signal; the frame image is synchronously transmitted to the noise suppression submodule and the wavelet transform submodule, and the noise suppression operation and radiation dose rate calculation are performed in parallel; the noise suppression operation is based on the noise suppression submodule, and the noise suppression image is output; the radiation dose rate is calculated based on the wavelet transform submodule, and the radiation dose rate value is output; the noise suppression image and the radiation dose rate are synchronously uploaded to the host computer, and the host computer temporally associates the two and displays them on the same screen.
[0014] A further technical solution of the present invention is: the process of the noise suppression operation is as follows:
[0015] I. Multi-frame image extraction and processing: Extract n consecutive frames of images output by the camera, extract the R, G, and B channels in each frame separately, and generate independent grayscale matrices for each, namely the R, G, and B channel grayscale matrices;
[0016] II. Pixel-by-pixel comparison of multiple frames: Compare the grayscale values of the same pixel in each of the n frames of image, one by one, for the grayscale matrices of the R, G, and B channels.
[0017] III. Temporary output of minimum grayscale value: Based on the premise that noise does not recur in consecutive frames, the minimum grayscale value is selected as the temporary output value of the target pixel position, replacing the current pixel value of the target pixel. A single temporary grayscale matrix for the R, G, and B channels is obtained, thus achieving the first noise filtering;
[0018] IV. Median filtering of each single channel: Median filtering is performed on the individual temporary grayscale matrices of the R, G, and B channels after the initial noise filtering. First, a neighborhood window is used to traverse each pixel point. Then, all pixel values within the neighborhood window are sorted and the median value is taken to replace the current pixel value. This results in a single final grayscale matrix for the R, G, and B channels, thus achieving secondary noise filtering.
[0019] V. Three-channel grayscale matrix merging: Merge the final grayscale matrices of the R, G, and B channels after secondary noise filtering into a frame of color image and output it to the host computer for display.
[0020] A further technical solution of the present invention is: the calculation process of the radiation dose rate is as follows:
[0021] From the continuous frame images output by the camera, one frame is taken from every consecutive n frames for Haar wavelet decomposition, and the diagonal component is extracted. Then, based on the "correspondence function between the number of pixels contained in a specific pixel value segment and the radiation dose rate" obtained by fitting based on the prior experimental data, the number of pixels contained in the specific pixel value segment obtained by statistics is substituted into the function to solve the radiation dose rate at the camera location and output it to the host computer for display.
[0022] A further technical solution of the present invention is that the n frames of images selected for radiation dose rate calculation are exactly the same as the n frames of images extracted by the noise suppression operation, so that the calculated radiation dose rate and the noise suppressed image are temporally correlated.
[0023] A further technical solution of the present invention is that the radiation field is a single gamma radiation field, and correspondingly, the specific pixel value range is 10-70.
[0024] Compared with the prior art, the present invention has the following advantages:
[0025] 1. It innovatively utilizes the dual response characteristics of CMOS sensors to radiation fields, enabling the simultaneous acquisition of visual images and radiation information through a single device. Specifically, while a CMOS sensor captures images of a visible light scene, its photosensitive elements are bombarded by radiation particles (such as gamma rays), generating radiation noise (such as bright white spots). This noise not only interferes with visual information but also reveals characteristics of radiation intensity.
[0026] 2. It achieves "integrated monitoring" through the following designs:
[0027] ① Dual analysis of noise: While suppressing radiation noise (such as multi-frame minimum filtering and median filtering) to restore a clear image, high-frequency components (such as Haar wavelet diagonal components) are extracted from the noise signal and converted into radiation dose rate through a priori functions;
[0028] ② Advantages of device integration: Abandoning traditional discrete visual cameras and radiation detectors (such as gamma cameras), only a single CMOS camera is needed to simultaneously output denoised on-site images and radiation data, significantly reducing hardware costs and deployment complexity, and avoiding the spatiotemporal deviation problem of multi-device data fusion;
[0029] ③ Signal homology guarantee: Since the visual images and radiation data are derived from the original signal of the same sensor, the two are strictly synchronized in time and space dimensions, providing high-confidence correlation information for nuclear leak positioning and contamination range assessment.
[0030] 3. It realizes nuclear environment monitoring by combining visual field and radiation field, and can display the scene image and radiation dose rate on the same screen of the host computer; it makes full use of the parallel architecture of the FPGA module to simultaneously perform image denoising operations and calculate radiation dose rates, greatly improving the efficiency of the method and meeting the rapid response requirements of nuclear leakage emergency scenarios.
[0031] 4. Image denoising has the following advantages:
[0032] ①. Taking advantage of the sparsity of radiation noise (noise at the same location rarely appears in consecutive frames), the pixel values at the same location in multiple frames are compared and the minimum value is selected for output, thereby effectively filtering out noise while retaining real scene details.
[0033] ②. Use shift registers to store multiple image frames to avoid DRAM access delays; decouple the camera pixel clock (24MHz) and Ethernet transmission clock (50MHz) through FIFO to reduce timing conflicts; fully utilize the parallel architecture of the FPGA module to process multiple image frames simultaneously, significantly improving data throughput and achieving millisecond-level noise filtering.
[0034] ③. It utilizes the differences in the distribution of radiation noise in the three channels R, G, and B to independently and parallelly process the three-channel grayscale matrices, improving operational efficiency and avoiding color distortion caused by traditional grayscale processing. It can also set differentiated thresholds for different channels (such as n=5 frames for the red channel and n=3 frames for the green / blue channels), thereby achieving targeted suppression of noise intensity.
[0035] 5. Advantages of radiation dose rate calculation process:
[0036] ①. Utilizing the computationally low-complexity Haar wavelet decomposition, we rapidly separate the high-frequency components of the image, particularly the diagonal high-frequency components, effectively capturing the characteristics of radiation noise. Prior experiments have shown that diagonal components are significantly more sensitive to radiation dose rate than horizontal / vertical components within specific pixel value ranges (e.g., the 10-70 range for gamma radiation fields), as evidenced by the distinct curves. This avoids feature overlap and improves computational accuracy.
[0037] ②. By fitting a function based on prior experimental data, the statistical number of pixels in the range of 10-70 pixels is directly converted into radiation values. The model is based on measured data in actual radiation environments, reducing theoretical assumption errors and ensuring the adaptability and reliability of the results for a single gamma radiation field.
[0038] ③. The algorithms for Haar wavelet decomposition and pixel value statistics are low in complexity and suitable for parallel implementation in FPGA modules. By extracting one frame for processing every n frames (synchronized with noise suppression), this reduces the computational load while avoiding sacrificing real-time performance, meeting the resource constraints of embedded devices.
[0039] ④. Radiation dose rate calculation and noise suppression use the same n-frame images, ensuring strict temporal synchronization between the two data sets to avoid errors caused by frame sequence misalignment. Furthermore, wavelet decomposition is used to filter out low-frequency background interference and focus on high-frequency radiation noise, enhancing resistance to environmental interference.
[0040] 6. Advantages of choosing FPGA modules;
[0041] 1. Millisecond Response: Nuclear leak scenarios require the system to rapidly output denoised images and radiation dose rates. Traditional deep learning models struggle to meet real-time requirements due to computational complexity (latency of tens to hundreds of milliseconds). FPGAs, however, directly implement algorithms through hardware logic, compressing processing time to milliseconds.
[0042] ②. Dual-module parallelism: The FPGA's hardware parallelism allows the noise suppression submodule and the wavelet transform submodule to synchronously process the same batch of image data. For example, when the camera outputs continuous frame images, the noise suppression submodule independently performs multi-frame comparison and median filtering on the three RGB channels, leveraging the FPGA's three-channel parallel computing capability to simultaneously process the R, G, and B data streams. Simultaneously, the wavelet transform submodule performs Haar wavelet decomposition on the extracted frames, extracting diagonal high-frequency components and counting pixel values to ensure that radiation dose rate calculation and image denoising are completed simultaneously. Dual-module parallelism avoids the data backlog problem of traditional serial processing and significantly improves data throughput.
[0043] ③. Customized Logic Design: The programmability of FPGAs allows for optimized hardware resource allocation for specific algorithms. For example, multi-frame processing optimization: Storing multiple frames of image data through shift register chains avoids external DRAM access latency and enables direct, fast pixel-level comparison. Differentiated Channel Processing: Addressing the differences in noise characteristics between RGB channels (e.g., more pronounced red channel radiated noise), independent processing logic (e.g., different frame count thresholds) is configured for each channel at the hardware level, avoiding color distortion caused by traditional grayscale processing.
[0044] ④. Low resource usage: Operations such as Haar wavelet decomposition and pixel statistics are implemented through FPGA hardware logic, eliminating the need for complex computing units and making it suitable for embedded devices with limited computing resources.
[0045] The present invention is further described below with reference to the figures and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] Figure 1 This is a diagram showing the electrical connections of the components of the present invention;
[0047] Figure 2 Schematic diagram of the camera structure;
[0048] Figure 3 Flowchart of the nuclear environment monitoring method of the present invention;
[0049] Figure 4 This is the relationship between the grayscale value and the number of pixels in the horizontal component obtained by Haar wavelet decomposition;
[0050] Figure 5 This is the relationship between the grayscale value and the number of pixels in the vertical component obtained by Haar wavelet decomposition;
[0051] Figure 6 This is the relationship between the grayscale value and the number of pixels in the diagonal component obtained by Haar wavelet decomposition;
[0052] Figure 7 This is a graph showing the relationship between the radiation dose rate and the number of pixels in the range of 10-70 pixel values.
[0053] Legend: CMOS sensor 11; optical lens module 12; circuit board 13; chip board 14; SOC chip 141. DETAILED DESCRIPTION
[0054] Example 1:
[0055] like Figure 1-2 As shown, the nuclear environment monitoring system combining visual field and radiation field includes a camera 1, an FPGA module 2 and a host computer 3.
[0056] Camera 1 includes a CMOS sensor 11, an optical lens module 12, a circuit board 13, and a chip board 14. The CMOS sensor 11 is fixedly mounted on the circuit board 13, which has a power supply interface. The optical lens module 12 is fixedly mounted on the circuit board 13 and coupled to the photosensitive surface of the CMOS sensor 11, directing external light to the photosensitive surface of the CMOS sensor 11. The chip board 14 is in communication with the circuit board 13 and is mounted with a SOC chip 141, which is used to output continuous frame images containing radiation response signals.
[0057] The FPGA module 2 includes a noise suppression submodule 21 and a wavelet transform submodule 22. The FPGA module 2 is communicatively connected to the chip board 14, and the noise suppression submodule 21 and the wavelet transform submodule 22 synchronously receive and parallelly process the image output by the camera 1 (output by the chip board 14). The noise suppression submodule 21 is used to perform independent and parallel noise suppression on the three channels (R, G, B channels) of the image and output the noise-suppressed image. The wavelet transform submodule 22 is used to perform Haar wavelet decomposition on the image and extract the diagonal high-frequency component, count the number of pixels contained in the specific pixel value segment therein, and then substitute the counted number of pixels into the "correspondence function between the number of pixels contained in the specific pixel value segment and the radiation dose rate" in the diagonal high-frequency component fitted based on the prior experimental data to solve and obtain the radiation dose rate.
[0058] The host computer 3 is in communication connection with the FPGA module 2 , and the host computer 3 synchronously receives and displays on the same screen the noise suppression image output by the noise suppression submodule and the radiation dose rate output by the wavelet transform submodule.
[0059] like Figure 3 As shown, the nuclear environment monitoring method based on data parallel processing is applied to the above-mentioned nuclear environment monitoring system combining the visual field and the radiation field, and the nuclear environment is limited to a single gamma radiation field.
[0060] The method is as follows: the environmental image is collected by a camera, and the environmental image is a continuous frame image containing only the γ response signal; the frame image is synchronously transmitted to the noise suppression submodule and the wavelet transform submodule, and the noise suppression operation and radiation dose rate calculation are performed in parallel; the noise suppression operation is based on the noise suppression submodule, and the noise suppression image is output; the radiation dose rate is calculated based on the wavelet transform submodule, and the radiation dose rate value is output; the noise suppression image and the radiation dose rate are synchronously uploaded to the host computer, and the host computer temporally associates the two and displays them on the same screen.
[0061] The noise suppression process is as follows:
[0062] Ⅰ. Multi-frame image extraction and processing: Extract n consecutive frames of images output by the camera, extract the R, G, and B channels in each frame of the image separately, and generate independent grayscale matrices, namely the R, G, and B channel grayscale matrices.
[0063] II. Pixel-by-pixel comparison of multiple frames: For the grayscale matrices of the R, G, and B channels, compare the grayscale values of the same pixel position in n frames of images one by one.
[0064] III. Temporary output of minimum grayscale value: Based on the premise that noise does not recur in consecutive frames, the minimum grayscale value is selected as the temporary output value of the target pixel position, replacing the current pixel value of the target pixel point, and obtaining a single temporary grayscale matrix for the R, G, and B channels, thereby achieving the first noise filtering. The first noise filtering filters out most of the random noise, but there may still be residual isolated noise points.
[0065] IV. Median filtering of each single channel: Median filtering is performed on the individual temporary grayscale matrices of the R, G, and B channels after the initial noise filtering. First, a neighborhood window (3×3 or 5×5) is used to traverse each pixel point (i.e., with each pixel point as the center, all pixel values in the surrounding 3×3 or 5×5 area are checked). Then, all pixel values in the neighborhood window are sorted, and the median value is taken to replace the current pixel value. The final grayscale matrices of the R, G, and B channels are obtained, respectively, thereby achieving secondary noise filtering (residual isolated noise points are eliminated).
[0066] V. Three-channel grayscale matrix merging: Merge the final grayscale matrices of the R, G, and B channels after secondary noise filtering into a frame of color image and output it to the host computer for display.
[0067] The calculation process of radiation dose rate is as follows:
[0068] From the continuous frame images output by the camera, one frame is taken from every consecutive n frames for Haar wavelet decomposition, and the diagonal component is extracted. Then, based on the "corresponding relationship function between the number of pixels contained in the 10-70 pixel value segment and the radiation dose rate" fitted based on the prior experimental data, the number of pixels contained in the 10-70 pixel value segment obtained by statistics is substituted into the function to solve the radiation dose rate at the camera location and output it to the host computer for display.
[0069] Preferably, the n frames of images selected for the radiation dose rate calculation are identical to the n frames of images extracted by the noise suppression operation, so that the calculated radiation dose rate and the noise suppressed image are temporally correlated.
[0070] Preferably, the radiation field is a single gamma radiation field, and correspondingly, the specific pixel value range is 10-70.
[0071] Description of extracting only the diagonal components after wavelet decomposition:
[0072] The image output by the camera is decomposed by Haar wavelet to obtain horizontal component, vertical component and diagonal component, which are respectively as follows: Figure 4 、 Figure 6 and Figure 6 As shown in the figure, the horizontal axis is the gray value, the vertical axis is the number of pixels, the curves of different colors represent different radiation dose rates, and the coordinates of any point in the curve represent the number of pixels contained in a specific pixel value in the image (for example, the coordinates of point A on the black curve are (69, 9×10 2 ), which means that when the radiation dose rate is 51.60 Gyh, the number of pixels with a pixel value of 69 in the image is 9×10 2 Combined Figure 4-6 It can be seen that in the graphs of the horizontal and vertical components, the curves corresponding to different radiation dose rates almost overlap and are difficult to distinguish. However, in the graph of the diagonal component, different radiation dose rates are clearly distinguished in the range of 10-70. Therefore, it is most appropriate to use the diagonal component to characterize the radiation dose rate.
[0073] Explanation of fitting function based on prior experimental data:
[0074] In a single gamma radiation field with a known radiation dose rate, a radiation experiment was conducted using a nuclear environment monitoring system that combines visual and radiation fields (the CMOS sensor model remained consistent throughout). The image captured by the camera was input into the wavelet transform submodule for Haar wavelet decomposition. The diagonal components were extracted, and the number of pixels within the 10-70 pixel value range was counted. The radiation dose rate in the single gamma radiation field was varied multiple times, and the above process was repeated to obtain the number of pixels within the 10-70 pixel value range. This ultimately yielded multiple sets of paired data, each containing the radiation dose rate and the corresponding number of pixels within the 10-70 pixel value range.
[0075] Establish a corresponding relationship diagram between radiation dose rate and the number of pixels in the range of 10-70 pixel values, see Figure 7 In the figure, the horizontal axis is the radiation dose rate, and the vertical axis is the number of pixels in the range of 10-70 pixel values. All the paired data obtained above are marked on Figure 7 The coordinate system of , and then fit the function ( Figure 7 straight line in the middle).
Claims
1. A nuclear environment monitoring system combining visual field and radiation field, characterized by: Including camera , FPGA module and host computer; The camera includes a CMOS sensor, an optical lens module, a circuit board, and a chip board. The CMOS sensor is fixedly mounted on a circuit board with a power supply interface. The optical lens module is fixedly mounted on the circuit board and coupled to the photosensitive surface of the CMOS sensor, which is used to guide external light to enter the photosensitive surface of the CMOS sensor. The chip board is connected to the circuit board and has a SOC chip installed on it, which is used to output continuous frame images containing radiation response signals. The FPGA module includes a noise suppression submodule and a wavelet transform submodule; the FPGA module is connected to the chip board for communication, and the noise suppression submodule and the wavelet transform submodule synchronously receive and parallelly process images output by the camera; The noise suppression submodule is used to perform three-channel independent and parallel noise suppression on the image and output the noise-suppressed image. The wavelet transform submodule is used to perform Haar wavelet decomposition on the image and extract the diagonal high-frequency component. The number of pixels contained in the specific pixel value segment is counted. Based on the "corresponding relationship function between the number of pixels contained in the specific pixel value segment and the radiation dose rate" in the diagonal high-frequency component fitted based on prior experimental data, the counted number of pixels is substituted into the function to solve for the radiation dose rate. The host computer is connected to the FPGA module for communication. The host computer synchronously receives and parallelly processes the image output by the noise suppression submodule and the radiation dose rate output by the wavelet transform submodule.
2. A nuclear environment monitoring method combining visual field and radiation field, applied to the nuclear environment monitoring system combining visual field and radiation field according to claim 1, characterized in that: The nuclear environment described is limited to a single radiation field; The method is as follows: an environmental image is captured by a camera, and the environmental image is a continuous frame image containing only a single type of radiation response signal; the frame image is synchronously transmitted to a noise suppression submodule and a wavelet transform submodule, which perform noise suppression operations and calculate radiation dose rates in parallel; the noise suppression operation is based on the noise suppression submodule, and a noise-suppressed image is output; The radiation dose rate is calculated based on the wavelet transform submodule and the radiation dose rate value is output; the noise suppression image and the radiation dose rate are uploaded to the host computer synchronously, and the host computer correlates the two in time and displays them on the same screen.
3. The nuclear environment monitoring method combining visual field and radiation field as claimed in claim 2, characterized in that: The noise suppression process is as follows: I. Multi-frame image extraction and processing: Extract n consecutive frames of images output by the camera, extract the R, G, and B channels in each frame separately, and generate independent grayscale matrices for each, namely the R, G, and B channel grayscale matrices; II. Pixel-by-pixel comparison of multiple frames: Compare the grayscale values of the same pixel in each of the n frames of image, one by one, for the grayscale matrices of the R, G, and B channels. III. Temporary output of minimum grayscale value: Based on the premise that noise does not recur in consecutive frames, the minimum grayscale value is selected as the temporary output value of the target pixel position, replacing the current pixel value of the target pixel. A single temporary grayscale matrix for the R, G, and B channels is obtained, thus achieving the first noise filtering; IV. Median filtering of each single channel: Median filtering is performed on the individual temporary grayscale matrices of the R, G, and B channels after the initial noise filtering. First, a neighborhood window is used to traverse each pixel point. Then, all pixel values within the neighborhood window are sorted and the median value is taken to replace the current pixel value. This results in a single final grayscale matrix for the R, G, and B channels, thus achieving secondary noise filtering. V. Three-channel grayscale matrix merging: Merge the final grayscale matrices of the R, G, and B channels after secondary noise filtering into a frame of color image and output it to the host computer for display.
4. The nuclear environment monitoring method combining visual field and radiation field as claimed in claim 3 is characterized by: The calculation process of radiation dose rate is as follows: From the continuous frame images output by the camera, one frame is taken from every consecutive n frames for Haar wavelet decomposition, and the diagonal components are extracted. Then, based on the "corresponding relationship function between the number of pixels contained in a specific pixel value segment and the radiation dose rate" obtained by fitting based on prior experimental data, the number of pixels contained in the specific pixel value segment obtained by statistics is substituted into the function to solve the radiation dose rate at the camera location and output it to the host computer for display.
5. The nuclear environment monitoring method combining visual field and radiation field as claimed in claim 4 is characterized by: The n frames of images selected for the radiation dose rate calculation are identical to the n frames of images extracted by the noise suppression operation, so that the calculated radiation dose rate and the noise suppressed image are temporally correlated.
6. The nuclear environment monitoring method combining visual field and radiation field as claimed in claim 5, characterized in that: The radiation field is a single gamma radiation field, and accordingly, the specific pixel value range is 10-70.