Light intensity measurement method and device based on single photon detection and electronic equipment

Abstract

The invention provides a light intensity measurement method and device based on single-photon detection and electronic equipment, and relates to the technical field of single-photon detection and laser radar imaging, and the method comprises the steps: collecting the arrival time information of photons in a plurality of single-photon detection events for each detection position; extracting a time sequence characteristic quantity related to the light intensity estimation based on the arrival time information; constructing a model input parameter for estimating the target light intensity based on the time sequence characteristic quantity; inputting the model input parameters into a light intensity estimation model, and predicting a light intensity estimation value of the target light intensity; and generating a corresponding light intensity distribution image according to the light intensity estimation value. According to the method, high-precision light intensity estimation can be realized only by a few photons, the requirements on the repeated detection times and the signal-to-background ratio are greatly reduced, and the imaging robustness is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of single-photon detection and laser radar imaging technology, in particular to a light intensity measurement method and device based on single-photon detection and electronic equipment. BACKGROUND

[0002] In a single-photon detection system, traditional light intensity measurement relies on long-time or multi-cycle cumulative photon counting for statistical estimation, resulting in low imaging efficiency and difficulty in applying to extremely weak light or high-speed imaging scenarios. The first-photon imaging method realizes fast intensity estimation through the pulse sequence number of the first detected photon, but its performance is severely dependent on high signal-to-background ratio (SBR) conditions, which is easily dominated by background noise in actual outdoor environments. The existing technology generally uses the time information of single-photon events only for time-of-flight ranging, without incorporating the timing characteristics into the intensity estimation model, resulting in the rich time statistical information carried by the photons being ignored, limiting the intensity reconstruction accuracy and system robustness under low photon flux. SUMMARY

[0003] The present application aims to provide a light intensity measurement method and device based on single-photon detection to alleviate the above technical problems existing in the prior art.

[0004] In a first aspect, the present application provides a light intensity measurement method based on single-photon detection, comprising: For each detection position, collecting the arrival time information of photons in multiple single-photon detection events; Extracting timing characteristic quantities related to light intensity estimation based on the arrival time information; Constructing model input parameters for estimating target light intensity based on the timing characteristic quantities; Inputting the model input parameters into the light intensity estimation model to predict the light intensity estimation value of the target light intensity; Generating a corresponding light intensity distribution image according to the light intensity estimation value.

[0005] In an optional implementation, for each detection position, collecting the arrival time information of photons in multiple single-photon detection events, comprises: For each detection position, synchronously detecting the incident light signal using multiple parallelly arranged photon detection units; Continuously detecting photon arrival events in multiple detection periods and recording the absolute time stamp of photon arrival in each detection event; Identifying multiple photons first detected in at least two consecutive detection events and extracting the corresponding arrival time information thereof.

[0006] In an optional implementation, extracting timing characteristic quantities related to light intensity estimation based on the arrival time information, comprises: The collected multiple photon arrival time information is sorted in an event sequence to generate a photon arrival sequence arranged in chronological order; Two or more photon events that first appear continuously in the photon arrival sequence are identified, and the arrival time corresponding to the photon events is determined; The arrival time interval between the first continuously appearing photon events is calculated; The arrival time interval is determined as a timing feature quantity related to the light intensity estimation.

[0007] In an optional embodiment, a model input parameter for estimating the target light intensity is constructed based on the timing feature quantity, including: At least one timing feature quantity extracted from the photon arrival time information is received; wherein the timing feature quantity includes the time interval, arrival order or time distribution feature of the photon events; The timing feature quantity is combined or functionally transformed to generate a model input variable that can be used for light intensity estimation; The model input variable is associated with the response parameter of the detection system to construct an input parameter set of the light intensity estimation model.

[0008] In an optional embodiment, the model input parameter is input into the light intensity estimation model to predict the light intensity estimation value of the target light intensity, including: The model input parameter is input into the parameter estimation model, which is a mathematical model constructed based on statistical principles or data-driven methods; The input parameter is numerically solved or optimized based on the parameter estimation model to obtain an initial estimation result of the target light intensity; The initial estimation result is corrected according to the noise statistical characteristics of the detection system to obtain the corrected light intensity estimation value of the target light intensity.

[0009] In an optional embodiment, a corresponding light intensity distribution image is generated according to the light intensity estimation value, including: A light intensity data matrix is constructed based on the light intensity estimation value corresponding to each detection position; The light intensity data matrix is spatially mapped to generate two-dimensional light intensity distribution data or three-dimensional light intensity distribution data corresponding to the detection field of view; The two-dimensional light intensity distribution data or three-dimensional light intensity distribution data is converted into a corresponding image format to generate a corresponding light intensity distribution image.

[0010] In an optional embodiment, single-photon detection events are acquired by a multi-channel parallel photon detection device, including: Multiple independent photon detection channels are configured at the same detection position, and each channel synchronously receives the incident light signal; Each detection channel detects photon arrival events in multiple detection periods, and records the photon arrival time information detected by each detection channel respectively; The photon arrival time information from the multiple detection channels is subjected to time alignment and event association processing; Based on the associated photon event sequence, two photon events that first appear continuously in the same or different detection periods are identified.

[0011] In a second aspect, the present application provides a light intensity measurement device based on single photon detection, comprising: The acquisition module is configured to acquire, for each detection position, the arrival time information of photons in a plurality of single photon detection events; The extraction module is configured to extract timing feature quantities related to light intensity estimation based on the arrival time information; The construction module is configured to construct model input parameters for estimating the target light intensity based on the timing feature quantities; The prediction module is configured to input the model input parameters into the light intensity estimation model to predict a light intensity estimation value of the target light intensity; The generation module is configured to generate a corresponding light intensity distribution image according to the light intensity estimation value.

[0012] In a third aspect, the present application provides an electronic device comprising a processor and a memory, the memory storing computer executable instructions capable of being executed by the processor, and the processor executes the computer executable instructions to implement the single photon detection based light intensity measurement method of any one of the preceding embodiments.

[0013] In a fourth aspect, the present application provides a computer readable storage medium storing computer executable instructions, and the computer executable instructions, when invoked and executed by a processor, cause the processor to implement the single photon detection based light intensity measurement method of any one of the preceding embodiments.

[0014] The single photon detection based light intensity measurement method, device and electronic device provided by the present application break through the limitation of traditional photon counting system which only relies on photon number statistics, fully exploit the time dimension information, and significantly improve the photon utilization efficiency by acquiring the arrival time information of single photon detection events and extracting timing feature quantities. Based on the timing feature construction model input parameters and light intensity prediction, the system can still effectively distinguish signal and noise photons under extremely low SBR conditions, solving the problem that existing methods such as first photon imaging (FPI) fail due to background noise interference in real environment. This method can achieve high precision light intensity estimation with only a few photons, significantly reducing the requirements for repeated detection times and SBR, enhancing the imaging robustness, and being suitable for complex outdoor scenes, realizing efficient and low photon budget light intensity distribution reconstruction, and having outstanding technical advantages and practical value. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the specific embodiments or the prior art of the present application, the drawings needed to be used in the description of the specific embodiments or the prior art will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0016] Figure 1 A flow chart of a light intensity measurement method based on single photon detection provided by an embodiment of the present application; Figure 2 A flow chart of a specific light intensity estimation method based on specific time information of single photon detection events provided by an embodiment of the present application; Figure 3 A structural diagram of a light intensity measurement device based on single photon detection provided by an embodiment of the present application; Figure 4 A structural diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, but not all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0018] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. All other embodiments obtained by those skilled in the art without creative labor based on the embodiments in the present application are within the scope of protection of the present application.

[0019] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0020] The embodiments of the present application provide a light intensity measurement method based on single photon detection, as shown in Figure 1 The method mainly includes the following steps: S110, for each detection position, collecting the arrival time information of photons in a plurality of single photon detection events.

[0021] The above detection position refers to a basic unit of spatial discretization sampling of a target scene in an imaging process of a single-photon laser radar (LiDAR) system, that is, a pixel point or a spatial pixel. Each detection position corresponds to a local region of a target surface, and the system independently collects and processes echo photon data for the position to reconstruct depth and light intensity information of the point.

[0022] Arrival time information of photons in multiple single-photon detection events: refers to an absolute arrival time stamp (t1, t2, …, t n ) of each echo photon recorded by the detector at a certain detection position relative to a system clock reference (such as a laser pulse emission time). The time information not only includes the photon flight time (for calculating the distance), but also reflects the time distribution characteristics of the photon arrival process, and the statistical law is closely related to the signal photon rate, the background noise intensity and the target reflectivity. The key of the present application lies in utilizing the time sequence relationship (such as the time interval between consecutive photons) in the time information, rather than relying only on the cumulative photon number, to realize high-precision light intensity estimation under low-photon budget.

[0023] In a specific implementation, first, the scene to be measured is divided into a plurality of discrete spatial detection positions (i.e., a pixel grid), and the system observes each detection position point by point or in parallel through a scanning mirror or a single-photon avalanche diode (SPAD) array. For any detection position, a multi-channel (SPAD) detector is used to receive a weak light signal reflected from the position. Each SPAD channel works independently and is connected to a high-precision time-to-digital converter (TDC) to realize accurate recording of the occurrence time of each incident photon event. The control system synchronizes the pulse emission time of the laser as the time reference zero point. Each time the detector responds to a photon, the TDC records the absolute arrival time and marks the corresponding detection position, forming a photon event sequence arranged in chronological order.

[0024] The system continuously collects photon events of the position until a preset effective event trigger condition is met, for example, two photons are continuously detected for the first time within the same laser pulse period, or two photons are continuously detected for the first time across the pulse period. At this time, the arrival times t1 and t2 of the two photons are extracted, and the time interval Δt = |t2-t1| is calculated as a key input feature for subsequent light intensity estimation. The process is repeated at all detection positions to complete the spatiotemporal photon data collection of the whole scene.

[0025] S120, extracting a time sequence feature quantity related to light intensity estimation based on the arrival time information.

[0026] The timing characteristic quantity related to the light intensity estimation refers to a physical quantity extracted from the arrival time information of the single-photon detection event, which can reflect the statistical characteristics of the target reflection intensity. Since the photon arrival process obeys a Poisson-type random process, the average arrival rate of the signal photons is proportional to the reflectivity of the target surface (i.e., the light intensity), and the background noise photons exhibit uniform or slowly varying random interference. Therefore, the timing characteristic quantity can include the arrival time interval between consecutive photons (such as the first two-photon time interval Δt), the statistical distribution characteristics of the time interval (such as the mean, variance, and probability density function value), the phase relationship with respect to the laser pulse period (such as the relative position within the gating time), and the like, all of which contain key information for distinguishing between signal and noise and then inverting the real light intensity. These timing characteristic quantities constitute the core input variables that distinguish the present application from the conventional statistical system, enabling high-precision light intensity estimation with only a small number of photons.

[0027] In a specific implementation, first, the absolute arrival time sequence {t1, t2, …, tN} of a plurality of single-photon events at each detection position is obtained. n Subsequently, photon pairs that satisfy the "valid two-photon event" condition are identified and extracted from the sequence, i.e., two photons that are first detected consecutively within the same laser pulse period or two photons that are first detected consecutively across different pulse periods. For each valid photon pair, the arrival time interval Δt = |t2-t1| is calculated as the most core timing characteristic quantity.

[0028] Further, the system can normalize Δt to a dimensionless parameter or map it to a time interval preset in the theoretical model in combination with the laser repetition frequency and pulse width information. In addition, auxiliary timing characteristics can also be extracted, such as the first photon time of flight t1 with respect to the time of emission of the most recent laser pulse, whether the two photons fall within the same pulse echo gating window, the statistical moments (such as the mean and variance) of the Δt set of a plurality of valid two-photon events, and the probability value of Δt appearing in the peak region of the theoretical signal photon interval distribution.

[0029] The above timing characteristic quantities are combined into a feature vector as the input parameter of the subsequent light intensity estimation model. This processing is completed in real time by an embedded processor or FPGA, ensuring efficient data flow and supporting fast imaging. By accurately modeling the physical relationship between Δt and the target reflectivity, the system can significantly suppress the influence of background noise using only 2 photons, achieving robust and high-resolution light intensity reconstruction.

[0030] S130, constructing a model input parameter for estimating the target light intensity based on the timing characteristic quantity.

[0031] The model input parameters for estimating the target light intensity refer to a set of numerical variables formed after mathematical transformation, normalization, combination or mapping of the timing feature quantities extracted from single-photon detection events, which are suitable as input of a light intensity estimation model (such as a physical model, a statistical model or a machine learning model). These parameters are not original time stamps or simple time intervals, but processed feature representations that can directly reflect the relationship between the target reflection intensity and the noise environment. For example, the first two-photon time interval Δt itself can be used as the core input, or can be further converted into a conditional probability value P(Δt| λ) (λ is the signal photon arrival rate), or combined with system parameters (such as laser power, pulse width, background light level) to construct dimensionless ratios, logarithmic features, segmented codes, etc. The design of this parameter needs to ensure that there is a monotonic, modelable function relationship between it and the target light intensity, thereby supporting subsequent high-precision inversion.

[0032] In a specific implementation, the system first receives timing feature quantities from the output of the previous processing module, mainly including the arrival time interval Δt of the effective two-photon event and the flight time t1 of the first photon. Then, according to the known system operating parameters (such as laser repetition frequency f_rep, pulse width τ, detection gate time window T_gate, background photon count rate b, etc.), Δt is normalized and conditionally modeled.

[0033] Further, a regularization term or prior knowledge (such as spatial smoothness constraint) can be introduced to associate the timing features of adjacent pixels, generating enhanced input parameters containing local context information. This set of parameters is formatted into a fixed-length feature vector, such as [Δt, 1 / Δt, t1, log(Δt), background_level], which is input into the light intensity estimation model (such as maximum likelihood estimator, neural network, Bayesian inference model) to predict the target light intensity.

[0034] S140, input the model input parameters into the light intensity estimation model to predict the light intensity estimation value of the target light intensity.

[0035] The light intensity estimation model refers to a statistical physical mechanism based on single-photon detection process (such as Poisson process, geometric distribution, probability density function of arrival time interval), which converts the information extracted from the time sequence characteristics into a quantitative estimate of the reflectivity of the target surface. Optionally, the model form can include a target function constructed based on maximum likelihood estimation (MLE) (obtained by optimization to obtain the optimal light intensity parameter), a regularized optimization model (for example, a model introducing spatial smoothness, sparsity and other prior constraints based on MLE), and a pre-trained neural network (convolutional neural network (CNN), multilayer perceptron (MLP) and the like). The nonlinear relationship between the input features and the true light intensity is learned using simulation or measured data). The core of the light intensity estimation model is that its input depends on the time information of single-photon events, rather than the cumulative photon number in the prior art, thereby realizing high-precision estimation under extremely low photon budget.

[0036] The light intensity estimation value of the target light intensity refers to the numerical estimation result of the target surface reflectivity corresponding to a certain detection position output by the model, which is usually represented as normalized reflectivity, relative light intensity or absolute photon emission rate. The estimation value directly reflects the optical properties of the local region of the target and is the basic unit for generating the final intensity image. Since the present method only requires two photons to complete an effective estimation, the value still has high robustness and accuracy under extremely weak light conditions, which is significantly better than the statistical counting method.

[0037] S150, generating a corresponding light intensity distribution image according to the light intensity estimation value.

[0038] After obtaining the light intensity estimation value of each detection position, an image generation step is performed to form a complete two-dimensional or three-dimensional light intensity distribution image. The process includes spatial mapping, data organization, optional post-processing and image output, etc.

[0039] First, according to the laser scanning path or the spatial coding information of the SPAD array, the light intensity estimation value corresponding to each detection position is mapped to the corresponding pixel point in the image coordinate system. For example, in a system using a galvanometer scanner, the detection position associated with each photon event is determined by the current scanning angle; in a system using a SPAD array, the mapping is directly performed according to the physical position of the pixel.

[0040] Further, the light intensity estimation values of all pixel points are organized in matrix form to form a preliminary intensity image matrix. For pixels that have not collected effective two-photon events (i.e. no light intensity estimation value), they can be set to zero or filled in by interpolation methods (such as nearest neighbor, bilinear interpolation) to ensure the integrity of the image.

[0041] Finally, the processed light intensity distribution image is stored in a standard format (such as TIFF, PNG, RAW) or displayed in real time on the monitoring interface, supporting subsequent analysis and application. The image truly reflects the reflection characteristic distribution of the target scene, and can achieve high-fidelity imaging under extremely low photon budget (only 2 photons per pixel) and low SBR conditions, significantly better than traditional photon counting methods.

[0042] The present application breaks through the limitation of traditional photon counting system which only relies on cumulative photon number for light intensity estimation by using the arrival time information of single photon detection events. By extracting the first two photon time interval and other timing characteristics, a light intensity estimation model based on time statistical characteristics is constructed, realizing high-precision intensity reconstruction under extremely low photon budget (only 2 photons per pixel). This method significantly reduces the requirement for SBR, effectively suppresses background noise interference, and solves the problem of FPI failure in low SBR environment. Compared with the prior art, the present scheme greatly improves the imaging efficiency, robustness and practicality of single photon detection system, and is especially suitable for complex real scenes such as long distance and strong background noise, providing a new efficient physical model and technical path for single photon LiDAR and weak light imaging fields.

[0043] For ease of understanding, the light intensity measurement method based on single photon detection provided by the embodiments of the present application is described in detail below.

[0044] In the optional implementation, the above-mentioned acquisition of the arrival time information of the photons in the multiple single photon detection events for each detection position can include the following steps 1-1 to 1-3 in specific implementation: Step 1-1, for each detection position, the incident light signal is synchronously detected by a plurality of parallel arranged photon detection units. Each detection position corresponds to an optical focusing point, and the return light is guided to a group of parallel arranged photon detection units (for example, 2-10 SPAD pixels integrated on the same chip). These detection units share the same field of view and time reference, independently output photon trigger pulses, and are connected to high-precision TDC modules. All channels are synchronously controlled by a unified clock to ensure that the time stamps recorded by each channel are comparable and consistent. Incident photons can randomly fall on any detection unit and be recorded, and the system does not need to distinguish which physical pixel responds, but only pays attention to whether continuous photon events occur within the same detection position. This structure can be used for single point detection head in scanning system, or integrated in each super pixel unit of SPAD array to realize full parallel imaging.

[0045] Step 1-2, continuously detect photon arrival events in multiple detection cycles and record the absolute timestamp of photon arrival in each detection event. A detection cycle refers to a time window from laser pulse emission to echo reception. During system operation, the laser periodically emits pulses at a fixed frequency (e.g., 10 MHz), with each pulse corresponding to a detection cycle. Within each cycle, all parallel detection units monitor in real time whether a photon has arrived. Once a channel responds, the TDC immediately records the absolute timestamp of the event (with a precision of picoseconds) and marks the corresponding detection position and laser pulse number. The control system stores all timestamps in sequence in a buffer, forming a time-ordered photon event stream. This process continues until the pre-set valid event condition is met (e.g., two photons are detected consecutively for the first time). The timestamp information not only includes the time of flight (for distance measurement), but also the relative interval characteristics between adjacent photons.

[0046] Step 1-3, identify the first two photons detected in at least two consecutive detection events and extract their corresponding arrival time information. Consecutive detection events refer to two photon responses that occur in time sequence, regardless of whether they belong to the same laser pulse cycle. The core of this step is to filter out valid photon pairs for intensity estimation from the original time sequence, excluding isolated noise or non-consecutive signal interference, thereby ensuring the accuracy of subsequent modeling. In actual operation, a time-ordered photon event queue is maintained for each detection position. When a new photon event arrives, it is checked to see if it is the first or second valid event since the last reset. If it is the first event, its timestamp is temporarily stored; if the second event arrives subsequently, it constitutes a "first two photon" combination, which is determined as a valid event pair. These two photons can appear within the same laser pulse echo gating window (i.e., in the same cycle) or in different pulse cycles (across cycles), as long as they are the first two consecutive photons in the time sequence for that position. The system then extracts the absolute arrival times t1 and t2 of the two photons, calculates the time interval Δt = |t2 - t1|, and uses this Δt and related timestamps as key inputs for subsequent processing. Once the extraction is complete, the system resets the state of that position and waits for the next round of valid event accumulation.

[0047] This approach effectively overcomes the problem of photon detection loss caused by dead time in single-channel detectors, significantly improving the capture efficiency of consecutive photon events. Combined with continuous monitoring across cycles and accurate timestamp recording, it ensures complete capture of valid signals in sparse photon environments. By clearly defining "the first two consecutive photons detected" as valid events, it achieves high robustness in identifying real signal photon pairs, greatly enhancing the stability of imaging and the accuracy of light intensity estimation in low signal-to-background conditions, providing solid technical support for high-efficiency imaging with only 2 photons per pixel.

[0048] Further, the above-mentioned extraction of time-of-arrival information based on the timing characteristics related to light intensity estimation, in specific implementation, can include the following steps 2-1 to step 2-4: Step 2-1, the collected multiple photon time-of-arrival information is sorted in event sequence to generate a photon arrival sequence arranged in time sequence. Since the multi-channel detector or cross-cycle detection may produce non-sequential storage of timestamp data, the real time sequence of photon events must be restored by sorting. This step ensures that the subsequent analysis is based on accurate time logic relationship, and provides reliable basis for identifying the "first continuous event". In specific implementation, all photon events from the same detection position (regardless of which detection channel or laser pulse cycle) are collected into a cache queue. Then, a quick sorting algorithm (such as merge sort or heap sort) is used to sort the absolute timestamp in ascending order, forming a photon arrival sequence strictly organized in time sequence {t1, t2, …, tN}, where t1< t2< …< tN. n n The sorted sequence clearly reflects the dynamic process of photon arrival, facilitating subsequent accurate identification of continuous event pairs.

[0049] Step 2-2, identify the first two or more photon events that continuously appear in the photon arrival sequence, and determine the arrival time of the photon events. The first continuous appearance refers to the first photon pair (or group) that constitutes a continuous response from the start of the sequence, ignoring the isolated noise events that may exist before. By traversing the sorted photon arrival sequence, the first two photon events that occur continuously in time are found. That is, starting from the first event t1, if there is a second event t2 immediately following it (without being in the same pulse cycle), and no other intermediate events are inserted between them, then it is determined as the "first continuous appearance" of two photon events. The control system records the absolute arrival time t1 and t2 of the two events as the time reference for subsequent calculation. This identification logic can be realized by hardware state machine or software conditional judgment, with low delay and high reliability. Even if the first few events in the sequence are background noise, as long as there is a pair of continuous signal photons, it can be effectively captured for intensity estimation.

[0050] ​Step 2-3, calculate the time interval between the first two consecutive photon events. After determining the arrival time of the first two consecutive photons t1 and t2, the time interval Δt = |t2-t1| is calculated by subtraction operation, which is an important statistical quantity for characterizing signal intensity: under the Poisson photon stream model, the higher the signal photon rate, the smaller the Δt; on the contrary, under the condition of low light intensity or high background, Δt is large and more random. Therefore, Δt has a clear physical connection with the target reflectivity, and can be used as the main input for light intensity estimation. Alternatively, multiple consecutive pairs (such as the interval between the first three photons) can also be calculated, but the first two photon interval is preferred to ensure the lowest photon budget and the highest efficiency.

[0051] Step 2-4, determine the arrival time interval as the time sequence feature quantity related to light intensity estimation. The calculated Δt is directly marked as the main time sequence feature quantity, and the related context information (such as the time of flight of the first photon, the background level estimate, etc.) is marked as a feature vector. The Δt value can be further normalized or mapped to a preset interval, adapting to different modeling methods (such as maximum likelihood function lookup table, neural network input).

[0052] This way ensures the accuracy of time sequence analysis by time sorting the photon event sequence; effectively improves the robustness of signal discrimination by identifying the first two consecutive photon pairs, and suppresses background noise interference; by calculating and establishing the time interval between the first two photons as the core time sequence feature quantity, it realizes the efficient conversion from raw detection data to light intensity sensitive parameters, breaks through the limitation of traditional photon counting system relying only on quantity statistics, significantly reduces the number of photons required for imaging and SBR requirements, and provides key technical support for ultra-high efficiency light intensity measurement of only 2 photons per pixel.

[0053] Further, the above-mentioned model input parameters for estimating the target light intensity based on the time sequence feature quantity can include the following steps 3-1 to 3-3 in specific implementation: Step 3-1, receive at least one time sequence feature quantity extracted from the photon arrival time information; wherein the time sequence feature quantity includes the time interval, arrival order or time distribution characteristics of the photon event. The time sequence feature quantity is not the original timestamp, but a filtered and refined statistical index: such as the time interval Δt between the first two photons reflecting the signal intensity level; the arrival order (such as whether it is the first consecutive pair) reflects the event effectiveness; the time distribution characteristics (such as the relative position within the pulse period, the interval variance of multiple events) can assist in distinguishing signal and background. In specific implementation, for each detection position, the input feature at least includes the core variable Δt (i.e. the time interval between the first two photons), and can be extended to include: the time of flight of the first photon t1, whether it occurs across the pulse period, the number of consecutive events, the mean and standard deviation of multiple intervals, etc.

[0054] Step 3-2, feature combination or function transformation is performed on the timing feature quantity to generate model input variables that can be used for light intensity estimation. Since there is usually a nonlinear relationship (such as exponential decay or inverse relationship) between the original time interval Δt and the target light intensity, it is difficult to model directly, so function transformation (such as taking the inverse, logarithm, square root) or constructing composite features (such as the ratio of Δt and background rate) is needed to make it closer to linear separable or conform to the physical model assumption.

[0055] Step 3-3, associate the model input variables with the response parameters of the detection system to construct the input parameter set of the light intensity estimation model. The response parameters of the detection system include laser repetition frequency, pulse width, detection gating time, SPAD quantum efficiency, background light level estimation value, etc. The model input variables generated in the previous step are fused with the system parameters calibrated in advance or estimated in real time. For example, in the maximum likelihood estimation framework, the background photon rate as a key prior parameter participates in the calculation of the probability density function; the laser pulse period is used to judge whether two photons come from the same signal source. These system parameters are weighted, normalized and spliced with the transformed timing features into a high-dimensional vector, which contains not only the dynamic features of the current event, but also the system working state, so that the model can maintain stable performance under different environmental and hardware conditions. Finally, the input parameter set is encapsulated and passed to the light intensity estimation model for subsequent prediction process.

[0056] This way comprehensively captures the statistical rules of the photon arrival process by receiving multiple types of timing feature quantities; enhances the modelability between the features and the target light intensity by function transformation and combination of the features; and by fusing the response parameters of the detection system, an input parameter set with dynamic perception and system adaptive ability is constructed, which significantly improves the accuracy, robustness and generalization ability of the light intensity estimation model, effectively overcoming the estimation failure problem at low SBR caused by ignoring time information and system coupling effect in traditional methods, and providing key support for realizing high-precision, low-photon budget single-photon imaging.

[0057] Further, the above inputting the model input parameters into the light intensity estimation model to predict the light intensity estimation value of the target light intensity can include the following steps 4-1 to 4-3: Step 4-1, input the model input parameters into a parameter estimation model, which is a mathematical model constructed based on statistical principles or data-driven methods. The parameter estimation model is a physical or learning model based on the statistical rules of single photon event time, which mainly associates the time sequence characteristics with the target reflection intensity by modeling the probability distribution of photon arrival process (such as Poisson point process). The model form can include two categories: one is an analytical model based on statistical principles (such as maximum likelihood estimation, Bayesian inference); the second is a data-driven model (such as neural network, random forest), which learns the mapping relationship between input parameters and real light intensity. In specific implementation, the input parameter set generated by the previous stage is input into the preset parameter estimation model. If a statistical model is used, for example, a maximum likelihood estimator, the target light intensity is used as the estimated variable to construct a conditional probability function. This function is derived based on the Poisson statistical characteristics of the photon arrival process, reflecting the probability of observing the current time sequence characteristics under a given signal intensity. If a data-driven model is used, a pre-trained lightweight neural network (such as MLP or shallow CNN) is used to directly map the output value.

[0058] Step 4-2, numerical solution or optimization processing is performed on the input parameters based on the parameter estimation model to obtain the initial estimation result of the target light intensity. For an analytical model, numerical optimization methods are used to search for parameter values that maximize the likelihood function; for a data-driven model, forward propagation operations are performed.

[0059] Step 4-3, the initial estimation result is corrected according to the noise statistical characteristics of the detection system to obtain the light intensity estimation value of the corrected target light intensity. Due to the influence of various noise sources such as dark count, background photons, and time jitter in single photon detection, the initial estimation may have bias or excessive variance. By introducing noise prior knowledge (such as background photon rate distribution, detection efficiency calibration value), the estimation result can be corrected to improve accuracy and stability.

[0060] This method realizes accurate mapping from time sequence characteristics to light intensity by introducing a parameter estimation model based on statistical principles or data-driven methods; guarantees the realizability and real-time performance of the algorithm by obtaining the initial estimation result through numerical solution or model reasoning; and effectively suppresses background interference and system bias by dynamically correcting the noise statistical characteristics of the detection system, significantly improving the estimation accuracy and robustness under low SBR and weak light conditions, so that stable and reliable light intensity estimation values can be obtained with only two photons per pixel, providing a solid method support for ultra-high efficiency single photon imaging.

[0061] Further, the above-mentioned generation of corresponding light intensity distribution image according to the light intensity estimation value can include the following steps 5-1 to 5-3 in specific implementation: Step 5-1, construct a light intensity data matrix based on the light intensity estimate values corresponding to each detection position. This light intensity data matrix is a two-dimensional array (or tensor), with each element corresponding to a light intensity estimate value of a spatial detection position (pixel point), forming the original data basis of the image. In specific implementation, fill the light intensity estimate value of each pixel into the pre-allocated memory matrix according to the spatial numbering order of the detection position (such as row-by-row scanning or array index). For pixels that have not successfully acquired valid two-photon events (i.e., no light intensity estimate value), mark them as invalid values (such as NaN or 0) for subsequent interpolation processing.

[0062] Step 5-2, perform spatial mapping processing on the light intensity data matrix to generate two-dimensional light intensity distribution data or three-dimensional light intensity distribution data corresponding to the detection field of view. Mapping processing is used to restore the logical data matrix to the geometric image in real space. Two-dimensional mapping is used for planar intensity imaging, while three-dimensional mapping combines depth information (obtained from the time of flight t1 extracted in the same system) to superimpose light intensity values on three-dimensional point clouds, forming data with brightness information. According to the laser scanning path, mirror angle, or physical layout of the SPAD array, establish the mapping relationship between the detection position and the actual spatial coordinates. For scanning systems, the position of each pixel is determined by the horizontal and vertical scanning angles; for area array detectors, the row and column indices are directly used to correspond to the physical pixel position. Through coordinate transformation algorithms (such as polar to Cartesian coordinates), map the light intensity value of each pixel to the Cartesian space to generate a two-dimensional grayscale image or a three-dimensional point cloud graph.

[0063] Step 5-3, convert the two-dimensional light intensity distribution data or three-dimensional light intensity distribution data into the corresponding image format to generate the corresponding light intensity distribution image. Image formats can include common PNG, TIFF, JPEG (for two-dimensional), or PLY, PCD (for three-dimensional point clouds), or can be output in real time through a video interface (such as HDMI). By calling the image encoding module, quantize the processed light intensity distribution data (such as normalized to the range of 0~255), color map (such as grayscale or heat map color matching), and compress and encode. For two-dimensional images, generate single-channel or multi-channel bitmaps; for three-dimensional data, encapsulate as point cloud files with intensity attributes. The output image can be attached with metadata (such as timestamp, laser parameters, SBR level, etc.) for subsequent analysis.

[0064] The method realizes the ordered organization of the scattered measurement values to the structured image data by constructing the light intensity data matrix, accurately restores the two-dimensional or three-dimensional geometric layout of the target scene through the space mapping processing, accurately aligns the light intensity information and the spatial position, and realizes the visualization and practical output of the imaging results through the standardized image format conversion. The whole process is a complete closed loop, which not only retains the advantages of single-photon efficient detection, but also generates a high-fidelity and analyzable light intensity distribution image, significantly improving the practicability and engineering application value of the imaging system in a weak light environment.

[0065] Further, the single-photon detection event is acquired by a multi-channel parallel photon detection device, which can include the following steps 6-1 to 6-4 in specific implementation: Step 6-1, a plurality of independent working photon detection channels are configured at the same detection position, and each channel synchronously receives the incident light signal. By deploying multiple parallel photon detection channels (such as multiple pixels in a SPAD array) under the same optical field of view, concurrent detection of the same target return light can be realized, signal loss caused by dead time can be significantly reduced, and the recognition probability of effective two-photon events can be improved. In specific implementation, the return light beams from the same spatial detection point are distributed to two or more independent photon detection channels (for example, SPAD pixels integrated on the same chip) through a beam splitter or direct coupling. Each channel has an independent avalanche trigger circuit and a time response unit, and can respond to a photon event independently. All channels share the same optical focusing system to ensure that the received signal is the reflection signal of the same target area. The number of detection channels can be set to 2-10 according to application requirements, which is not limited here.

[0066] Step 6-2, each detection channel detects photon arrival events in multiple detection periods and records the photon arrival time information detected by each channel. Each channel operates independently, continuously monitors photon arrival, and records the absolute time stamp through a high-precision TDC. This distributed acquisition mechanism ensures that even if a channel is in a dead time state, other channels can still respond to photons, thereby improving the overall detection sensitivity. Within the laser pulse repetition period, each detection channel monitors whether there is a photon trigger. Once a response occurs, the TDC of the corresponding channel immediately records the absolute arrival time (relative to the system master clock or the first laser pulse) of the event, and marks the channel number, laser pulse sequence number and belonging detection position. All time stamp data are cached by channel and have a unique identifier. The process continues across multiple detection periods to form an independent time sequence for each channel. Since different channels may use different TDC modules, there may be slight deviations in the original time stamp, which needs to be calibrated later.

[0067] Step 6-3, time alignment and event association of photon arrival time information from multiple detection channels. Time calibration is performed in the initialization phase: a standard light source with known time interval (such as a narrow pulse laser) is used to illuminate all channels, and the response time difference of each channel is measured to generate a time offset compensation table between channels. In normal operation, the time stamps output by each channel are corrected according to the compensation table to achieve global time alignment. Subsequently, the photon events of all channels are combined into a unified time sequence according to the corrected absolute time, and repeated or false triggering events (such as crosstalk) are removed.

[0068] Step 6-4, based on the associated photon event sequence, identify the first two consecutive photon events that appear in the same or different detection periods. In specific implementation, traverse the joint photon event sequence after time alignment, find the first two photon events that constitute consecutive arrival. That is, starting from the first item of the sequence, if the second event follows the first event (without intermediate insertion events), it is considered as the "first consecutive appearance" of the effective two-photon event. The two photons can come from the same or different detection channels, or can span different laser pulse periods (such as the first photon appearing in the nth pulse echo window, and the second appearing in the n+1th). The control system extracts the absolute arrival time of the two photons for calculating the time interval.

[0069] This way, by configuring multiple channels in parallel detection structure at the same detection position, by recording time information independently and achieving high-precision time alignment, the accuracy of the joint event sequence is ensured; and by integrating multi-channel data to identify the first two consecutive photon events, the capture ability of real signals in low SBR and high background noise environment is significantly improved. This scheme not only enhances the photon utilization efficiency of the system, but also greatly improves the stability of light intensity estimation and imaging quality, providing strong hardware support and technical feasibility guarantee for realizing high-efficiency single-photon imaging with only two photons per pixel.

[0070] Figure 2 A specific light intensity estimation method based on specific time information of single-photon detection events is shown, which mainly includes the following steps: Step S101, detect whether two consecutive photons have arrived; if so, record the absolute arrival times t1, t2 and time interval Δt of the two photons, and store them in the two-photon event database.

[0071] In the single photon event acquisition process, all photon arrival events are monitored in real time by a multi-channel single photon detector within each laser pulse cycle. For each spatial detection point, the absolute arrival time information of the single photon detection event is accurately recorded, and a complete timestamp sequence is constructed. When two photons are detected consecutively for the first time within the same pulse cycle, or two photons are detected consecutively for the first time between different pulse cycles, the system will extract the absolute arrival time (t1, t2) of the two photons and the time interval (Δt) thereof, and store the group of parameters as an effective two-photon event in the database. This method is universal and does not depend on specific acquisition processes or device configurations, and can be widely applied to various single photon counting systems.

[0072] In step S102, based on the statistical distribution of Δt, the laser pulse parameters and the system noise are combined to construct an intensity maximum likelihood estimation model, and a regularized convex optimization method is used to globally reconstruct the intensity distribution.

[0073] Based on the accurate time information of the collected single photon detection events, a maximum likelihood estimation model of the target intensity is constructed. Taking the first two photon arrival time interval Δt as the core statistical quantity, and comprehensively considering the laser pulse parameters and the system noise characteristics, the quantitative estimation of the target intensity is systematically realized. This process can be solved by using general mathematical methods such as regularized convex optimization, and is not limited to specific optimization algorithms. It should be particularly pointed out that the core innovation point of the present scheme is to directly use the time information of the effective photon events itself for intensity estimation, rather than simply relying on the total photon count or its statistical distribution characteristics. Taking an embodiment as an example, the following maximum likelihood estimation model of the target intensity can be established:

[0074]

[0075] wherein M is the number of pulses emitted when two photons appear consecutively for the first time within the same pulse cycle, or two photons are detected consecutively for the first time between different pulse cycles, is the spatial coordinate, is the number of photons detected by the detector at position in unit time, is the arrival time probability density function of the photon detected at β is the regularization term weight, Φ is the regularization term, and α x,y =1 / Δt, the intensity distribution image is reconstructed in the global range by using the regularized convex optimization method.

[0076] In step S103, the final intensity distribution result is output, and single photon imaging is realized.

[0077] Based on the above physical model, a high-precision intensity distribution image is reconstructed to realize high-resolution LiDAR imaging. The technology can accurately invert the target intensity information under the condition that only a very small number of photons (such as 2) are received per pixel, thereby significantly improving the imaging efficiency of the system and greatly reducing the dependence on SBR. According to the requirements of specific application scenarios, the imaging results can be processed to further optimize the image quality.

[0078] In summary, compared with the way of photon counting system for photon detection, the above-mentioned way provided by the embodiments of the application evaluates the light signal intensity through the specific time information of single photon detection events, so that only a small amount of detection (such as 2 times of photon detection per pixel) is needed to accurately recover the target depth and light intensity distribution, which greatly improves the detection efficiency, greatly reduces the requirement of the system for the number of repeated sampling, and improves the imaging capability in low signal scenarios. The application proposes a light intensity evaluation method based on single photon detection timing information, and takes the first two photon time interval scheme as a typical representative, successfully applies the accurate time information of single photon detection to the single photon LiDAR system. This technology fundamentally breaks through the dependence of the traditional FPI method on SBR conditions. Compared with the limitation of FPI using only the first photon information, the method innovatively uses the following strategies: the complete timing information (including the time interval and the absolute time of each) of the two photons detected continuously for the first time in the same pulse period, or the two photons detected continuously for the first time in different pulse periods. This technology breakthrough significantly improves the system's ability to identify noise photons, so that even in a very low SBR (0.01-0.1) actual field environment, the system still maintains high-precision, high-efficiency intensity and depth joint imaging reconstruction capability.

[0079] Compared with the traditional FPI technology, the application has the following significant advantages: first, it greatly reduces the technical requirements for SBR; second, it significantly improves the imaging robustness and practical performance of the system; especially worth emphasizing is that this technology has excellent adaptability in complex application scenarios such as outdoor strong background and long distance, providing reliable technical support for the practical engineering application of single photon LiDAR.

[0080] By constructing the above-mentioned new parameter estimation paradigm with effective photon event time information as the core, all time statistical characteristics of continuous single photons can be systematically mined and fully utilized, thereby significantly improving the photon utilization efficiency, and the performance is far superior to existing technical solutions. Compared with the traditional FPI method and its improved scheme, which are limited to hardware upgrade or algorithm optimization level technical improvement, this research has achieved a breakthrough in the bottom modeling level of the imaging physical mechanism. This theoretical breakthrough not only builds a new theoretical framework for the field of high-photon-efficiency imaging, but also opens up an important scientific value of technical development direction. It needs to be specially pointed out that the parameter estimation process can flexibly use various mathematical methods such as maximum likelihood estimation and regularized convex optimization, which is not specifically limited here.

[0081] In summary, the present application not only breaks through the technical bottleneck of the failure of FPI and its improved scheme under low SBR conditions, but also realizes high-resolution, high-robustness intensity and depth joint imaging under the condition of extremely low photon number (such as 2 photons) of single pixel in practical application. This technical breakthrough provides reliable technical support and theoretical basis for single-photon LiDAR technology in long-distance detection and complex environment application, and has significant industrial application value and broad market prospect.

[0082] Based on the above method embodiment, the embodiment of the present application provides a light intensity measuring device based on single-photon detection, as shown in Figure 3 The device includes the following parts: The acquisition module 310 is used for acquiring the arrival time information of photons in a plurality of single-photon detection events for each detection position; The extraction module 320 is used for extracting time sequence characteristic quantities related to light intensity estimation based on the arrival time information; The construction module 330 is used for constructing model input parameters for estimating target light intensity based on the time sequence characteristic quantities; The prediction module 340 is used for inputting the model input parameters into the light intensity estimation model to predict the light intensity estimation value of the target light intensity; The generation module 350 is used for generating a corresponding light intensity distribution image according to the light intensity estimation value.

[0083] In a feasible implementation manner, the above-mentioned acquisition module 310 is used for: For each detection position, a plurality of parallelly arranged photon detection units are used to synchronously detect the incident light signal; Continuously detect photon arrival events in a plurality of detection periods, and record the absolute time stamp of photon arrival in each detection event; Identify a plurality of photons first detected in at least two continuous detection events, and extract the corresponding arrival time information thereof.

[0084] In an implementation, the extraction module 320 is configured to: sequence the collected photon arrival time information to generate a photon arrival sequence in time order; identify two or more photon events that first appear consecutively in the photon arrival sequence and determine the arrival time corresponding to the photon events; calculate the arrival time interval between the first consecutively appearing photon events; determine the arrival time interval as a timing feature quantity related to the light intensity estimation.

[0085] In an implementation, the construction module 330 is configured to: receive at least one timing feature quantity extracted from the photon arrival time information; wherein the timing feature quantity includes a time interval, arrival order, or time distribution feature of the photon events; perform feature combination or function transformation on the timing feature quantity to generate a model input variable that can be used for light intensity estimation; associate the model input variable with the response parameter of the detection system to construct an input parameter set of the light intensity estimation model.

[0086] In an implementation, the prediction module 340 is configured to: input the model input parameter into a parameter estimation model, which is a mathematical model constructed based on statistical principles or data-driven methods; numerically solve or optimize the input parameter based on the parameter estimation model to obtain an initial estimation result of the target light intensity; correct the initial estimation result according to the noise statistical characteristics of the detection system to obtain a light intensity estimation value of the corrected target light intensity.

[0087] In an implementation, the generation module 350 is configured to: construct a light intensity data matrix based on the light intensity estimation values corresponding to each detection position; perform spatial mapping processing on the light intensity data matrix to generate two-dimensional light intensity distribution data or three-dimensional light intensity distribution data corresponding to the detection field of view; convert the two-dimensional light intensity distribution data or the three-dimensional light intensity distribution data into a corresponding image format to generate a corresponding light intensity distribution image.

[0088] In an implementation, the single-photon detection events are acquired by a multi-channel parallel photon detection device, including: configuring multiple independently operating photon detection channels at the same detection position, and each channel synchronously receives the incident light signal; Each detection channel detects photon arrival events in multiple detection cycles and records the arrival time information of the photons it detects. Time alignment and event correlation processing are performed on photon arrival time information from multiple detection channels; Based on the correlated photon event sequence, identify two photon events that appear consecutively for the first time in the same or different detection periods.

[0089] The light intensity measurement device based on single-photon detection provided in this application has the same implementation principle and technical effect as the aforementioned method embodiment. For the sake of brevity, any parts of the light intensity measurement device based on single-photon detection not mentioned in the embodiment can be referred to the corresponding content in the aforementioned light intensity measurement method embodiment based on single-photon detection.

[0090] This application also provides an electronic device, such as... Figure 4 The diagram shows the structure of the electronic device 100, which includes a processor 41 and a memory 40. The memory 40 stores computer-executable instructions that can be executed by the processor 41. The processor 41 executes the computer-executable instructions to implement any of the above-mentioned light intensity measurement methods based on single-photon detection.

[0091] exist Figure 4 In the illustrated embodiment, the electronic device further includes a bus 42 and a communication interface 43, wherein the processor 41, the communication interface 43, and the memory 40 are connected via the bus 42.

[0092] The memory 40 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 43 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 42 may be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 42 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0093] The processor 41 can be an integrated circuit chip with processing capability. In implementation process, each step of the above method can be completed by integrated logic circuit of hardware in the processor 41 or by instructions in the form of software. The processor 41 described above can be a general processor, including a central processing unit (CPU), a network processor (NP), etc.; can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component. The general processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as a hardware code processor to execute, or be executed by a combination of hardware and software modules in the code processor. The software module can be located in a random access memory, a flash memory, a read only memory, a programmable read only memory or an electrically erasable programmable memory, a register, or other mature storage medium in the art. The storage medium is located in the storage, and the processor 41 reads the information in the storage, and combines the hardware to complete the steps of the single-photon-detection-based light intensity measurement method of the foregoing embodiments.

[0094] The embodiment of the present application further provides a computer readable storage medium, which stores computer executable instructions. When the computer executable instructions are called and executed by a processor, the computer executable instructions cause the processor to implement the single-photon-detection-based light intensity measurement method described above. For specific implementation, refer to the foregoing method embodiments, which will not be described here again.

[0095] The computer program product of the single-photon-detection-based light intensity measurement method, device and electronic equipment provided by the embodiment of the present application includes a computer readable storage medium storing program codes. The instructions included in the program codes can be used to execute the method described in the foregoing method embodiments. For specific implementation, refer to the method embodiments, which will not be described here again.

[0096] Unless otherwise specifically stated, the relative steps, numerical expressions and values of the components and steps set forth in these embodiments do not limit the scope of the present application.

[0097] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a nonvolatile computer readable storage medium executable by a processor. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0098] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of measuring light intensity based on single photon detection, characterized in that, The method comprises the following steps: For each detection position, the arrival time information of photons in multiple single-photon detection events is collected; Based on the arrival time information, time sequence characteristic quantities related to light intensity estimation are extracted; Based on the time sequence characteristic quantities, model input parameters for estimating the target light intensity are constructed; The model input parameters are input into the light intensity estimation model to predict the light intensity estimation value of the target light intensity; According to the light intensity estimation value, a corresponding light intensity distribution image is generated.

2. The single-photon-detection-based optical-intensity measurement method according to claim 1, wherein, For each detection position, the arrival time information of photons in multiple single-photon detection events is collected, including: For each detection position, the incident light signal is synchronously detected by multiple parallel photon detection units; In multiple detection periods, photon arrival events are continuously detected, and the absolute time stamp of photon arrival in each detection event is recorded; Identify the first detected photons in at least two consecutive detection events and extract their corresponding arrival time information.

3. The single-photon-detection-based optical-intensity measurement method according to claim 1, wherein Based on the arrival time information, time sequence characteristic quantities related to light intensity estimation are extracted, including: Sort the collected multiple photon arrival time information in event sequence to generate a photon arrival sequence arranged in time sequence; Identify two or more photon events that first appear continuously in the photon arrival sequence, and determine the arrival time corresponding to the photon events; Calculate the arrival time interval between the first continuously appearing photon events; Determine the arrival time interval as the time sequence characteristic quantity related to light intensity estimation.

4. The single-photon-detection-based optical-intensity measurement method according to claim 1, wherein, Based on the time sequence characteristic quantities, model input parameters for estimating the target light intensity are constructed, including: Receive at least one time sequence characteristic quantity extracted from the photon arrival time information; wherein the time sequence characteristic quantity includes the time interval, arrival order or time distribution characteristic of the photon event; Feature combination or function transformation is performed on the time sequence characteristic quantity to generate a model input variable that can be used for light intensity estimation; Associate the model input variable with the response parameter of the detection system to construct an input parameter set of the light intensity estimation model.

5. The single-photon-detection-based optical-intensity measurement method according to claim 1, wherein, The model input parameters are input into the light intensity estimation model to predict the light intensity estimation value of the target light intensity, including: Input the model input parameters into the parameter estimation model, which is a mathematical model constructed based on statistical principles or data-driven methods; Based on the parameter estimation model, the input parameters are numerically solved or optimized to obtain an initial estimation result of the target light intensity; According to the noise statistical characteristics of the detection system, the initial estimation result is corrected to obtain the light intensity estimation value of the corrected target light intensity.

6. The single-photon-detection-based optical-intensity measurement method according to claim 1, wherein According to the light intensity estimation value, a corresponding light intensity distribution image is generated, including: Based on the light intensity estimation value corresponding to each detection position, a light intensity data matrix is constructed; Perform spatial mapping processing on the light intensity data matrix to generate two-dimensional light intensity distribution data or three-dimensional light intensity distribution data corresponding to the detection field of view; Convert the two-dimensional light intensity distribution data or three-dimensional light intensity distribution data into a corresponding image format to generate a corresponding light intensity distribution image.

7. The single-photon-detection-based optical-intensity measurement method according to claim 1, wherein, The single-photon detection events are obtained by a multi-channel parallel photon detection device, including: At the same detection position, multiple independent photon detection channels are configured, and each channel synchronously receives the incident light signal; Each detection channel detects photon arrival events in multiple detection periods and records the respective detected photon arrival time information; The photon arrival time information from the multiple detection channels is time-aligned and event-associated processed; Based on the associated photon event sequence, two photon events that first appear continuously in the same or different detection periods are identified.

8. An optical intensity measuring device based on single photon detection, characterized in that The method comprises: A collection module is configured to collect, for each detection position, arrival time information of photons in multiple single-photon detection events; An extraction module is configured to extract timing characteristic quantities related to light intensity estimation based on the arrival time information; A construction module is configured to construct model input parameters for estimating target light intensity based on the timing characteristic quantities; A prediction module is configured to input the model input parameters into a light intensity estimation model to predict a light intensity estimation value of the target light intensity; A generation module is configured to generate a corresponding light intensity distribution image according to the light intensity estimation value.

9. An electronic device, comprising: The computer readable storage medium stores computer executable instructions, and the computer executable instructions, when invoked and executed by the processor, cause the processor to implement the single-photon detection-based light intensity measurement method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer executable instructions, and the computer executable instructions, when invoked and executed by the processor, cause the processor to implement the single-photon detection-based light intensity measurement method of any one of claims 1 to 7.

Images