A single photon detection method and system for weak PMT single photon event confirmation

Abstract

The application discloses a single-photon detection method and system for weak PMT single-photon event confirmation, and particularly relates to the technical field of single-photon detection, which comprises the following steps: acquiring a continuous electric pulse signal output by an electric sensor, searching for a wave peak point and a wave trough point, determining a signal segment between two adjacent wave trough points and higher than the electric level of the two wave trough points as a candidate pulse, extracting a time of arrival, a peak amplitude, a rising section slope, a half-peak width and a peak recovery difference value, and outputting a candidate pulse sequence; the candidate pulse is extracted from a continuous electric pulse signal output by a PMT, a previous continuation graph is constructed in combination with a difference relationship between a current candidate pulse and a previous confirmed single-photon event, a candidate explanation path is generated, a counterfactual elimination is performed, and an independent origin path determination is executed, so as to distinguish a new photon trigger from a previous event continuation pulse.

Description

Technical Field

[0001] This invention relates to the field of single-photon detection technology, and more specifically, to a single-photon detection method and system for confirming weak PMT single-photon events. Background Technology

[0002] In single-photon detection technology based on photomultiplier tubes, the mainstream approach in the industry is to identify single-photon events from the output electrical signal of the PMT under weak light input conditions. This usually involves amplifying, shaping, thresholding, integrating, matching templates, or filtering time windows on the pulse signal output from the PMT anode, and outputting pulses that meet the conditions as single-photon events.

[0003] For example, in time-resolved fluorescence measurement or weak scintillator readout scenarios, PMT needs to work for a long time under extremely weak incident, continuous acquisition and high gain readout conditions. At the same time, it is subject to hard constraints such as the continuous existence of dark count, the possibility of back pulse and reflected pulse caused by the pre-multiplication process, the existence of ringing response in the front-end amplification chain, and the requirement that the system cannot rely on additional reference optical paths or redundant detectors.

[0004] Under this constraint, the mainstream approach will consistently reveal an observable and verifiable defect: in the output pulses identified as new single-photon pulses, subsequent pulses formed by the continuation of previously confirmed events in the multiplication chain or readout chain will be continuously mixed in. This manifests as an abnormal increase in the counting result as the distribution of preceding events changes, significant fluctuations in the number of new events under different discrimination thresholds, and difficulty in distinguishing some candidate pulses that are highly close to preceding events in time position. The reason for this is that the existing methods mainly perform discrimination based on the amplitude, width, or waveform similarity of the current pulse itself, and cannot determine whether the pulse is a new photocathode trigger or a continuation of a preceding event.

[0005] The technical problem to be solved by this application is: how to determine whether a candidate pulse of PMT output has the qualification of origin independent of the preceding event without relying on an additional reference detector structure, so as to achieve accurate confirmation of weak PMT single-photon events. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a single-photon detection method and system for confirming weak PMT single-photon events. This method extracts candidate pulses from the continuous electrical pulse signal output by the PMT, constructs a precedent continuation graph by combining the difference between the current candidate pulse and a previously confirmed single-photon event, generates candidate interpretation paths, performs counterfactual elimination, and determines independent origin paths. This distinguishes between new photon triggers and preceding event continuation pulses, thereby solving the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a single-photon detection method for confirming weak PMT single-photon events, comprising:

[0008] S1. Acquire the continuous electrical pulse signal output by the electrical sensor, search for peak points and trough points, identify the signal segment between two adjacent trough points where the peak level is higher than the level of the two trough points as candidate pulses, and extract the arrival time, peak amplitude, rising slope, half-peak width and post-peak recovery difference, and output the candidate pulse sequence.

[0009] S2. For the current candidate pulse, read the previously confirmed single-photon events, calculate the time interval, peak amplitude difference, half-peak width difference, and recovery difference between each confirmed single-photon event and the current candidate pulse, and construct a preceding continuation graph with the preceding events as nodes and the corresponding difference groups as boundary values.

[0010] S3. Generate single-event interpretation paths for each node in the preceding continuation graph, generate joint interpretation paths for time-adjacent nodes, and determine the paths that do not contain any node as empty interpretation paths, and output the candidate path set.

[0011] S4. Perform counterfactual elimination operation on each candidate path, remove the corresponding preceding events in the path one by one, and use the sum of the absolute values ​​of the time interval before and after removal, the peak amplitude difference, the half-peak width difference, and the recovery difference as the explanation residual. The candidate path that ranks first and corresponds to an empty explanation path is taken as the independent origin path.

[0012] S5. Based on the independent origin path, write the current candidate pulse into the confirmed single-photon event sequence or remove it as a continuation pulse of the preceding event, and output the single-photon event confirmation result.

[0013] In a preferred embodiment, S1 includes:

[0014] S1-1. Calculate the level difference between adjacent sampling points for a continuous electrical pulse signal according to the sampling order. Determine the sampling point where the level difference changes from positive to negative as the peak point and the sampling point where the level difference changes from negative to positive as the trough point. Output the peak point sequence and trough point sequence in time order.

[0015] S1-2. For each peak in the peak sequence, read the two valleys located on both sides of the peak and adjacent to the peak in time. Determine the signal segment between the two valleys as the candidate signal segment, and calculate the front peak-valley difference and the rear peak-valley difference between the peak level and the two valley levels.

[0016] S1-3. When the difference between the peak and valley on the front side and the difference between the peak and valley on the back side are both greater than zero, the candidate signal segment is determined as the candidate pulse. Based on the peak position, peak level, half-peak level corresponding boundary point and post-peak end level in the candidate signal segment, the arrival time, peak amplitude, rising slope, half-peak width and post-peak recovery difference are calculated, and the candidate pulse sequence is output.

[0017] In a preferred embodiment, S2 includes:

[0018] S2-1. For the current candidate pulse, read the previously confirmed single-photon events, calculate the time interval, peak amplitude difference, half-peak width difference, and recovery difference between each confirmed single-photon event and the current candidate pulse, and form an event difference vector set by combining the four differences corresponding to each confirmed single-photon event.

[0019] S2-2. Input the event difference vector set into the cost solution process. For each confirmed single-photon event, calculate the single-event interpretation cost consisting of time interval, peak amplitude difference, half-peak width difference, and recovery difference. For multiple confirmed single-photon events arranged in time adjacent order, calculate the joint interpretation cost formed by the sum of the corresponding single-event interpretation costs. Output the single-event interpretation results and the joint interpretation results.

[0020] S2-3. Input the single-event interpretation results and the joint interpretation results into the consistency verification process. Calculate the cost offset between the single-event interpretation cost and the cost of participating in the joint interpretation for each confirmed single-photon event. Delete confirmed single-photon events whose cost offset exceeds the corresponding single-event interpretation cost from the joint interpretation results and output the correction event set.

[0021] In a preferred embodiment, S2 further includes:

[0022] S2-4. Input the set of corrected events into the confidence update process. Calculate the event continuation confidence value for each confirmed single-photon event based on the single-event interpretation cost, joint interpretation cost, and cost offset. Sort each confirmed single-photon event in descending order according to the event continuation confidence value and output the preceding event sorting set.

[0023] S2-5. Using each confirmed single-photon event in the prior event sorting set as a node, construct a prior continuation graph with the corresponding time interval, peak amplitude difference, half-peak width difference, recovery difference, and event continuation confidence value as boundary values. Stop writing new nodes when the event continuation confidence value corresponding to a new node is lower than the event continuation confidence value corresponding to the previous node, and output the prior continuation graph corresponding to the current candidate pulse.

[0024] In a preferred embodiment, S3 includes:

[0025] S3-1. Read each node and edge value in the preceding continuation graph, generate a single event interpretation path for each node, and use the time interval, peak amplitude difference, half-peak width difference and recovery difference corresponding to the node as the path difference group of the single event interpretation path.

[0026] S3-2. Perform path splicing on multiple nodes that are adjacent in time order in the preceding continuation graph to generate a joint interpretation path containing multiple nodes, and combine the edge values ​​corresponding to each node in the order of the nodes to form a combined difference group of the joint interpretation path.

[0027] S3-3. Determine the path that does not contain any node as an empty interpretation path, and merge each single-event interpretation path, each joint interpretation path and the empty interpretation path to output a candidate path set.

[0028] In a preferred embodiment, S4 includes:

[0029] S4-1. Read each candidate path in the candidate path set, perform counterfactual removal on each preceding event in each candidate path, recalculate the time interval, peak amplitude difference, half-peak width difference and recovery difference after removal, and output the sequence of the difference group after removal for each candidate path.

[0030] S4-2. Match the difference group sequence after removal of each candidate path with the difference group before removal item by item. For each candidate path, calculate the sum of the absolute values ​​of the time interval change, peak amplitude difference change, half-peak width difference change and recovery difference change, and use it as the path residual of the candidate path. Output the path residual sequence.

[0031] In a preferred embodiment, S4 further includes:

[0032] S4-3. Sort the path residual sequence in ascending order of value. Read the overlap of the corresponding preceding events between two adjacent candidate paths. Remove the candidate paths with overlapping preceding events and path residual values ​​higher than the adjacent paths from the candidate path set. Re-sort the candidate path set after deletion in ascending order of value until two adjacent candidate paths no longer contain the same preceding event. Output the converged path sequence.

[0033] S4-4. Read the sequence of removed difference groups corresponding to each candidate path in the convergence path sequence. For each candidate path, count the number of items with zero values ​​among the time interval change, peak amplitude difference change, half-peak width difference change, and recovery difference change. Write the item count into the corresponding candidate path to form a consistent item count sequence.

[0034] S4-5. Perform joint sorting on the converged path sequence and the sequence of consistent items in ascending order of path residual value and descending order of consistent item number. Take the candidate path that is first in the sort and has an empty explanation path as the independent origin path.

[0035] In a preferred embodiment, S5 includes:

[0036] S5-1. Read the independent origin path and the current candidate pulse. When the independent origin path corresponds to an empty interpretation path, write the arrival time, peak amplitude, rising slope, half-peak width, and post-peak recovery difference of the current candidate pulse to the end of the confirmed single-photon event sequence and generate a confirmation write mark. When the independent origin path does not correspond to an empty interpretation path, generate a continuation removal mark for the current candidate pulse.

[0037] In a preferred embodiment, S5 further includes:

[0038] S5-2. Read the preceding events in the independent origin path. When the current candidate pulse has a continuation elimination mark, write the time interval, peak amplitude difference, half-peak width difference and recovery difference between the current candidate pulse and the preceding event into the continuation record bit of the preceding event, delete the current candidate pulse from the new event count bit, and output the continuation elimination result.

[0039] S5-3. Merge the confirmation write mark and the continued elimination results to generate the single-photon event confirmation result corresponding to the current sampling position, and write it into the single-photon event confirmation result sequence in chronological order.

[0040] A single-photon detection system for confirming weak PMT single-photon events, the system comprising a pulse extraction module, a mapping module, a path generation module, a judgment module, and a result confirmation module:

[0041] The pulse extraction module is used to acquire continuous electrical pulse signals output by electrical sensors, search for peak points and trough points, identify signal segments between two adjacent trough points whose peak level is higher than the levels of the two trough points as candidate pulses, and extract arrival time, peak amplitude, rising slope, half-peak width and post-peak recovery difference, and output candidate pulse sequence.

[0042] The mapping module is used to read the previously confirmed single-photon events for the current candidate pulse, calculate the time interval, peak amplitude difference, half-peak width difference and recovery difference between each confirmed single-photon event and the current candidate pulse, and construct a preceding continuation graph with the preceding events as nodes and the corresponding difference groups as boundary values.

[0043] The path generation module is used to generate single-event interpretation paths for each node in the preceding continuation graph, generate joint interpretation paths for time-adjacent nodes, and determine paths that do not contain any nodes as empty interpretation paths, and output a set of candidate paths.

[0044] The determination module is used to perform counterfactual elimination operation on each candidate path, remove the corresponding preceding events in the path one by one, and use the sum of the absolute values ​​of the time interval before and after removal, the peak amplitude difference, the half-peak width difference, and the recovery difference as the explanation residual, and take the candidate path that ranks first and has an empty explanation path as the independent origin path.

[0045] The result confirmation module is used to write the current candidate pulse into the confirmed single-photon event sequence or remove it as a continuation pulse of the preceding event based on the independent origin path, and output the single-photon event confirmation result.

[0046] The technical effects and advantages of this invention are as follows:

[0047] 1. This scheme calculates the time interval, peak amplitude difference, half-peak width difference, and recovery difference between the current candidate pulse and the previously confirmed single-photon event, and determines the independent origin path by combining the empty interpretation path. This can distinguish the new photon trigger from the continuation products of the previous event, thereby relatively suppressing the false count phenomenon caused by the previous event.

[0048] 2. Perform unified boundary extraction on the peak points, trough points and candidate signal segments in the continuous electrical pulse signal, and output the arrival time, peak amplitude, rising slope, half-peak width and post-peak recovery difference according to a fixed caliber, so that the field composition of the candidate pulse is consistent, thereby providing repeatable input for subsequent difference calculation and path determination;

[0049] 3. By constructing a pre-sequence continuation graph with preceding events as nodes and four differences and event continuation confidence values ​​as boundary values, and generating single-event interpretation paths, joint interpretation paths, and empty interpretation paths within the graph structure, the explanatory relationship between preceding events and the current candidate pulse is expressed in a structured way, thereby relatively improving the decision organization method under complex preceding influences.

[0050] 4. Counterfactual removal is performed on each candidate path one by one, and the path residual is formed by the sum of the absolute values ​​of the changes in the four differences before and after removal. This allows for comparison of the influence of different interpretation paths on the current candidate pulse under a unified numerical standard, thus providing a verifiable calculation basis for the selection of independent origin paths.

[0051] 5. By deleting candidate paths with overlapping preceding events and high path residuals, and performing joint sorting on the converged path sequence in conjunction with the number of consistent items, the interference caused by multiple paths repeatedly occupying the same preceding event can be reduced, thereby keeping the selection results of independent origin paths relatively stable. Attached Figure Description

[0052] Figure 1 This is a flowchart outlining the method steps of the present invention;

[0053] Figure 2 This is a schematic diagram of the system module structure of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] Refer to the instruction manual appendix Figure 1-2 The present invention provides a single-photon detection method for confirming weak PMT single-photon events, comprising:

[0056] S1. Acquire the continuous electrical pulse signal output by the electrical sensor, search for peak points and trough points, identify the signal segment between two adjacent trough points where the peak level is higher than the level of the two trough points as candidate pulses, and extract the arrival time, peak amplitude, rising slope, half-peak width and post-peak recovery difference, and output the candidate pulse sequence.

[0057] The purpose of S1 in this implementation is to form a candidate pulse sequence with clear boundaries, complete fields, and direct readability for subsequent prequence continuation diagram construction from the continuous electrical pulse signal output by the electrical sensor. The principle is to first identify peaks and troughs based on the level changes of adjacent sampling points, then define candidate signal segments by two time-adjacent troughs on either side of the peak, and finally calculate the arrival time, peak amplitude, rising slope, half-peak width, and post-peak recovery difference within the candidate signal segments using a unified value caliber, thereby converting the original continuous sampling level sequence into a structured pulse object sequence. Each sampling point in the continuous electrical pulse signal includes at least a sampling number, sampling time, and level value. The sampling time is obtained by converting the sampling number and sampling period, and the sampling period is taken from a fixed configuration of the acquisition device. This implementation process includes the following steps:

[0058] The purpose of S1-1 is to first obtain the time-unique peak and trough points from the continuous electrical pulse signal, providing a boundary reference for subsequent candidate signal segment extraction; the input is the sampling point sequence corresponding to the continuous electrical pulse signal; the processing actions include: reading the level values ​​of two adjacent sampling points in ascending order of sampling number, calculating the level value of the later sampling point minus the level value of the previous sampling point to form a level difference sequence; when the level difference between two adjacent sampling points changes from positive to negative, the sampling point at the corresponding turning point is determined as the peak point; when the level difference between two adjacent sampling points changes from negative to positive, the sampling point at the corresponding turning point is determined as the trough point. Valley points; when multiple consecutive sampling points have the same level to form a plateau region, if the level difference before the plateau region is positive and the level difference after the plateau region is negative, then the sampling point corresponding to the midpoint of the plateau region time is determined as the peak point; if the level difference before the plateau region is negative and the level difference after the plateau region is positive, then the sampling point corresponding to the midpoint of the plateau region time is determined as the valley point; the output is a peak point sequence and a valley point sequence arranged in chronological order, and they are written into the peak-valley index area for S1-2 to read; when the number of sampling points of the continuous electrical pulse signal is less than three, peak-valley recognition is not performed, and empty peak point sequence and empty valley point sequence are directly output;

[0059] The purpose of S1-2 is to form candidate signal segments with closed boundaries centered on each peak point, and to exclude invalid segments that do not have a pulse fluctuation structure by using the difference between the peaks and valleys on both sides; the input quantities are the peak point sequence, the valley point sequence, and the original continuous electrical pulse signal; the processing actions include: reading the peak points one by one in chronological order, searching for the valley point that is chronologically earlier and closest to it in the valley point sequence for each peak point as the preceding valley point, and then searching for the valley point that is chronologically later and closest to it as the following valley point, and determining the continuous signal segment between the preceding valley point and the following valley point as the candidate signal segment. Select a signal segment; then calculate the peak level minus the previous trough level to obtain the previous peak-trough difference, and calculate the peak level minus the subsequent trough level to obtain the subsequent peak-trough difference; output the candidate signal segment and its corresponding previous and subsequent peak-trough differences, and write them to the candidate segment buffer for S1-3 to read; when a peak point lacks a previous trough point or a subsequent trough point, delete the record corresponding to that peak point and do not generate a candidate signal segment; when the previous or subsequent peak-trough difference is not greater than zero, retain the candidate signal segment number but write an invalid mark, and do not participate in the feature calculation of S1-3;

[0060] The purpose of S1-3 is to perform unified field extraction on candidate signal segments with complete fluctuation structures, forming candidate pulse sequences directly used in subsequent judgment steps; the input quantities are candidate signal segments in the candidate segment buffer, the peak-to-valley difference on the front side, and the peak-to-valley difference on the back side; the processing actions include: determining candidate signal segments with both the peak-to-valley difference on the front side and the peak-to-valley difference on the back side as candidate pulses; determining the peak point time in the candidate signal segment as the peak position, the peak point level as the peak level, and the median of the peak level and the trough level on the front side as the half-peak level; in the candidate signal segment In the rising segment, the first sampling point with a level not lower than half the peak level is searched as the half-peak front boundary point. In the falling segment of the candidate signal segment, the first sampling point with a level not higher than half the peak level is searched as the half-peak back boundary point. The arrival time of the half-peak front boundary point is determined. The peak level minus the level of the preceding trough point is determined as the peak amplitude. The rising segment slope is determined by dividing the peak level minus the level of the preceding trough point by the peak position time minus the time of the preceding trough point. The half-peak width is determined by subtracting the half-peak front boundary point from the time of the half-peak back boundary point. The peak recovery difference is determined by subtracting the level of the preceding trough point from the peak level.

[0061] Subsequently, the candidate pulse sequence is written in the order of fixed fields such as pulse number, arrival time, peak amplitude, rising slope, half-peak width, and post-peak recovery difference; the output is the candidate pulse sequence and written to the subsequent S2 reading area; when there is no sampling point with a level not lower than half-peak level in the rising segment, the peak time is taken as the arrival time; when there is no sampling point with a level not higher than half-peak level in the falling segment, the trough time is taken as the post-half-peak boundary point, thus ensuring that the fields of each valid candidate pulse are completely output.

[0062] This implementation method converts a continuous sampling level sequence into a candidate pulse sequence with clear boundaries, unique time sequence, and closed field composition by first determining peaks and valleys, then signal segments, and finally characteristic fields. This solves the problems of non-unique peak plateau regions, non-closed candidate signal segment boundaries, and inconsistent values ​​for arrival time and half-peak width. It also provides a consistent input for the unified calculation of subsequent time intervals, peak amplitude differences, half-peak width differences, and recovery difference differences.

[0063] In practical applications: When the PMT outputs a continuous sampling signal in a weak fluorescence measurement scenario, the acquisition device first generates a sampling point sequence according to a fixed sampling period, and then identifies the peak and trough points according to this implementation method. Invalid segments that lack bilateral trough points or whose bilateral peak-trough differences do not meet the conditions are removed. Subsequently, the arrival time, peak amplitude, rising slope, half-peak width, and post-peak recovery difference are uniformly calculated for the retained candidate signal segments, and candidate pulse sequences are formed in chronological order for direct use in subsequent pre-sequence continuation graph construction and independent origin path determination.

[0064] S2. For the current candidate pulse, read the previously confirmed single-photon events, calculate the time interval, peak amplitude difference, half-peak width difference, and recovery difference between each confirmed single-photon event and the current candidate pulse, and construct a preceding continuation graph with the preceding events as nodes and the corresponding difference groups as boundary values.

[0065] The purpose of S2 in this implementation is as follows: For the current candidate pulse, pre-order events with a possible continuation relationship are screened from the confirmed single-photon event sequence, and discrete event records are converted into a pre-order continuation graph that can be directly read for subsequent path generation. The principle is to first uniformly calculate the four differences between the current candidate pulse and each pre-order event, then use the four differences to construct the single-event interpretation cost and joint interpretation cost, delete unstable joint interpretation results through consistency verification, then update and sort the event continuation confidence values ​​of the retained events, and finally generate a pre-order continuation graph with node and boundary value closure according to a unified writing rule; wherein, each event in the confirmed single-photon event sequence includes at least an event number, arrival time, peak amplitude, half-peak width, post-peak recovery difference, and continuation record bit, and the current candidate pulse includes at least a pulse number, arrival time, peak amplitude, half-peak width, and post-peak recovery difference. The calculation direction of the time interval, peak amplitude difference, half-peak width difference, and recovery difference difference is uniformly reduced by the field corresponding to the current candidate pulse minus the field corresponding to the pre-order event; the implementation process includes the following steps:

[0066] The purpose of S2-1 is to first convert the current candidate pulse and each preceding event into a common-caliber difference object, providing a unified input for subsequent cost calculation. The input consists of the current candidate pulse and the sequence of confirmed single-photon events. The processing actions include: reading confirmed single-photon events whose arrival time is earlier than the current candidate pulse in reverse order of arrival time; stopping reading when the number of reads reaches the reading limit or the time interval exceeds the analysis duration, where the reading limit and analysis duration are both taken from the acquisition equipment configuration table; for each confirmed single-photon event read, calculating the arrival time of the current candidate pulse minus the formation time interval of the confirmed single-photon event, and calculating the peak amplitude of the current candidate pulse minus the confirmed single-photon event. The peak amplitude of a single-photon event forms the peak amplitude difference. The half-peak width difference is calculated by subtracting the half-peak width of the confirmed single-photon event from the half-peak width of the current candidate pulse. The recovery difference difference is calculated by subtracting the recovery difference of the confirmed single-photon event from the recovery difference of the current candidate pulse. Event difference vectors are formed in a fixed order of time interval, peak amplitude difference, half-peak width difference, and recovery difference difference. The output is a set of event difference vectors, and the event number and the corresponding event difference vector are written into the difference buffer for S2-2 to read. When the confirmed single-photon event sequence is empty, an empty event difference vector set is output and an empty marker is written, so that a prequence continuation graph containing only empty interpretation paths can be directly generated later.

[0067] The purpose of S2-2 is to convert the four differences into comparable interpretation costs and simultaneously generate single-event interpretation results and joint interpretation results. The input is the event difference vector set in the difference buffer. The processing actions include: for each confirmed single-photon event, taking the absolute values ​​of the time interval, peak amplitude difference, half-peak width difference, and recovery difference and summing them to obtain the single-event interpretation cost of the confirmed single-photon event; arranging the read confirmed single-photon events in ascending order of arrival time, forming joint groups for two or three consecutive nodes, and directly accumulating the single-event interpretation costs contained in each joint group to obtain the joint interpretation cost of the joint group; then combining each single-event interpretation cost with the corresponding event number to form a single-event interpretation result, and combining each joint interpretation cost with the corresponding set of joint group node numbers to form a joint interpretation result; the output is the single-event interpretation result and the joint interpretation result, which are written to the cost result area for S2-3 to read; when the event difference vector set is empty, no single-event interpretation result or joint interpretation result is generated, only an empty result marker is retained.

[0068] The purpose of S2-3 is to remove preceding events that cause imbalances between individual and joint interpretations, preventing unstable nodes from entering subsequent mapping. The input consists of single-event interpretation results and joint interpretation results in the cost result area. The processing includes: for each confirmed single-photon event, retrieving all joint interpretation results it participates in, and subtracting the single-event interpretation cost of the confirmed single-photon event from the corresponding joint interpretation cost to form a cost offset; when a cost offset is greater than the corresponding single-event interpretation cost, the confirmed single-photon event is deleted from the corresponding joint interpretation result; after deletion, if the remaining node numbers are still continuous, the reconstructed joint interpretation result is retained; if the remaining node numbers are split into multiple continuous segments, multiple joint interpretation results are reconstructed according to the continuous segments; if only one node remains after deletion, the result is converted to a single-event interpretation result; the output is a correction event set, which includes at least the retained event number, the corresponding single-event interpretation cost, the participating joint interpretation cost, and the cost offset, and is written to the correction area for S2-4 to read; when a confirmed single-photon event does not participate in any joint interpretation result, its cost offset is recorded as zero and it is directly retained.

[0069] The purpose of S2-4 is to establish a unified sorting basis for the corrected preceding events, ensuring a unique writing order for subsequent graph nodes. The input is the set of corrected events in the correction area. The processing actions include: reading the single-event interpretation cost, joint interpretation cost, and cost offset for each confirmed single-photon event, calculating the event continuation confidence value by dividing 1 by (1 plus single-event interpretation cost plus joint interpretation cost plus cost offset), where the joint interpretation cost of confirmed single-photon events that do not participate in the joint interpretation result is counted as zero; then, arranging each confirmed single-photon event in descending order of event continuation confidence value, and when the event continuation confidence values ​​are the same, performing parallel resolution in ascending order of time interval, ascending order of absolute value of peak amplitude difference, and ascending order of event number; the output is the preceding event sorting set, and writing the event number, corresponding four differences, and event continuation confidence value into the sorting area for S2-5 to read in sorting order; when the corrected event set is empty, an empty preceding event sorting set is output.

[0070] The purpose of S2-5 is to write the sorted preceding events into a closed preceding continuation graph with nodes and boundary values, which can be directly called for subsequent candidate path generation. The input is the sorted set of preceding events in the sorting region and the current candidate pulse. The processing actions include: reading the confirmed single-photon events one by one in the order of the sorted preceding events, writing each confirmed single-photon event as a graph node, and writing the time interval, peak amplitude difference, half-peak width difference, recovery difference, and event continuation confidence value corresponding to the node as the boundary values ​​pointing to the current candidate pulse in a fixed field order; after writing the first node, continue reading. Take the next node and compare the event continuation confidence value corresponding to the newly added node with the event continuation confidence value corresponding to the previous node. If they are equal, continue writing; if they are not equal and the joint interpretation cost after the addition is greater than the joint interpretation cost before writing, stop writing the new node; if the joint interpretation cost after the addition is not greater than the joint interpretation cost before writing, continue writing the next node; the output is the preceding continuation graph corresponding to the current candidate pulse, and it is written to the path generation area for S3 to read; when the preceding event sorting set is empty, output the preceding continuation graph of empty nodes and write an empty graph mark in the graph type field;

[0071] This implementation solves the problems of unclear preceding event reading range, inconsistent difference signs, suspended cost process, inability to connect after deletion of joint interpretation results, and unclear writing boundaries of preceding continuation graph nodes by unifying the difference direction, fixing the cost calculation caliber, supplementing the joint interpretation result deletion and reconstruction rules, and providing a direct calculation method for event continuation confidence value. It also supplements the sorting parallel resolution and node stopping writing rules. This enables S2 to stably output a preceding continuation graph with closed fields, unique node order, and clear edge value source.

[0072] In practical applications: When the arrival time of the current candidate pulse is t8, and events e5, e6, and e7 have been confirmed in the single-photon event sequence, the system first reads e5, e6, and e7 within the analysis time according to the configuration table, calculates the time interval, peak amplitude difference, half-peak width difference, and recovery difference between t8 and e5, e6, and e7 respectively, forming an event difference vector set; then, it obtains the interpretation cost of each single event by summing absolute values, and forms a joint interpretation cost for time-continuous e5-e6, e6-e7, and e5-e6-e7; subsequently, it deletes nodes with unbalanced cost offsets, updates the event continuation confidence value by reciprocal formula, and completes the sorting; finally, it writes the preceding events that are ranked first and still meet the condition that the joint interpretation cost does not increase after being added into the preceding continuation graph, so that they can be directly read by the subsequent generation of single-event interpretation paths, joint interpretation paths, and empty interpretation paths.

[0073] S3. Generate single-event interpretation paths for each node in the preceding continuation graph, generate joint interpretation paths for time-adjacent nodes, and determine the paths that do not contain any node as empty interpretation paths, and output the candidate path set.

[0074] The purpose of S3 in this implementation is to convert the pre-order event nodes that have been filtered and sorted in the pre-order continuation graph into a set of candidate paths that can be directly read by the subsequent counterfactual elimination operation. The principle is to first generate a single-event explanation path for each node, then perform path concatenation on multiple nodes with adjacent time sequences to form a joint explanation path, and finally add empty explanation paths that do not contain any nodes, thereby uniformly converting the graph structure object into a path structure object. Each node in the pre-order continuation graph includes at least an event number, and each edge value includes at least a time interval, peak amplitude difference, half-peak width difference, recovery difference, and event continuation confidence value. Candidate paths include at least a path number, path type, node set, and difference group sequence. This implementation process includes the following steps:

[0075] The purpose of S3-1 is to first convert each individual node in the preceding continuation graph into a single-event interpretation path with complete fields, providing a basic path object for subsequent joint path splicing; the input is each node and edge value in the preceding continuation graph; the processing actions include: reading nodes one by one in the order they are written in the preceding continuation graph, generating a single-event interpretation path for each node that only contains the event number of that node, forming a path difference group in a fixed order based on the time interval, peak amplitude difference, half-peak width difference, and recovery difference of the node pointing to the current candidate pulse, and writing the path type as a single-event interpretation path; then writing the path number incrementally in the order of path generation; the output is a set of single-event interpretation paths, and writing the path number, node set, path difference group, and path type into the path buffer for S3-2 and S3-3 to read; when the preceding continuation graph is empty, no single-event interpretation path is generated, only an empty graph marker is written;

[0076] The purpose of S3-2 is to combine multiple nodes with a temporally continuous relationship into a joint interpretation path, so that the joint interpretation relationship of multiple preceding events on the current candidate pulse can be expressed in a unified path structure. The input is the node order information in the path buffer and the edge values ​​of each node in the preceding continuation graph. The processing actions include: reading the nodes in the preceding continuation graph according to the node writing order, selecting two or three nodes with consecutive numbers to form a node group, performing path concatenation on each node group to generate a joint interpretation path containing multiple node event numbers; and then reading the time interval, peak amplitude difference, and half-peak value of each node according to the order of their positions in the preceding continuation graph. The width difference and recovery difference are combined in the same order to form a combined difference group for the joint interpretation path, and the path type is written as the joint interpretation path. Only node groups with consecutive node numbers are concatenated; node groups with discontinuous numbers are not concatenated. The output is a set of joint interpretation paths, and the path number, node set, combined difference group, and path type are written to the path buffer for S3-3 to read. When the number of nodes in the current sequence continuation graph is less than two, no joint interpretation path is generated; when the number of nodes is equal to two, only a two-node joint interpretation path is generated; when the number of nodes is greater than two, both a two-node joint interpretation path and a three-node joint interpretation path are generated simultaneously.

[0077] The purpose of S3-3 is to: fill in empty interpretation paths that indicate no preceding events involved in the interpretation, and merge the three types of paths into a candidate path set for subsequent unified processing; the input is the single-event interpretation path set and the joint interpretation path set in the path buffer; the processing actions include: generating an empty interpretation path with an empty node set, an empty difference group sequence, and an empty interpretation path type, and assigning an unused path number to this empty interpretation path; then merging the single-event interpretation path set, the joint interpretation path set, and the empty interpretation path in ascending order of path number to form a candidate path set; after merging, uniformly writing the path number, path type, node set, and difference group sequence fields to each path in the candidate path set; the output is the candidate path set, which is written to the counterfactual operation area for S4 to read; when there are no single-event interpretation paths and joint interpretation paths in the path buffer, only the candidate path set containing empty interpretation paths is output;

[0078] This implementation method first generates single-event interpretation paths, then generates joint interpretation paths, and finally fills in empty interpretation paths. This method stably converts the nodes and edge values ​​in the preceding continuation graph into a set of candidate paths with unified fields, clear types, and unique order. This solves the problems of unclosed path object fields, unclear splicing range of joint paths, and lack of independent structure in empty interpretation paths. This allows subsequent counterfactual removal operations to process different types of candidate paths one by one with the same reading caliber.

[0079] In practical applications: When the preceding continuation graph corresponding to a current candidate pulse contains three nodes e5, e6, and e7, the system first generates three single-event interpretation paths containing e5, e6, and e7 respectively. Then, it generates three joint interpretation paths e5-e6, e6-e7, and e5-e6-e7 in node order. Subsequently, an empty interpretation path without any nodes is added, and the above paths are merged into a candidate path set and written into the counterfactual operation area for subsequent execution of preceding event removal and interpretation residual calculation.

[0080] S4. Perform counterfactual elimination operation on each candidate path, remove the corresponding preceding events in the path one by one, and use the sum of the absolute values ​​of the time interval before and after removal, the peak amplitude difference, the half-peak width difference, and the recovery difference as the explanation residual. The candidate path that ranks first and corresponds to an empty explanation path is taken as the independent origin path.

[0081] The purpose of S4 in this implementation is to perform counterfactual removal and difference recalculation on each candidate path in the candidate path set, eliminate the impact of duplicate interpretation caused by common preceding events, and determine independent origin paths under a unified residual caliber. The principle is to first remove each preceding event in each candidate path and recalculate the difference group after removal; then, to form path residuals by matching the removed difference group with the original difference group item by item; subsequently, to delete candidate paths containing duplicate preceding events and with large residuals; to continue counting the number of zero-change items in the four change quantities for each candidate path; and to perform joint sorting according to the path residuals and the number of consistent items, thereby outputting the independent origin paths corresponding to the empty interpretation paths. A candidate path includes at least a path number, path type, node set, and difference group sequence. The path type is limited to single-event interpretation paths, joint interpretation paths, and empty interpretation paths. Precedence event overlap is limited to the existence of the same event number in the node sets of two candidate paths. This implementation process includes the following steps:

[0082] The purpose of S4-1 is to construct a counterfactual state for each candidate path after removing preceding events, forming a sequence of post-removal difference sets required for subsequent residual calculations. The inputs are the candidate path set, the current candidate pulse, and the confirmed single-photon event sequence. The processing actions include: reading candidate paths sequentially in ascending order of path number; performing counterfactual removal on each preceding event in each candidate path; when a candidate path is a single-event interpretation path, after removing the unique preceding event, setting the time interval, peak amplitude difference, half-peak width difference, and recovery difference to zero, forming a set of post-removal difference sets; when a candidate path is a joint interpretation path, after removing one of the preceding events... For each preceding event, the time interval, peak amplitude difference, half-peak width difference, and recovery difference are recalculated between the remaining preceding events and the current candidate pulse, and a set of removed difference groups is formed according to the node order. When the candidate path is an empty interpretation path, the preceding event removal is not performed, and a set of removed difference groups with all four values ​​being zero is directly generated. The output is the sequence of removed difference groups corresponding to each candidate path, and the path number and the corresponding sequence of removed difference groups are written to the counterfactual result area for S4-2 to read. When a joint interpretation path removes a certain preceding event and no longer contains any remaining preceding events, the removal result is processed as an empty node state and a zero-value difference group is output.

[0083] The purpose of S4-2 is to uniformly compress the difference changes before and after removal into a single scalar residual, enabling different types of candidate paths to be compared within the same numerical space. The input consists of the sequence of difference groups before removal and the sequence of difference groups after removal in the counterfactual result area of ​​the candidate path set. The processing actions include: reading the difference groups before removal and the corresponding difference groups after removal for each candidate path, calculating the time interval change, peak amplitude difference change, half-peak width difference change, and recovery difference change respectively, where each change is calculated by subtracting the value before removal from the value after removal and taking the absolute value; then summing the four changes of each group of removal results to obtain the group residual of the group of removal results, and then accumulating all group residuals under the same path to obtain the path residual of the candidate path; the output is the path residual sequence, and the path number, path type, node set, path residual, and the sequence of difference groups after removal are written together into the residual sorting area for S4-3 to read; when the candidate path is an empty interpretation path, its path residual is directly recorded as zero.

[0084] The purpose of S4-3 is to: delete candidate paths with overlapping interpretation ranges and high residuals, and retain the convergent path sequence that does not involve duplicate usage; the input is the path residual sequence in the residual sorting region and the node set of each candidate path; the processing actions include: first, sorting all candidate paths in ascending order of path residual values; then, reading adjacent candidate paths one by one in the sorting order, and checking whether the node sets of the two candidate paths have the same event number; when the same event number exists, it is determined that the preceding events coincide, and the candidate path with the higher path residual value is deleted from the candidate path set; when the path residual values ​​of two candidate paths are the same, the nodes are compared. The system calculates the number of nodes and deletes candidate paths with a larger number of nodes. If the number of nodes is still the same, the system compares the path numbers and deletes the candidate path with the larger path number. After one round of deletion, the remaining candidate path set is re-sorted in ascending order of path residual values, and the above adjacent comparison and deletion process is repeated until the node sets of any two adjacent candidate paths do not have the same event number. The node set of an empty interpretation path is empty and does not participate in the determination of overlapping preceding events. The output is a converged path sequence, which is written to the consistency statistics area for S4-4 to read. When only one candidate path remains in the candidate path set, that candidate path is directly written as the converged path sequence.

[0085] The purpose of S4-4 is to supplement the consistency index of each candidate path in the converged path sequence, providing a second ranking basis for subsequent joint ranking. The input is the converged path sequence in the consistency statistics area and its corresponding sequence of difference groups after removal. The processing actions include: reading the time interval change, peak amplitude difference change, half-peak width difference change, and recovery difference change for each candidate path group by group, and determining whether the value of each change is zero; when the value of a change is zero, the item is recorded as a consistent item; then, the total number of consistent items in all removal results of the candidate path is counted to form the number of consistent items for the candidate path; the output is a sequence of consistent item counts, and the path number and the corresponding number of consistent items are written into the joint ranking area for S4-5 to read; when the candidate path is an empty interpretation path, its number of consistent items is generated by accumulating according to the rule that all four changes are zero.

[0086] The purpose of S4-5 is to determine the final sorting order using path residuals and the number of consistent items, and to identify independent origin paths. The input consists of the convergent path sequence and the sequence of consistent items in the joint sorting area. The processing actions include: first, sorting all candidate paths in ascending order of path residual values; then, sorting candidate paths with the same path residual values ​​in descending order of the number of consistent items; when the path residual values ​​and the number of consistent items are the same, performing parallel resolution in ascending order of node count and then path number; after sorting, reading candidate paths one by one starting from the first sorted path, retrieving their path types, and when the first candidate path with an empty explanation path type is read, determining that candidate path as an independent origin path; when there are no empty explanation paths in the sorting results, outputting an independent origin path missing flag; the output is the independent origin path, and the path number, path type, and sorting position are written into the confirmation decision area for S5 to read.

[0087] This implementation solves the problems of inconsistent generation of difference groups after removal, non-unique correspondence of path residuals, unclear deletion boundaries of overlapping preceding events, and unstable selection order of independent origin paths by supplementing the zero-value recalculation caliber after counterfactual removal, unifying the accumulation method of path residuals, clarifying the deletion rules of overlapping preceding events, providing a statistical method for the number of consistent items, and establishing joint sorting and parallel resolution rules for path residuals and the number of consistent items. This enables S4 to output a unique independent origin path under the same executable caliber.

[0088] In practical applications: When the candidate path set corresponding to a current candidate pulse contains single-event interpretation paths, two-node joint interpretation paths, three-node joint interpretation paths, and empty interpretation paths, the system first performs counterfactual removal on each preceding event in each path and recalculates the four differences. Then, it calculates the absolute value of each of the four differences before and after removal and accumulates them to generate path residuals. Subsequently, it deletes candidate paths that contain the same preceding events and have high residuals, retains the convergent path sequence that does not repeatedly occupy preceding events, counts the number of zero-value items in the four changes of each candidate path, and completes joint sorting by ascending path residuals and descending number of consistent items. Finally, it reads the first empty interpretation path from the sorting results, identifies it as an independent origin path, and writes it into the confirmation judgment area for subsequent confirmation writing or continued removal.

[0089] S5. Based on the independent origin path, write the current candidate pulse into the confirmed single-photon event sequence or remove it as a continuation pulse of the preceding event, and output the single-photon event confirmation result.

[0090] The purpose of S5 in this implementation is to: perform new event writing or continuation elimination on the current candidate pulse based on the independent origin path output by S4, and form a traceable single-photon event confirmation result. The principle is to first perform splitting processing on the current candidate pulse according to the path type of the independent origin path, then, in the continuation elimination scenario, write the difference relationship between the current candidate pulse and the corresponding preceding event back to the continuation record bit of the preceding event, and finally merge the confirmation writing result and the continuation elimination result into a single-photon event confirmation result sequence. Each event in the confirmed single-photon event sequence includes at least an event number, arrival time, peak amplitude, rising slope, half-peak width, post-peak recovery difference, and continuation record bit. The current candidate pulse includes at least a pulse number, arrival time, peak amplitude, rising slope, half-peak width, and post-peak recovery difference. The single-photon event confirmation result includes at least a candidate pulse number, confirmation status, independent origin path number, and a set of preceding event numbers. This implementation process includes the following steps:

[0091] The purpose of S5-1 is to: first determine the processing branch of the current candidate pulse based on the independent origin path, so that the new event writing and continuation rejection have a unique entry point; the input is the independent origin path and the current candidate pulse; the processing actions include: reading the path type of the independent origin path; when the path type is an empty interpretation path, generating a new event number according to the position of the tail of the confirmed single-photon event sequence, and writing the arrival time, peak amplitude, rising slope, half-peak width, and post-peak recovery difference of the current candidate pulse along with the event number to the tail of the confirmed single-photon event sequence in sequence, while writing the confirmation status of the corresponding candidate pulse as a new event and generating a confirmation write mark; when the path type is not an empty interpretation path, the current candidate pulse is not written to the confirmed single-photon event sequence, but a continuation rejection mark is written to the status field of the current candidate pulse; the output is a confirmation write mark or a continuation rejection mark, and the mark result is written to the result buffer for S5-2 and S5-3 to read; when the independent origin path is missing, the current candidate pulse is directly written to the manual review queue, and a review mark is written to the result buffer;

[0092] The purpose of S5-2 is to establish a write-back relationship between the current candidate pulse and the preceding events in the independent origin path in the continuation elimination scenario, thereby completing the registration of continuation information and the clearing of new event counts. The inputs are the continuation elimination flag, the independent origin path, the current candidate pulse, and the confirmed single-photon event sequence in the result buffer. The processing actions include: when the current candidate pulse has a continuation elimination flag, reading all preceding event numbers in the independent origin path; calculating the time interval between the arrival time of the current candidate pulse and the arrival time of the preceding event for each preceding event; calculating the peak amplitude difference between the peak amplitude of the current candidate pulse and the peak amplitude of the preceding event; calculating the half-peak width difference between the half-peak width of the current candidate pulse and the half-peak width of the preceding event; and calculating... The recovery difference is calculated by subtracting the recovery difference of the preceding event from the recovery difference of the current candidate pulse peak. Then, the current candidate pulse number, time interval, peak amplitude difference, half-peak width difference, and recovery difference are appended to the continuation record bit of the corresponding preceding event. The new event count bit corresponding to the current candidate pulse is changed from 1 to 0. The output is the continuation elimination result, and the current candidate pulse number, the corresponding preceding event number set, and the new event count bit are written to the result buffer for S5-3 to read. When an independent origin path contains multiple preceding events, the appending is performed on each preceding event one by one. When a preceding event fails to be retrieved in the confirmed single-photon event sequence, the preceding event is skipped, and a missing marker is written to the abnormal field of the continuation record bit.

[0093] The purpose of S5-3 is to unify the new event writing results and the continuation elimination results into a single-photon event confirmation result, forming a confirmation record that is traceable in chronological order. The inputs are the confirmation writing flag, the continuation elimination result, and the current candidate pulse in the result buffer. The processing actions include: when a confirmation writing flag exists, reading the current candidate pulse number, the corresponding new event number, and the independent origin path number, writing the confirmation status as a new event, and writing the previous event number set to an empty set; when a continuation elimination result exists, reading the current candidate pulse number, the independent origin path number, and the previous event number set, writing the confirmation status as a continuation pulse; then combining the candidate pulse number, confirmation status, independent origin path number, and previous event number set to form the single-photon event confirmation result corresponding to the current sampling position, and writing it into the single-photon event confirmation result sequence in ascending order of the arrival time of the current candidate pulse; the output is the single-photon event confirmation result sequence, which is written into the result output area for subsequent statistics, backhaul, or archiving; when a verification flag exists in the result buffer, writing the confirmation status of the current candidate pulse as pending verification, and storing it separately from the normal confirmation result.

[0094] This implementation method, by first splitting, then writing back, and then merging, completes the object writing position, field composition, and state transition relationship of the current candidate pulse in the new event writing scenario and the continuation elimination scenario. It solves the problems of unclear write-back caliber of continuation record bit, new event count bit, and single photon event confirmation result sequence, inability to synchronously register when the joint interpretation path corresponds to multiple preceding events, and no landing point of the result when the independent origin path is missing. It enables S5 to output single photon event confirmation results with complete fields, unique states, and traceability.

[0095] In practical applications: when the independent origin path corresponding to a current candidate pulse is an empty interpretation path, the system writes the current candidate pulse to the end of the confirmed single-photon event sequence and assigns a new event number, while generating a new event confirmation result; when the independent origin path corresponding to a current candidate pulse is a joint interpretation path, the system does not write the current candidate pulse to the confirmed single-photon event sequence, but instead writes the time interval, peak amplitude difference, half-peak width difference, and recovery difference between the current candidate pulse and each preceding event in the joint interpretation path into the continuation record bit of the corresponding preceding event, clears the new event count bit of the current candidate pulse to zero, and then generates a continuation pulse confirmation result for the current candidate pulse and writes it into the single-photon event confirmation result sequence in chronological order.

[0096] Furthermore, the present invention also includes a single-photon detection system for confirming weak PMT single-photon events, the system comprising a pulse extraction module, a mapping module, a path generation module, a judgment module, and a result confirmation module:

[0097] The pulse extraction module is used to acquire continuous electrical pulse signals output by electrical sensors, search for peak points and trough points, identify signal segments between two adjacent trough points whose peak level is higher than the levels of the two trough points as candidate pulses, and extract arrival time, peak amplitude, rising slope, half-peak width and post-peak recovery difference, and output candidate pulse sequence.

[0098] The mapping module is used to read the previously confirmed single-photon events for the current candidate pulse, calculate the time interval, peak amplitude difference, half-peak width difference and recovery difference between each confirmed single-photon event and the current candidate pulse, and construct a preceding continuation graph with the preceding events as nodes and the corresponding difference groups as boundary values.

[0099] The path generation module is used to generate single-event interpretation paths for each node in the preceding continuation graph, generate joint interpretation paths for time-adjacent nodes, and determine paths that do not contain any nodes as empty interpretation paths, and output a set of candidate paths.

[0100] The determination module is used to perform counterfactual elimination operation on each candidate path, remove the corresponding preceding events in the path one by one, and use the sum of the absolute values ​​of the time interval before and after removal, the peak amplitude difference, the half-peak width difference, and the recovery difference as the explanation residual, and take the candidate path that ranks first and has an empty explanation path as the independent origin path.

[0101] The result confirmation module is used to write the current candidate pulse into the confirmed single-photon event sequence or remove it as a continuation pulse of the preceding event based on the independent origin path, and output the single-photon event confirmation result.

[0102] The working principle of this scheme is as follows: First, the continuous electrical pulse signal output by the PMT as an electrical sensor is analyzed point by point to identify peaks and troughs. Candidate pulses are extracted from the effective signal segments between adjacent troughs, and the arrival time, peak amplitude, rise slope, half-peak width, and post-peak recovery difference are uniformly obtained. Then, using the current candidate pulse as a reference, the previously confirmed single-photon events are read, and the time interval, peak amplitude difference, half-peak width difference, and recovery difference between the two are calculated to construct a prequence continuation graph. Finally, a single-event interpretation path is generated from the prequence continuation graph, and the combined... For each candidate path, counterfactual removal and residual calculation are performed to eliminate paths that repeatedly occupy preceding events. The paths are sorted by residual and number of consistent items to determine whether there is an independent origin path corresponding to the empty explanation path. Finally, if an independent origin path exists, the current candidate pulse is confirmed as a new single-photon event and written into the confirmed single-photon event sequence. If no such path exists, it is removed as a continuation pulse of the preceding event and written back to the corresponding continuation record, thereby further distinguishing signals that look like pulses into new single-photon events or preceding continuation products.

[0103] In practical applications, such as in time-resolved fluorescence measurement scenarios, the sample emits extremely weak photons, and the PMT continuously outputs a large number of weak electrical pulses, which may include real single-photon responses, as well as dark counts, subsequent pulses, and continuation pulses left by preceding events. This scheme first divides the continuous waveform into candidate pulses, and then compares each candidate pulse with previously confirmed events to determine whether it is a new photon trigger or a continuation of a previous event. If a candidate pulse cannot be explained by any preceding event, it is recorded as a new single-photon event. If it forms a continuation relationship with a preceding event in terms of time position and pulse characteristics, it is not counted again, but is recorded as a continuation pulse of that preceding event. In this way, even in environments with weak light, high noise, and easily confused pulses, the system can still provide more stable confirmation results of real single-photon events.

[0104] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A single-photon detection method for confirming weak PMT single-photon events, characterized in that, include: S1. Acquire the continuous electrical pulse signal output by the electrical sensor, search for peak points and trough points, identify the signal segment between two adjacent trough points where the peak level is higher than the level of the two trough points as candidate pulses, and extract the arrival time, peak amplitude, rising slope, half-peak width and post-peak recovery difference, and output the candidate pulse sequence. S2. For the current candidate pulse, read the previously confirmed single-photon events, calculate the time interval, peak amplitude difference, half-peak width difference, and recovery difference between each confirmed single-photon event and the current candidate pulse, and construct a prequence continuation graph with preceding events as nodes and corresponding difference groups as boundary values; including: S2-1. For the current candidate pulse, read the previously confirmed single-photon events, calculate the time interval, peak amplitude difference, half-peak width difference, and recovery difference between each confirmed single-photon event and the current candidate pulse, and form an event difference vector set by combining the four differences corresponding to each confirmed single-photon event. S2-2. Input the event difference vector set into the cost solution process. For each confirmed single-photon event, calculate the single-event interpretation cost consisting of time interval, peak amplitude difference, half-peak width difference, and recovery difference. For multiple confirmed single-photon events arranged in time adjacent order, calculate the joint interpretation cost formed by the sum of the corresponding single-event interpretation costs. Output the single-event interpretation results and the joint interpretation results. S2-3. Input the single-event interpretation results and joint interpretation results into the consistency verification process. Calculate the cost offset between the single-event interpretation cost and the cost of participating in the joint interpretation for each confirmed single-photon event. Delete confirmed single-photon events whose cost offset exceeds the corresponding single-event interpretation cost from the joint interpretation results and output the correction event set. S2-4. Input the set of corrected events into the confidence update process. Calculate the event continuation confidence value for each confirmed single-photon event based on the single-event interpretation cost, joint interpretation cost, and cost offset. Sort each confirmed single-photon event in descending order according to the event continuation confidence value and output the preceding event sorting set. S2-5. Using each confirmed single-photon event in the prior event sorting set as a node, construct a prior continuation graph with the corresponding time interval, peak amplitude difference, half-peak width difference, recovery difference, and event continuation confidence value as boundary values. Stop writing new nodes when the event continuation confidence value corresponding to the newly added node is lower than the event continuation confidence value corresponding to the previous node, and output the prior continuation graph corresponding to the current candidate pulse. S3. Generate single-event interpretation paths for each node in the preceding continuation graph, generate joint interpretation paths for time-adjacent nodes, and determine the paths that do not contain any node as empty interpretation paths, and output the candidate path set. S4. Perform counterfactual elimination on each candidate path, removing the corresponding preceding events from each path one by one. Use the sum of the absolute values ​​of the time interval before and after removal, the difference in peak amplitude, the difference in half-peak width, and the difference in recovery value as the explanatory residual. The candidate path that ranks first and corresponds to an empty explanatory path is considered the independent origin path; including: S4-1. Read each candidate path in the candidate path set, perform counterfactual removal on each preceding event in each candidate path, recalculate the time interval, peak amplitude difference, half-peak width difference and recovery difference after removal, and output the sequence of the difference group after removal for each candidate path. S4-2. Match the difference group sequence after removal of each candidate path with the difference group before removal item by item. For each candidate path, calculate the sum of the absolute values ​​of the time interval change, peak amplitude difference change, half peak width difference change and recovery difference change, and use it as the path residual of the candidate path. Output the path residual sequence. S4-3. Sort the path residual sequence in ascending order of value. Read the overlap of the corresponding preceding events between two adjacent candidate paths. Remove the candidate paths with overlapping preceding events and path residual values ​​higher than the adjacent paths from the candidate path set. Re-sort the candidate path set after deletion in ascending order of value until two adjacent candidate paths no longer contain the same preceding event. Output the converged path sequence. S4-4. Read the sequence of removed difference groups corresponding to each candidate path in the convergence path sequence. For each candidate path, count the number of items with zero values ​​among the time interval change, peak amplitude difference change, half-peak width difference change, and recovery difference change. Write the item count into the corresponding candidate path to form a consistent item count sequence. S4-5. Perform joint sorting on the converged path sequence and the sequence of consistent items in ascending order of path residual value and descending order of consistent item number. Take the candidate path that is first in the sort and has an empty explanation path as the independent origin path. S5. Based on the independent origin path, write the current candidate pulse into the confirmed single-photon event sequence or remove it as a continuation pulse of the preceding event, and output the single-photon event confirmation result.

2. The single-photon detection method for confirming weak PMT single-photon events according to claim 1, characterized in that: S1 includes: S1-1. Calculate the level difference between adjacent sampling points for a continuous electrical pulse signal according to the sampling order. Determine the sampling point where the level difference changes from positive to negative as the peak point and the sampling point where the level difference changes from negative to positive as the trough point. Output the peak point sequence and trough point sequence in time order. S1-2. For each peak in the peak sequence, read the two valleys located on both sides of the peak and adjacent to the peak in time. Determine the signal segment between the two valleys as the candidate signal segment, and calculate the front peak-valley difference and the rear peak-valley difference between the peak level and the two valley levels. S1-3. When the difference between the peak and valley on the front side and the difference between the peak and valley on the back side are both greater than zero, the candidate signal segment is determined as the candidate pulse. Based on the peak position, peak level, half-peak level corresponding boundary point and post-peak end level in the candidate signal segment, the arrival time, peak amplitude, rising slope, half-peak width and post-peak recovery difference are calculated, and the candidate pulse sequence is output.

3. The single-photon detection method for confirming weak PMT single-photon events according to claim 1, characterized in that: S3 includes: S3-1. Read each node and edge value in the preceding continuation graph, generate a single event interpretation path for each node, and use the time interval, peak amplitude difference, half-peak width difference and recovery difference corresponding to the node as the path difference group of the single event interpretation path. S3-2. Perform path splicing on multiple nodes that are adjacent in time order in the preceding continuation graph to generate a joint interpretation path containing multiple nodes, and combine the edge values ​​corresponding to each node in the order of the nodes to form a combined difference group of the joint interpretation path. S3-3. Determine the path that does not contain any node as an empty interpretation path, and merge each single-event interpretation path, each joint interpretation path and the empty interpretation path to output a candidate path set.

4. The single-photon detection method for confirming weak PMT single-photon events according to claim 3, characterized in that: S5 includes: S5-1. Read the independent origin path and the current candidate pulse. When the independent origin path corresponds to an empty interpretation path, write the arrival time, peak amplitude, rising slope, half-peak width, and post-peak recovery difference of the current candidate pulse to the end of the confirmed single-photon event sequence and generate a confirmation write mark. When the independent origin path does not correspond to an empty interpretation path, generate a continuation removal mark for the current candidate pulse.

5. The single-photon detection method for confirming weak PMT single-photon events according to claim 4, characterized in that: S5 also includes: S5-2. Read the preceding events in the independent origin path. When the current candidate pulse has a continuation elimination mark, write the time interval, peak amplitude difference, half-peak width difference and recovery difference between the current candidate pulse and the preceding event into the continuation record bit of the preceding event, delete the current candidate pulse from the new event count bit, and output the continuation elimination result. S5-3. Merge the confirmation write mark and the continued elimination results to generate the single-photon event confirmation result corresponding to the current sampling position, and write it into the single-photon event confirmation result sequence in chronological order.

6. A single-photon detection system for confirming weak PMT single-photon events, used to implement the single-photon detection method for confirming weak PMT single-photon events as described in any one of claims 1-5, the system comprising a pulse extraction module, a mapping module, a path generation module, a judgment module, and a result confirmation module, characterized in that: The pulse extraction module is used to acquire continuous electrical pulse signals output by electrical sensors, search for peak points and trough points, identify signal segments between two adjacent trough points whose peak level is higher than the levels of the two trough points as candidate pulses, and extract arrival time, peak amplitude, rising slope, half-peak width and post-peak recovery difference, and output candidate pulse sequence. The mapping module is used to read previously confirmed single-photon events for the current candidate pulse, calculate the time interval, peak amplitude difference, half-peak width difference, and recovery difference between each confirmed single-photon event and the current candidate pulse, and construct a preceding continuation graph with preceding events as nodes and corresponding difference groups as boundary values. The path generation module is used to generate single-event interpretation paths for each node in the preceding continuation graph, generate joint interpretation paths for time-adjacent nodes, and determine paths that do not contain any nodes as empty interpretation paths, and output a set of candidate paths. The determination module is used to perform counterfactual elimination operation on each candidate path, remove the corresponding preceding events in the path one by one, and use the sum of the absolute values ​​of the time interval before and after removal, the peak amplitude difference, the half-peak width difference, and the recovery difference as the explanation residual, and take the candidate path that ranks first and has an empty explanation path as the independent origin path. The result confirmation module is used to write the current candidate pulse into the confirmed single-photon event sequence or remove it as a continuation pulse of the preceding event based on the independent origin path, and output the single-photon event confirmation result.

Images