A method for correcting saturation signals of spaceborne photon counting lidar

Abstract

A method for correcting saturated signals of a satellite-borne photon counting laser radar includes target surface identification; saturated signal identification; extracting the number of pulse signal photon events and the number of surface photon events after saturated and unsaturated data segments; calculating the coefficient K of the latter; calculating the number of real signal photons received by the detector at the saturated position of the target surface; fitting the cumulative distribution histogram of the unsaturated data segment to obtain the root mean square pulse width of the cumulative histogram of the target surface signal; using the number of real signal photon events on the target surface and the root mean square pulse width as input parameters of the first photon deviation calculation model to obtain the first photon deviation correction value. The present invention has the advantages of being independent of external auxiliary data, low computational complexity, and strong robustness, and can be applied to various surfaces such as ice lakes, small pools, calm water surfaces, and ice surfaces that are prone to mirror reflection and cause signal saturation. And the correction result of the present invention can achieve "self-verification" without relying on complex on-site calibration.

Description

Technical Field

[0001] The invention relates to a method for correcting a saturation signal of a satellite-borne photon counting laser radar, and belongs to the technical field of satellite-borne laser radars. Background Art

[0002] As an active three-dimensional remote sensing observation technology, space-borne photon counting lidar satellites can obtain high-precision three-dimensional coordinates of the surface and can penetrate the water surface to measure the underwater topography. The underwater topographic data obtained by space-borne photon counting lidar can be used to construct high-precision water depth control points, which can be combined with passive optical images to invert shallow water depth. Photon counting lidar uses a single photon detector, which enables it to use high repetition rate, micro-pulse and multi-beam designs, and can obtain data with higher spatial and distance resolution than traditional linear lidar.

[0003] However, the existence of the "dead zone" of single-photon detectors causes the photon counting radar to produce significant nonlinear effects and first-photon deviations in the data due to detector saturation when receiving strong reflected signals. This phenomenon is particularly significant when detecting calm water surfaces, which restricts the depth measurement accuracy of photon counting lidar. Once saturation occurs in the detection data, the number of signal photon events or the count rate will decrease instead of increase, and the ranging value will be small, which can reach tens of centimeters. On surfaces such as ice lakes, sea ice, and shallow water in salt marshes, mirror reflections are easily generated, resulting in 50-80% of the data containing saturated signals, which seriously restricts the application of space-borne photon counting lidar data. Summary of the invention

[0004] The present invention proposes a method for correcting saturated signals of space-borne photon counting lidar, which recovers the number of signal photons missing due to the saturation of single-photon detectors through post-pulse signals in the data, and corrects the first-photon error caused by signal saturation in combination with the single-photon detector detection model.

[0005] This method uses the photon point cloud with geographic coordinates of the spaceborne photon counting lidar and its auxiliary data as input (such as the ATL03 data product of ICESat-2). The spaceborne photon counting lidar saturation signal correction method includes the following steps:

[0006] Step 1: Target surface identification: Use the signal photon flags in the satellite-borne lidar photon point cloud (such as the ICESat-2 ATL03 data product) to extract the signal photons in the point cloud data, and accurately identify the target ground by combining the signal photons with passive optical images or ground object classification mask data.

[0007] Step 2, saturation signal identification: Use the detector saturation signal mark provided by the satellite-borne lidar photon point cloud to match the photon point cloud, and judge whether the detector is saturated according to the saturation signal mark (the higher the label value, the higher the degree of detector saturation), and separate the saturated data segment and the unsaturated data segment.

[0008] Step 3, the relationship between the number of post-pulse signal photon events and the number of surface photon events: for the saturated and unsaturated data segments extracted in step 2, detrend the data by photon accumulation. Make a photon cumulative distribution histogram of the detrended photon events, and extract the number of surface signal photon events and the number of post-pulse photon events in the saturated and unsaturated data segments, respectively.

[0009] Step 4, recovering the number of saturation missing photon events: calculating the coefficient K of the number of post-pulse signal photon events and the number of surface signal photon events in the unsaturated data segment;

[0010] Using this coefficient and the number of pulse signal photon events after the saturation signal segment obtained in step 3, the actual number of signal photons received by the detector at the saturation position of the target surface is calculated.

[0011] Step 5, pulse width fitting: for the cumulative distribution histogram of the unsaturated data segment obtained in step 3, the photon cumulative distribution histogram is fitted using a fitting algorithm to obtain the root mean square pulse width of the target surface signal cumulative histogram.

[0012] Step 6: First photon bias correction: The number of real signal photon events on the target surface calculated in step 4 and the root mean square pulse width fitted in step 5 are used as input parameters of the first photon bias calculation model to obtain the first photon bias correction value.

[0013] In step 1, the satellite-borne lidar photon point cloud uses the ICESat-2 ATL03 data product.

[0014] In step 2, before determining whether the detector is saturated according to the tag value, the target surface is first segmented.

[0015] In step 3, before detrending the data by photon accumulation, the photon events are first segmented according to whether they are saturated or not.

[0016] In step 4, the relationship K between the number of surface photon events and the number of post-pulse photon events is as follows: ,

[0017] In the formula, N surface represents the number of surface photon events, N ap represents the number of post-pulse photons.

[0018] The step 6 is specifically as follows:

[0019] According to the response characteristics of the PMT extended dead zone detector, in ( t , t + t ) The detector detects the photon event within a certain time p The probability can be expressed as the product of the probability of not receiving photons in the previous dead time and the probability of receiving photons in the current time period. If it is assumed that the number of photons of the echo signal received by the detector is Gaussian distributed, the probability can be expressed as:

[0020] ,

[0021] In the formula, n s represents the number of real signal photons on the target surface recovered by coefficient K in step 4, f n represents the noise rate, K pe represents the number of photon events received, η q Represents the quantum efficiency of the detector.

[0022] For the multi-channel detector carried by ICESat-2, each time slice t The detection probability of an internal photon event can be expressed as follows:

[0023] ,

[0024] In the formula, P ji The extended dead zone single photon detector is j The channel gate i The detection probability within a time slice; N D Indicates the number of channels;

[0025] The first photon deviation of the photon counting lidar can be expressed as the difference between the first-order moment of the probability function of its photon event distribution and the true value of the distance measurement. When the target point position is set to T target , the first photon deviation can be expressed as:

[0026] ,

[0027] In the formula, c is the speed of light. By dividing the number of recovered saturation missing photons and the fitted RMS pulse width σ t Substituting into the above formula, the first photon deviation correction value can be calculated.

[0028] The present invention realizes the recovery of the number of photons of the saturated signal based on the "post-pulse" signal regarded as noise in the photon counting laser radar receiving system, and corrects the first photon deviation in combination with the single-photon detection model. The method provided by the present invention is independent of external auxiliary data, has low computational complexity and strong robustness, and can be applied to various surfaces such as ice lakes, small pools, calm water surfaces and ice surfaces that are prone to mirror reflection and signal saturation. In addition, since the "post-pulse" comes from multiple reflections of the optical end face in the receiving system, the relative time interval between the "post-pulse" and the main echo is fixed, that is, the correction result of the present invention can be "self-verified" without relying on complex on-site calibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is the overall flow chart of the saturation signal correction method for spaceborne photon counting lidar.

[0030] Figure 2 is a schematic diagram of the location of the post-pulse noise signal.

[0031] The data in the figure are the photon point cloud data (gt2l) of the ICESat-2 satellite passing over Lake Michigan at 5:40 local time on January 2, 2019. In the figure, ap_d is the signal output when the PMT single-photon detector is saturated, ap1, ap2 and ap3 are the after-pulses caused by multiple reflections on the optical fiber end face inside the ATLAS receiving system, and ap is the after-pulse caused by the ionization effect inside the PMT single-photon detector.

[0032] Figure 3 This is the distribution diagram of the first photon deviation of different signal pulse widths. Data from an ice lake on the Emery Ice Shelf, an ice lake in Greenland, a section of Lake Ontario, and ground calibration experiment data were selected to fit the pulse widths, which are 2.56ns, 2.5ns, 3ns, and 2.5ns, respectively.

[0033] Figure 4 This is the first photon deviation correction effect diagram.

[0034] in, Figure 4 (a) The trajectory of the ICESat-2 satellite passing over an ice lake on the Amery Ice Shelf on January 3, 2019. Figure 4 (b) is the ICESat-2 ice lake point cloud data. The red one is the photon point cloud of the ice lake surface, and the green line segment is the corrected lake surface height, with a correction amount of 16.52 cm. DETAILED DESCRIPTION

[0035] A method for calculating the first photon deviation of a saturated signal of a spaceborne photon counting lidar is used as an example to calculate the first photon deviation of a saturated ice lake. Figure 4 As shown, Figure 4(a) is the track image of ICESat-2 passing over an ice lake on the Amery Ice Shelf on January 3, 2019. Figure 4 (b) is the ICESat-2 ice lake point cloud profile. The first photon deviation calculation method includes the following steps:

[0036] Step 1, ice lake identification: Use the photon confidence label value given in the ATL03 data to match the photon label value confidence with the photon event according to time, set the photon confidence label value to high confidence, and extract each high-confidence photon event. Then calculate the height distribution standard deviation of the high-confidence photon event to identify the ice lake, and determine it as a suspected ice lake. Finally, match the suspected ice lake photon event with the passive optics of the same period according to longitude and latitude to accurately identify the ice lake.

[0037] Step 2, saturation signal identification: In the ice lake data, ATL03 provides saturation data label values, which are calculated by segmenting the data according to the satellite flight time. The time of the label value needs to be matched with the time of the ice lake photon event. Then set the threshold of the saturation data label value, and record the label value time point above the saturation threshold. Finally, find the time point of the saturation data point in the photon event, intercept the data of the time slice length adjacent to the target time point as the saturation data, and record the other data as unsaturated ice lake data.

[0038] Step 3, the relationship between the number of post-pulse signal photon events and the number of surface photon events: the unsaturated ice lake data extracted in step 2 is divided into several segments according to a fixed length, and the maximum value is found as the surface elevation by binning each segment in the height direction to detrend the data. The detrended data is binned according to height to make a photon cumulative distribution histogram, and the number of surface photon events and post-pulse photon events are counted at the same length at a fixed position in the photon cumulative distribution histogram. Repeat this step for each segment of unsaturated ice lake data extracted in step 2, and statistically analyze the relationship between the number of post-pulse signal photon events and the number of surface photon events. K .

[0039] Step 4, restore the number of saturated missing photons: divide the saturated ice lake data extracted in step 2 into several segments according to a fixed length, and perform data detrending by binning and accumulating the maximum value in the height direction of each segment as the surface elevation. Bin the detrended data according to the height to make a photon cumulative distribution histogram, count the number of pulse photon events at a fixed position in the photon cumulative distribution histogram, and substitute it into formula (1) to calculate the surface photon event number.

[0040] Step 5, pulse width fitting: The unsaturated ice lake data extracted in step 2 are divided into several segments according to fixed lengths, and the maximum value is found as the surface elevation by binning each segment in the height direction to detrend the data. The detrended data is binned according to the height to make a photon cumulative distribution histogram, and the cumulative histogram is fitted using the Gaussian fitting algorithm. Take 3 s The photons within the range are counted, the pulse width is calculated, and finally the pulse width units are converted into time.

[0041] Step 6: First photon deviation correction. The number of signal photons reflected from the actual surface of the object calculated in step 4 and the pulse width fitted in step 5 are used as input parameters of the first photon deviation calculation model and substituted into formula (4) to obtain the first photon deviation correction value.

[0042] Example verification

[0043] Figure 3 The first photon deviation is calculated based on the pulse waveform fitted from Greenland, the Amery Ice Shelf, Lake Ontario and ground calibration data. The pulse widths of the Gaussian model fitted from an ice lake on the Amery Ice Shelf, an ice lake on Greenland, a section of Lake Ontario and ground calibration experiments are 2.56ns, 2.5ns, 3ns and 2.5ns respectively. When the average number of photon events is less than 13, the first photon deviations obtained from the four signals are similar. When the average number of photon events is greater than 13, the first photon deviations obtained from the ice lake on the Amery Ice Shelf and Lake Ontario are slightly larger than the first photon deviations obtained from the ice lake on the Greenland Ice Shelf and the ground calibration experiment, and the maximum difference is ≤2cm.

[0044] Figure 4 Data of a saturated ice lake on the Amery Ice Shelf and the first photon bias correction results, Figure 4 The red point cloud in (b) is the photon point cloud of the ice lake surface. Due to the saturation of the detector, its elevation is higher than the actual height. The green solid line is the water surface height after the first photon deviation correction, and the correction amount is 16.52cm.

[0045] The present invention proposes a saturation signal correction method for spaceborne photon counting lidar, which can effectively restore the actual number of photons in the saturation signal and perform first photon deviation correction. This method can greatly improve the data availability of spaceborne photon counting lidar on highly reflective surfaces such as sea ice, ice lakes, and small lakes on the plateau, and provide a guarantee for high-precision depth inversion in shallow water.

Claims

1. A method for correcting saturation signals of a spaceborne photon counting laser radar, characterized in that The following steps are involved: Step 1: Target surface identification: Use the signal photon flags in the satellite-borne laser radar photon point cloud to extract the signal photons in the point cloud data, and accurately identify the target surface by combining the signal photons with passive optical images or ground object classification mask data; Step 2, saturation signal identification: match the detector saturation signal mark provided by the satellite-borne laser radar photon point cloud with the photon point cloud, determine whether the detector is saturated according to the saturation signal mark, and separate the saturated data segment and the unsaturated data segment; Step 3, the relationship between the number of post-pulse signal photon events and the number of surface signal photon events: for the saturated and unsaturated data segments extracted in step 2, respectively, detrend the data by photon accumulation; make a photon cumulative distribution histogram of the detrended photon events, and extract the number of surface signal photon events and the number of post-pulse signal photon events of the saturated and unsaturated data segments respectively; Step 4, recovering the number of saturation missing photon events: calculating the coefficient K of the number of post-pulse signal photon events and the number of surface signal photon events in the unsaturated data segment; The coefficient and the number of pulse signal photon events after the saturation data segment obtained in step 3 are used to calculate the number of real signal photon events at the target surface saturation position received by the detector; Step 5, pulse width fitting: fitting the photon cumulative distribution histogram of the unsaturated data segment obtained in step 3 using a fitting algorithm to obtain the root mean square pulse width of the target surface signal cumulative histogram; Step 6, first photon bias correction: The number of real signal photon events on the target surface calculated in step 4 and the root mean square pulse width fitted in step 5 are used as input parameters of the first photon bias calculation model to obtain the first photon bias correction value.

2. The method according to claim 1, characterized in that In step 1, the satellite-borne laser radar photon point cloud uses the photon point cloud data provided by ICESat-2 ATL03.

3. The method according to claim 1, characterized in that In step 2, before judging whether the detector is saturated according to the saturation signal mark, the target surface is first segmented.

4. The method according to claim 1, characterized in that In step 3, before detrending the data by photon accumulation, the photon events are first segmented according to whether they are saturated or not.

5. The method according to claim 1, characterized in that In step 4, the relationship K between the number of surface signal photon events and the number of post-pulse signal photon events is as follows: ， In the formula, N surface represents the number of surface signal photon events, N ap Represents the number of post-pulse signal photon events.

6. The method according to claim 1, characterized in that The step 6 is specifically as follows: According to the response characteristics of the PMT extended dead zone detector, The detector detects the photon event within a certain time p The probability can be expressed as the product of the probability of not receiving photons in the previous dead time and the probability of receiving photons in the current time period. If it is assumed that the number of photons of the echo signal received by the detector is Gaussian distributed, the probability can be expressed as: ， In the formula, n s represents the number of real signal photon events on the target surface recovered by coefficient K in step 4, f n represents the noise rate, K pe represents the number of received photon events, η q Represents the quantum efficiency of the detector.

Images