A low-energy section background optimization method based on SNIP gamma spectrum

Abstract

The application discloses a kind of based on SNIP gamma spectrum low-energy section background optimization method.Method includes: the original gamma spectrum to be processed is collected, vertical mirror image transformation is carried out to map high-count background area as low-count background area, and mirror image spectrum vector is obtained;After logarithm-log square transformation is carried out to mirror image spectrum vector, then iteration screening large value processing is carried out, and mirror image domain background estimation is obtained;Mirror image domain background estimation is carried out anti-mirror image restoration operation, and the background estimation result of original gamma spectrum after optimization is obtained.The application can fundamentally inhibit gamma spectrum low-energy section background deduction error, effectively avoid low-energy section background deduction incomplete and false peak misjudgment, reduce parameter sensitivity and parameter adjustment difficulty, significantly improve low-energy weak signal analysis precision, significantly improve the background deduction precision and reliability of low-energy weak signal peak area, keep algorithm universality and implementation simplicity, easy to integrate deployment in existing gamma spectrum analysis software or embedded system.

Description

Technical Field

[0001] This invention relates to a gamma spectrum background optimization method, which relates to the field of nuclear radiation detection and energy spectrum signal processing technology, and specifically to a low-energy background optimization method based on SNIP gamma spectrum. Background Technology

[0002] In the field of nuclear physics and environmental radiation monitoring, gamma spectra obtained through detectors (such as NaI(Tl) detectors and high-purity germanium (HPGe) detectors) are fundamental for identifying radionuclides and calculating their activities. Due to internal material effects within the detector (such as Compton scattering and electron escape) and the natural background radiation of the surrounding environment, the raw gamma spectrum contains a large amount of continuous, featureless background signals. Accurately subtracting the background and extracting the net photoelectric peak area is a crucial prerequisite for subsequent qualitative and quantitative analysis of nuclides and their activities. Therefore, efficient, robust, and adaptive background subtraction algorithms are a core technical requirement in this field.

[0003] The Statistically Sensitive Nonlinear Iterative Peak-Clipping (SNIP) algorithm is one of the most widely used background subtraction algorithms. Its core idea is to compress the dynamic range of the data and suppress the influence of strong peaks through a log-log-squared LLS (Log-Log Square) transform. Iteratively performing peak clipping operations within a sliding window (preserving local minima) approximates a smooth background profile, and finally, the estimated background energy spectrum is obtained through an inverse transform. This algorithm is widely used in various gamma-ray spectroscopy analysis systems due to its advantages of not requiring a pre-defined background model and its high computational efficiency.

[0004] However, with the increasing complexity of application scenarios and the diversification of detector types, the shortcomings of traditional SNIP algorithms in processing low-energy (typically below 100 keV) gamma spectra are becoming increasingly prominent, mainly in the following two aspects: 1. Low-Energy Compton Flat Residue and False Peaks: In the low-energy range, the continuous background (Compton flat) formed by Compton scattering has a steep slope and a high count rate. The LLS transform of the traditional SNIP algorithm maps this high-count region to a region with larger numerical values ​​and higher data density in the transform domain. When performing iterative peak reduction operations in this region, any small numerical fluctuations or estimation errors will be significantly amplified in the subsequent inverse LLS transform, resulting in incomplete background subtraction. Some background residues are misjudged as weak false peaks, seriously affecting the detection limit and quantitative accuracy of low-activity nuclides.

[0005] 2. Adaptability and parameter dependence for energy peaks of different intensities: The effectiveness of the SNIP algorithm is highly dependent on the key parameter "iteration number m". A smaller m value is needed to prevent over-subtraction when processing weak peaks, while a larger m value is needed to ensure that the peak is fully stripped when processing strong peaks. This contradiction makes the algorithm heavily reliant on operator experience for manual parameter tuning in practical applications, resulting in low automation, poor robustness, and difficulty in handling the analysis of samples with large dynamic ranges of energy spectrum intensity or unknown complex samples.

[0006] Existing improvement schemes mostly focus on iterative parameter adjustment, external noise filtering, or specific hardware adaptation. No method has yet been found to effectively circumvent the fundamental flaw in the traditional SNIP algorithm—the inherent error amplification effect of the LLS transform in the low-energy, high-background region—from a mathematical perspective. Therefore, there is an urgent need for a SNIP optimization method that starts from the core mechanism of the algorithm, fundamentally suppressing the low-energy background subtraction error, reducing parameter sensitivity, and improving adaptive capabilities. Summary of the Invention

[0007] To address the problems existing in the background technology, this invention provides a low-energy background optimization method based on SNIP gamma spectrum. This method solves the technical problems of traditional SNIP algorithms, such as large background estimation bias, susceptibility to false peaks, and high sensitivity to the iteration parameter m in the Compton plateau region of low energy ranges (energy below 100 keV).

[0008] The technical solution adopted in this invention is: The present invention provides a method for optimizing the low-energy background of the SNIP gamma spectrum, comprising: Step S1: The raw gamma spectrum to be processed is acquired based on the gamma detector. The raw gamma spectrum is processed based on the improved statistically sensitive nonlinear iterative peak stripping SNIP algorithm. First, the raw gamma spectrum is subjected to vertical mirror transformation to map the high count background region in the raw gamma spectrum to the low count background region, and the mirror spectrum vector is obtained.

[0009] Step S2: Then, the mirror energy spectrum vector is iteratively filtered to remove large values, and the background estimate of the mirror domain is obtained.

[0010] Step S3: Finally, the mirror domain background estimate is reversed and restored to obtain the optimized background estimate of the original gamma spectrum.

[0011] In step S1, the maximum count value of the original gamma spectrum is obtained for vertical mirror transformation, as detailed below: y udr (i) = y max - y(i) Among them, y udr (i) represents the mirror energy spectrum vector, and i is the energy spectrum channel address index; ymax y(i) is the maximum count value of the original gamma spectrum; y(i) is the vector of the original gamma spectrum to be processed.

[0012] In step S2, during the iterative screening of large values, the mirror energy spectrum vector is first subjected to a log-log-squared transformation to obtain the transform domain vector v(i). Then, a sliding window iteration is performed on the transform domain vector v(i). In each iteration, the larger value in the sliding window is retained to fill the valley corresponding to the original high background region in the mirror domain. After the iteration is completed, a log-log-squared inverse transformation is performed to obtain the background estimate of the mirror domain.

[0013] The sliding window iteration is described in detail below: v p (i) = max{ v p-1 (i), [v p-1 (i+p) + v p-1 (ip) ] / 2} Among them, v p (i) and v p-1 (i) are the mirror domain background estimates for the p-th and p-1-th iterations, respectively, where p = 1, 2, ..., m, and m is the preset iteration number, v0(i) = v(i); v p-1 (ip) and v p-1 (i+p) are the mirror domain background estimates corresponding to the left boundary ip and right boundary i+p of the computation window centered at i and with radius p in the (p-1)th iteration, respectively; max{} is the maximum value.

[0014] In step S3, the reverse mirror restoration operation is specifically a vertical mirror inverse transformation, as follows: b(i) = y max - b udr (i) Where b(i) is the optimized background estimate of the original gamma spectrum; b udr (i) is the estimated background vector of the mirror domain after the log-log-square inverse transform.

[0015] The original gamma spectrum is a low-energy gamma spectrum with energies below 100 keV.

[0016] The electronic device of the present invention includes: a memory and a processor coupled to each other, wherein the memory stores program data, and the processor invokes the program data to execute the method described above.

[0017] The present invention provides a computer-readable storage medium having program data stored thereon, characterized in that the program data, when executed by a processor, implements the method described above.

[0018] The beneficial effects of this invention are: 1. Fundamentally suppress low-energy background subtraction error: By using vertical mirror transformation, the high-count background region (such as Compton plateau) in the original energy spectrum is mapped to a low-count valley region, so that it is in a low-density region with a small ordinate value in the LLS transform domain; in this region, the numerical fluctuations introduced by iterative operations are greatly weakened even after LLS inverse transformation, thus effectively avoiding incomplete low-energy background subtraction and false peak misjudgment.

[0019] 2. Reduced parameter sensitivity and tuning difficulty: Since the mirror domain iteration adopts the "valley filling" strategy of retaining larger values, it works in conjunction with the mirror transformation to solve the problem that the original SNIP algorithm is prone to overcomputation or noise amplification in the slender high background region; the optimized algorithm is significantly less sensitive to the iteration parameter m, and the processing effect of energy peaks of different intensities is more stable and consistent, thus improving the automation level of the algorithm.

[0020] 3. Significantly improves the accuracy of low-energy weak signal analysis: In the Compton plateau scenario with no peaks below 100 keV, the net energy spectrum residual count obtained by the method of this invention is reduced from 2348 in the traditional SNIP algorithm to 929, and the background residual is significantly reduced; for Americium-241 at 59.5 keV ( 241 The actual net peak area of ​​the weak peak (Am) is 1698, while the net peak area calculated by the traditional SNIP algorithm is 3091, with a relative error of 82.04%. The net peak area calculated by the method of this invention is 1979, and the relative error is reduced to 16.55%, which significantly improves the accuracy and reliability of background subtraction in the low-energy weak signal peak region.

[0021] 4. Maintaining the universality and ease of implementation of the algorithm: While keeping the computational framework of the traditional SNIP algorithm basically unchanged, this invention only adds mirror transformation and changes the iterative value selection strategy to achieve optimization. The increase in computational complexity is limited and it is easy to integrate and deploy in existing gamma spectroscopy analysis software or embedded systems. Attached Figure Description

[0022] Figure 1 A flowchart illustrating a method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram comparing the background estimation results of the method of this invention and the traditional SNIP algorithm in the low-energy-range scenario without energy peaks; Figure 3 This diagram illustrates a comparison of the net peak area calculation results using the method of this invention and the traditional SNIP algorithm in a low-energy weak-peak scenario. Detailed Implementation

[0023] To make the above-mentioned objectives, technical solutions, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that, unless otherwise specified, the parameters in the following applications can be adjusted according to actual circumstances, and these adjustments do not affect the determination of the scope of protection of the present invention.

[0024] In practice, a gamma-ray spectrometer equipped with a multichannel analyzer was used as the implementation platform, and a 3-inch NaI(Tl) scintillation detector was selected for measuring Americium-241 (… 241 Am) and potassium-40 in the environment ( 40 The γ-ray spectrum of K). The optimization method is implemented as a software algorithm in an industrial control computer, and the steps are programmed into functional modules using C / C++ programming language. The input of the software algorithm is the raw gamma-ray spectrum vector y generated by the detector and multichannel analyzer. The channel address range of the raw gamma-ray spectrum vector y is 0 to 1023, and the energy resolution corresponding to each channel is related to the detector performance. The output of the software algorithm is the estimated background energy spectrum vector b. In specific implementation, the method is deployed as a software algorithm in an industrial control computer, embedded microprocessor, field-programmable gate array (FPGA), or high-performance computer. The user can select whether to enable the low-energy optimization mode through the human-machine interface. When the low-energy optimization mode is enabled, the software automatically calls the algorithm flow of the following steps of the present invention.

[0025] Specifically, such as Figure 1 As shown, the execution flow of the low-energy background optimization method based on SNIP gamma spectrum of the present invention is as follows: First, the raw gamma spectrum y(i), i = 0, 1, 2, ..., 1023, is acquired using a NaI(Tl) scintillator gamma detector. The raw gamma spectrum is a low-energy gamma spectrum with energies below 100 keV. The raw gamma spectrum vector y(i) is smoothed to suppress random statistical noise; the smoothing window width is preferably 3 to 7 channels. The raw gamma spectrum is then processed using an improved statistically sensitive nonlinear iterative peak stripping SNIP algorithm. First, the raw gamma spectrum vector y(i) is traversed to obtain the maximum count value of the raw gamma spectrum. Then, a vertical mirror transformation is performed on each channel address, as follows: y udr (i) = y max - y(i) y max = max{y(i)} Among them, y udr (i) represents the mirror energy spectrum vector, and i is the energy spectrum channel address index; y max y(i) is the maximum count value of the original gamma spectrum; y(i) is the vector of the original gamma spectrum to be processed.

[0026] The high count background region in the original gamma spectrum is mapped to a low count background region to obtain the mirror energy spectrum vector.

[0027] Then, the mirror energy spectrum vector is iteratively filtered to remove large values. During the iterative filtering process, each element of the mirror energy spectrum vector is first subjected to a logarithmic-logarithmic-squared LLS transform to obtain the transform domain vector v(i), as follows: v(i) = ln[ ln( sqrt(y udr (i)) + 1 ) + 1 ] Where sqrt() represents the square root.

[0028] Then, a sliding window iteration is performed on the transformed domain vector v(i), starting from p = 1 and iterating to p = m. In each iteration, the larger value in the sliding window is retained to fill the valley corresponding to the original high background region in the mirror domain, as follows: v p (i) = max{ v p-1 (i), [v p-1 (i+p) + v p-1 (ip) ] / 2} Among them, v p (i) and v p-1 (i) are the background estimates of the mirror domain for the p-th and p-1-th iterations, respectively, where p = 1, 2, ..., m, and m is the preset iteration number. The iteration number is set to m = 4, but the value of m can be adjusted between 4 and 8 depending on the channel number and energy resolution of the spectrometer. Initialize v0(i) = v(i); p-1 (ip) and v p-1 (i+p) represents the background estimates of the mirror domain corresponding to the left boundary ip and right boundary i+p of the calculation window centered at i and with radius p in the (p-1)th iteration. The boundary processing adopts the symmetric extension method, that is, when i+p or ip exceeds the energy spectrum range, the value at the boundary is taken as the replacement; max{} is to take the maximum value.

[0029] After completing m iterations, the result v of the m-th iteration is... m (i) Perform a log-log-squared LLS inverse transform to obtain the baseline estimation vector b in the mirror domain. udr (i) Estimation vector b of the mirror domain background udr (i) Perform the opposite inverse mirror restoration operation, specifically the vertical mirror inverse transformation, as follows: b(i) = y max - b udr (i) b udr (i) = { exp[ exp( vm (i) ) - 1 ] - 1} 2 Where b(i) is the optimized background estimate of the original gamma spectrum; b udr (i) is the estimated background vector of the mirror domain after the log-log-square inverse transform.

[0030] Subtracting the original gamma spectrum vector y(i) from the background estimation vector b(i) yields the net spectrum vector n(i) = y(i) - b(i), which is used for subsequent nuclide identification and activity calculation.

[0031] To verify the beneficial effects of the method of this invention, the same experimental setup and measurement conditions as described above were used. Background estimation and net energy spectrum calculation were performed on the same set of raw energy spectrum data using both the traditional SNIP algorithm and the method of this invention. The performance of the two methods was then compared and analyzed. The experimental conditions were set as follows: a 3-inch NaI(Tl) scintillation detector, a multichannel analyzer with 1024 channels, and an energy range of approximately 0 keV to 3000 keV. The energy range below 100 keV was selected for focused analysis.

[0032] like Figure 2 As shown, in the Compton plateau peakless scenario, where the energy is below 100 keV, the ideal net energy spectrum should be close to zero. The total net energy spectrum count calculated using the traditional SNIP algorithm is 2348, indicating a significant background surplus; the total net energy spectrum count calculated using the method of this invention is 929, a reduction of approximately 60.4% in background surplus, indicating that the method of this invention can more thoroughly subtract the background in peakless scenarios.

[0033] like Figure 3 As shown, in the low-energy weak peak scenario, americium-241 exists at 59.5 keV. 241 A weak peak (Am) is known to have an actual net peak area of ​​1698. The net peak area calculated using the traditional SNIP algorithm is 3091, with a relative error of 82.04%, nearly double the actual net peak area. This indicates that the traditional algorithm miscounts a large amount of residual background as peak area, forming false peaks or causing deviations in quantitative results. The net peak area calculated using the method of this invention is 1979, with a relative error of 16.55%, significantly reducing the error compared to the traditional method. This demonstrates that the method of this invention has a significant accuracy advantage in low-energy weak peak scenarios.

[0034] The experimental results above show that the method of the present invention can effectively solve the problems of severe background residue, easy generation of false peaks and high sensitivity to parameters in the background estimation of low-energy gamma spectrum by the traditional SNIP algorithm. It significantly improves the background subtraction accuracy and reliability of low-energy weak signal peak region and has important practical application value in applications with extremely high requirements for the resolution accuracy of low-energy weak signals, such as environmental radiation monitoring and nuclear emergency screening.

[0035] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented using various computer languages. This application is described with flowcharts of methods, systems, and computer program products according to embodiments of this application.

[0036] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, this invention is intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0037] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if these modifications and variations of this application fall within the scope of the equivalent technology of this invention, this application also intends to include these modifications and variations.

[0038] The above specific embodiments are used to explain and illustrate the present invention, but not to limit the present invention. Any modifications and changes made to the present invention within the spirit and scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for optimizing the low-energy background of SNIP gamma spectrum, characterized in that, include: Step S1: The raw gamma spectrum to be processed is acquired based on the gamma detector. The raw gamma spectrum is processed based on the improved statistically sensitive nonlinear iterative peak stripping SNIP algorithm. First, the raw gamma spectrum is subjected to vertical mirror transformation to map the high count background region in the raw gamma spectrum to the low count background region, and the mirror spectrum vector is obtained. Step S2: Then, the mirror energy spectrum vector is iteratively filtered to remove large values, and the background estimate of the mirror domain is obtained. Step S3: Finally, the mirror domain background estimate is reversed and restored to obtain the optimized background estimate of the original gamma spectrum.

2. The method for optimizing the low-energy background of SNIP gamma spectrum according to claim 1, characterized in that: In step S1, the maximum count value of the original gamma spectrum is obtained for vertical mirror transformation, as detailed below: y udr (i) = y max - y(i) Among them, y udr (i) represents the mirror energy spectrum vector, and i is the energy spectrum channel address index; y max y(i) is the maximum count value of the original gamma spectrum; y(i) is the vector of the original gamma spectrum to be processed.

3. The method for optimizing the low-energy background of SNIP gamma spectrum according to claim 1, characterized in that: In step S2, during the iterative screening of large values, the mirror spectrum vector is first subjected to a logarithmic-logarithmic squared transform to obtain the transform domain vector v(i). Then, a sliding window iteration is performed on the transform domain vector v(i). In each iteration, the larger value in the sliding window is retained. After the iteration is completed, a logarithmic-logarithmic squared inverse transform is performed to obtain the background estimate of the mirror domain.

4. The method for optimizing the low-energy background of the SNIP gamma spectrum according to claim 3, characterized in that: The sliding window iteration is described in detail below: v p (i) = max{ v p-1 (i) , [ v p-1 (i+p) + v p-1 (i-p) ] / 2} Among them, v p (i) and v p-1 (i) are the mirror domain background estimates for the p-th and p-1-th iterations, respectively, where p = 1, 2, ..., m, and m is the preset iteration number, v0(i) = v(i); v p-1 (ip) and v p-1 (i+p) are the mirror domain background estimates corresponding to the left boundary ip and right boundary i+p of the computation window centered at i and with radius p in the (p-1)th iteration, respectively; max{} is the maximum value.

5. The method for optimizing the low-energy background of SNIP gamma spectrum according to claim 2, characterized in that: In step S3, the reverse mirror restoration operation is specifically a vertical mirror inverse transformation, as follows: b(i) = y max - b udr (i) Where b(i) is the optimized background estimate of the original gamma spectrum; b udr (i) is the estimated background vector of the mirror domain after the log-log-square inverse transform.

6. The method for optimizing the low-energy background of the SNIP gamma spectrum according to claim 1, characterized in that: The original gamma spectrum is a low-energy gamma spectrum with energies below 100 keV.

7. An electronic device, characterized in that, include: A memory and a processor are coupled to each other, wherein the memory stores program data, and the processor invokes the program data to perform the method as described in any one of claims 1-6.

8. A computer-readable storage medium storing program data thereon, characterized in that, When the program data is executed by the processor, it implements the method as described in any one of claims 1-6.

Images