WO2024040912A1 - 激光雷达回波信号处理方法及装置、激光雷达探测系统 - Google Patents

激光雷达回波信号处理方法及装置、激光雷达探测系统 Download PDF

Info

Publication number
WO2024040912A1
WO2024040912A1 PCT/CN2023/079470 CN2023079470W WO2024040912A1 WO 2024040912 A1 WO2024040912 A1 WO 2024040912A1 CN 2023079470 W CN2023079470 W CN 2023079470W WO 2024040912 A1 WO2024040912 A1 WO 2024040912A1
Authority
WO
WIPO (PCT)
Prior art keywords
pulse
time
signal
intensity
pulses
Prior art date
Application number
PCT/CN2023/079470
Other languages
English (en)
French (fr)
Inventor
朱红雷
沈国峰
向少卿
Original Assignee
上海禾赛科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 上海禾赛科技有限公司 filed Critical 上海禾赛科技有限公司
Publication of WO2024040912A1 publication Critical patent/WO2024040912A1/zh

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
    • G01S7/4802—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
    • G01S17/02—Systems using the reflection of electromagnetic waves other than radio waves
    • G01S17/06—Systems determining position data of a target
    • G01S17/08—Systems determining position data of a target for measuring distance only
    • G01S17/10—Systems determining position data of a target for measuring distance only using transmission of interrupted, pulse-modulated waves
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
    • G01S17/88—Lidar systems specially adapted for specific applications
    • G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
    • G01S17/931—Lidar systems specially adapted for specific applications for anti-collision purposes of land vehicles
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C3/00—Measuring distances in line of sight; Optical rangefinders

Definitions

  • the invention relates to the field of signal processing technology, and specifically to a laser radar echo signal processing method and device, and a laser radar detection system.
  • Perceptual recognition technology for roads and obstacles includes sensing the precise distance, shape, and road type of obstacles and other perceptual characteristics.
  • lidar detection based on the direct time-of-flight (dTOF) principle which forms a point cloud of object shapes depicted by scene depth information.
  • dToF detects the distance to an object by emitting short pulses of light and then measuring the time it takes for the emitted light to return. It is suitable for applications such as industrial robots that require rapid ranging and obstacle avoidance detection, augmented reality (Augmented Reality, AR) applications, etc. Various fields. However, existing lidar detection devices based on the dToF principle still have problems such as insufficient ranging capabilities and high power consumption.
  • the present invention provides a lidar echo signal processing method and device, and a lidar detection system to improve the echo processing capability based on dTOF, thereby improving the lidar system. ranging performance and accuracy.
  • embodiments of the present invention provide a laser radar echo signal processing method, which method includes:
  • the arrival time of one pulse from the candidate pulses is selected as the TOF value.
  • the DC component of each statistical unit is the intensity statistical value of the historical statistical unit before the current statistical unit of the received signal; the AC component of each statistical unit is the intensity value of the current statistical unit. Subtract the DC component.
  • the method further includes waveform reshaping of the pulse to obtain a reshaped received signal.
  • the waveform reshaping includes: determining whether the signal strength value of the current statistical unit is in a wave trough, and the current statistical unit Whether the difference between the signal strength of the unit and the peak intensity of the previous pulse is greater than the associated swing of the AC component of the current statistical unit, the associated swing of the AC component is the relationship between the AC component of the current statistical unit and the preset coefficient product; if yes, return the intensity value of the current statistical unit to zero; if not, keep the intensity value of the current statistical unit unchanged.
  • screening the plurality of pulses according to pulse information of the received signal to obtain candidate pulses includes:
  • the pulse width of the first screening pulse is less than the first threshold and the signal-to-noise ratio of the first screening pulse is less than the second threshold, the first screening pulse is deleted, otherwise the first screening pulse is retained as a candidate. pulse.
  • the method further includes: calculating the arrival time of each of the first screening pulses or each of the candidate pulses.
  • selecting the arrival time of a pulse from the candidate pulses as the TOF value includes:
  • the arrival time of the pulse with the latest arrival time and an intensity value greater than the set value is selected from the candidate pulses as the TOF value.
  • calculating the arrival time of each first screening pulse or each candidate pulse includes:
  • the arrival time of the pulse is calculated based on the arrival intensity.
  • determining the edge moment of the pulse based on the peak moment of the pulse includes:
  • the threshold time is determined to be the edge time of the pulse.
  • calculating the arrival intensity according to the intensity of the pulse in the time period from the edge moment to the peak moment includes: calculating the median or average value of the pulse intensity between the edge moment and the peak moment as reach intensity.
  • calculating the arrival time of the pulse according to the arrival intensity includes:
  • the time weight is calculated according to the arrival intensity, the first intensity and the second intensity
  • the arrival time of the pulse is calculated based on the first time, the second time and the time weight.
  • the method further includes: filtering the received signal to smooth the received signal.
  • the method further includes: calibrating the TOF value.
  • the method further includes: acquiring the electrical signal generated by the detection unit in the associated time and associated space; converting the electrical signal into a digital signal, and the digital signal is the received signal.
  • the correlation space includes: the theoretical area corresponding to the echo signal in the detection unit array and the adjacent area, the adjacent area does not overlap or partially overlaps with the theoretical area; the correlation time and the Statistical unit correspondence.
  • converting the electrical signal into a digital signal includes: converting the electrical signal into a digital signal using a time-to-digital conversion method; or converting the electrical signal into a digital signal using an analog-to-digital conversion method.
  • the electrical signal is converted into a digital signal.
  • converting the electrical signal into a digital signal using a time-to-digital conversion method includes: generating a set of histograms based on the number of the detection units illuminated in the associated time and associated space; according to the histogram Get the digital signal.
  • generating a set of histograms based on the correlation time and the number of the detection units lit in the correlation space includes: calculating the total number of the detection units lit in the correlation space in each of the statistical units, generate a set of histograms according to the total number of each statistical unit; or count the number of lighted detection units in the theoretical area in each statistical unit, and count the number of detection units in each statistical unit
  • the number of lighted detection units in the adjacent area is used to generate a set of histograms based on the number of lighted detection units in the theoretical area and the number of lighted detection units in the adjacent area.
  • using an analog-to-digital conversion method to convert the electrical signal into a digital signal includes: obtaining the current signal or voltage signal corresponding to the detection unit in the associated time and associated space, and converting the current signal or voltage signal into a digital signal. Signal.
  • embodiments of the present invention also provide a laser radar echo signal processing device, which includes:
  • a signal acquisition module used to acquire received signals, where the received signals include multiple pulses and their corresponding time information and intensity information;
  • Orthogonal component calculation module used to decompose the intensity value of the received signal in each statistical unit into a DC component and an AC component, where the statistical unit is a time point or a clock beat;
  • a pulse detection module configured to screen the plurality of pulses to obtain candidate pulses according to the pulse information of the received signal, where the pulse information includes the DC component and the AC component;
  • TOF analysis module used to calculate the arrival time of each pulse
  • the TOF decision module is used to select the arrival time of a pulse from the candidate pulses as the TOF value.
  • the device further includes: a resampling filtering module, configured to filter the received signal to smooth the received signal.
  • a resampling filtering module configured to filter the received signal to smooth the received signal.
  • the device further includes: a waveform reshaping module for waveform reshaping the pulse to obtain a reshaped received signal;
  • the waveform reshaping includes: determining whether the signal strength value of the current statistical unit is is in the wave trough, and whether the difference between the signal intensity of the current statistical unit and the peak intensity of the previous pulse is greater than the associated swing of the AC component of the current statistical unit, the associated swing of the AC component is the current statistical unit The product of the AC component and the preset coefficient; if yes, return the intensity value of the current statistical unit to zero; if not, keep the intensity value of the current statistical unit unchanged;
  • the pulse detection module is specifically configured to select a set number of pulses with the highest peak intensity from the plurality of pulses as first screening pulses; calculate the pulse width of each first screening pulse, and Calculate the signal-to-noise ratio of each first screening pulse according to the DC component and the AC component; if the pulse width of the first screening pulse is less than the first threshold, and the signal-to-noise ratio of the first screening pulse is less than the second threshold, delete the first screening pulse, otherwise retain the first screening pulse as a candidate pulse.
  • the TOF analysis module is specifically configured to calculate each of the first screening pulses, or calculate the arrival time of each of the candidate pulses.
  • the device further includes: a TOF calibration module, used to calibrate the TOF value.
  • a TOF calibration module used to calibrate the TOF value.
  • the signal acquisition module includes: a signal acquisition module, used to acquire the electrical signal generated by the detection unit in the associated time and associated space, and convert the electrical signal into a digital signal, and the digital signal is the received signal. .
  • the signal acquisition module includes a TDC module or an ADC module;
  • the TDC module is used to convert the electrical signal into a digital signal using a time-to-digital conversion method
  • the ADC module is used to convert the electrical signal into a digital signal using an analog-to-digital conversion method.
  • embodiments of the present invention also provide a lidar detection system for determining the distance of a detected object based on TOF time.
  • the system includes: a transmitting device, a receiving device, and the aforementioned laser echo signal processing device;
  • the transmitting device is used to transmit optical pulse signals
  • the receiving device includes a plurality of detection units for receiving optical signals to generate receiving signals;
  • the laser echo signal processing device is connected to the receiving device and is used to determine the TOF time of emitting the optical pulse signal based on the received signal, and determine the distance of the detection object based on the TOF time.
  • the lidar echo signal processing method and device and the lidar detection system provided by the present invention obtain the received signal, decompose the intensity value of each statistical unit in the received signal into a DC component and an AC component, and analyze the received signal based on this information. Screen multiple pulses to select the most likely candidate pulses, and finally select the arrival time of a pulse from the candidate pulses as the TOF value.
  • the solution of the present invention not only improves the dTOF echo processing capability, but also reduces the power consumption and overhead of hardware implementation, reduces the complexity of the application, and can greatly improve the engineering and commercialization of the product.
  • the peak intensity, pulse width, and signal-to-noise ratio of the pulses are comprehensively considered to eliminate pulses that do not meet the requirements due to interference, making the final TOF value more accurate.
  • Figure 1 is a schematic diagram of the working principle of a laser radar detection device in the prior art
  • FIG. 2 is a flow chart of the laser radar echo signal processing method according to the embodiment of the present invention.
  • Figure 3 is a schematic diagram of the strongest pulse recorded in the embodiment of the present invention.
  • Figure 4 is a schematic diagram of a close-range echo pulse in an embodiment of the present invention.
  • Figure 5 is a flow chart for calculating the arrival time of a pulse in an embodiment of the present invention.
  • Figure 6 is a flow chart of converting electrical signals into digital signals using a time-to-digital conversion method in an embodiment of the present invention
  • Figure 7 is a schematic diagram of a histogram formed based on the single photon counting mechanism in the embodiment of the present invention.
  • Figure 8 is a schematic diagram of macro pixels in the SPAD array that match the size and shape of the aperture in the embodiment of the present invention.
  • Figure 9 is a schematic diagram comparing before and after filtering the jitter caused by the influence of noise on the waveform peak value in the embodiment of the present invention.
  • Figure 10 is a schematic diagram of the histogram after filtering the histogram shown in Figure 7;
  • Figure 11 is a schematic structural diagram of a laser radar echo signal processing device according to an embodiment of the present invention.
  • Figure 12 is another structural schematic diagram of a laser radar echo signal processing device according to an embodiment of the present invention.
  • Figure 13 is another structural schematic diagram of a laser radar echo signal processing device according to an embodiment of the present invention.
  • Figure 14 is another structural schematic diagram of a laser radar echo signal processing device according to an embodiment of the present invention.
  • Figure 15 is a schematic structural diagram of a lidar detection system according to an embodiment of the present invention.
  • the existing lidar measurement device mainly consists of two parts, namely the transmitting device TX and the receiving device RX.
  • the radar emits a very narrow infrared light pulse to the surface of the object 10, and the pulse echo reflected back by the object 10 is detected by the detector.
  • the detector is a single photon avalanche diode (SPAD) array.
  • the reflected pulse echo bombards the SPAD array.
  • the illuminated points in the SPAD array can participate in counting or collecting changes in current, that is, reflection
  • the correlation time during which the echo returns forms a single photon count or acquisition current change.
  • the maximum detection distance is the speed of light multiplied by 1024 nanoseconds (half of the laser round-trip time).
  • the core components of a dToF include: laser, Single Photon Avalanche Diode (SPAD) and Time to Digital Convert (TDC).
  • SPAD is a photodetection avalanche diode with single-photon detection capability, which can generate current as long as there is a weak light signal.
  • the laser may be a vertical cavity surface emitting laser (VCSEL), an edge emitting laser (EEL), or other types of lasers.
  • VCSEL vertical cavity surface emitting laser
  • EEL edge emitting laser
  • the laser of the dToF module emits pulse waves into the scene, and the SPAD receives the pulse waves reflected from the target object.
  • TDC can record the flight time of each received optical signal, which is the time interval between the transmitted pulse and the received pulse.
  • dToF will emit and receive N optical signals within one measurement time (that is, the time corresponding to measuring one pixel point).
  • the time corresponding to each emitted and received optical signal is a time window, and then make a histogram of the recorded N flight times.
  • Graph statistics in which the flight time t with the highest frequency is used to calculate the object to be measured depth.
  • the current actual accuracy of dToF can only reach the cm level. .
  • embodiments of the present invention provide a lidar echo signal processing method, an echo signal processing method and device, and a lidar detection system to improve the echo processing capability based on dTOF, thereby improving the ranging performance of the lidar device. and accuracy.
  • Step 201 Obtain a received signal, which includes multiple pulses and their corresponding time information and intensity information.
  • the time information and intensity information include: the start time, end time, time point corresponding to the peak value and intensity value of the pulse, and also include the start time, end time and intensity value of each statistical unit.
  • the received signal refers to a digital signal, or a digital signal obtained by analog-to-digital conversion of the analog signal generated by the lidar detection unit.
  • Step 202 Decompose the intensity value of the received signal at each statistical unit into a DC component and an AC component.
  • the statistical unit is a time point or a clock beat.
  • noise statistics are required. To do this, the intensity value of each statistical unit can be decomposed into DC and AC components.
  • the received signal refers to a detector corresponding to a certain angle in the lidar field of view, where the detector at a certain angle may include at least one detection unit (that is, the received signal is a detector corresponding to a certain angle in the lidar field of view).
  • the statistical unit can be based on time points as a unit, that is, counting the intensity information at each time point; or the statistical unit can be based on a clock beat as a unit, that is, counting the intensity information within each clock beat, where the clock beat refers to a fixed time interval.
  • the detector selects the SPAD array (that is, the detection corresponding to one pixel The detector includes multiple SPADs), the received signal includes the number of SPADs lit in the detector within one measurement time (emitting and receiving N optical signals), the statistical unit is a clock beat, and the one measurement time includes multiple consecutive clock beats , each statistical unit calculates the number of SPADs lit in the detector during this time period as the intensity value of the optical signal received under this statistical unit.
  • the received signal output by the detector is an analog signal, in which the size of the analog signal is used as the intensity value of the received optical signal (the analog signal needs to be converted into a digital signal to facilitate subsequent calculations).
  • the unit is a time point or a time beat; when the time point is used as the statistical unit, the size of the analog signal at that time point is the intensity value of the statistical unit; when the clock beat is used as the statistical unit, the time period corresponding to the clock beat.
  • the statistical value of the simulated signal size (such as the mean or median value or other statistical methods) is the intensity value of the statistical unit.
  • the intensity statistical value of the historical statistical unit before the current statistical unit of the received signal is used as the DC component of the statistical unit, and the intensity value of the current statistical unit minus the DC component is used as the DC component.
  • the AC component of the statistical unit is used for each statistical unit.
  • DC and AC respectively represent the statistical values of the current statistical unit of the DC component and AC component, that is, the DC component and the AC component.
  • i represents the sequence number of the statistical unit
  • F2(i) represents the intensity of the i-th statistical unit.
  • Funcd() is a custom function (such as average, median, variance statistics, etc.), and Funca() is the intensity value of the statistical unit minus the DC component.
  • Step 203 Screen the plurality of pulses to obtain candidate pulses according to the pulse information of the received signal, where the pulse information includes the DC component and the AC component.
  • a certain number of pulses with the highest peak intensity that is, the first screening pulses, can be selected first, and then candidate pulses are determined based on the pulse widths and signal-to-noise ratios of these pulses.
  • the first N strongest pulses can be recorded and their time information can be recorded. If the intensity of the new pulse is greater than the pulse with the smallest intensity in the N records, the original smallest intensity pulse is replaced with the new pulse, thus obtaining Top N strongest pulses.
  • PD0: PD0_TOF [1780 (start time), 1806 (peak time), 1826 (end time)]
  • PD1_TOF [1038 (start time), 1049 (peak time), 1058 (end time)]
  • PD2_TOF [1990 (start time), 2007 (peak time), 2017 (end time)]
  • PD3_TOF [1914 (start time), 1924 (peak time), 1935 (end time)]
  • PD0_peak [24.826171875 (intensity value)]
  • PD1_peak [23.866015625 (intensity value)]
  • PD2_peak [36.39296875 (intensity value)]
  • PD3_peak [25.03359375 (intensity value)]
  • the candidate pulses are selected from the 4 strongest pulses recorded. Specifically, the pulse width and signal-to-noise ratio of each first screening pulse are calculated. If the pulse width of the first screening pulse is less than the first threshold, and the signal-to-noise ratio of the first screening pulse is less than the second threshold, then The first screening pulse is deleted, otherwise the first screening pulse is retained as a candidate pulse.
  • the pulse width can be calculated based on the time information corresponding to the pulse, that is, the start time and end time of the pulse.
  • the signal-to-noise ratio of the pulse can be calculated based on the DC component and AC component in the pulse intensity value.
  • SNR(n) represents the signal-to-noise ratio of the n-th pulse
  • peak(n) represents the intensity of the pulse
  • dc and ac respectively represent the average DC component and the average AC component of the statistical unit included in the pulse
  • PW(n) represents the pulse width of the n-th pulse
  • tof_end(n) represents the end time of the pulse
  • tof_start(n) represents the start time of the pulse.
  • the pulse width of the first screening pulse can be first judged. If the pulse width is greater than or equal to the first threshold, the first screening pulse is retained; if the pulse width is less than the first threshold, the pulse width is processed again. Determination of signal-to-noise ratio, if the signal-to-noise ratio is less than the second threshold, delete the first screening pulse (determined to be noise or false echo), otherwise retain the first screening pulse.
  • pulse width and signal-to-noise ratio is mainly based on the fact that the echo tail at a short distance (such as within 10 meters) is long and close to saturation, and the signal-to-noise ratio is too low due to a large pulse width. Since the DC component of the statistical unit is the average of the intensities of all units before the current statistical unit, in a close-range scenario, the intensity of the pulse is very high and the width is very wide. The calculated DC component is larger and the AC component is smaller. , the signal-to-noise ratio is relatively low.
  • the basic difference in intensity of each statistical unit is very small, so The calculated dc is approximately equal to the intensity value of the pulse, that is, if dc is large, ac will be small, and the calculated signal-to-noise ratio is very small. If the candidate pulses are determined based solely on signal-to-noise ratio screening, this part of the pulses will be missed. By increasing the judgment of the pulse width, this situation can be avoided and the pulses corresponding to echoes that may be close to each other can be retained.
  • Step 204 Select the arrival time of a pulse from the candidate pulses as the TOF value.
  • the selection can be based on different decision-making principles, such as but not limited to the following principles:
  • the first stronger echo principle select the arrival time of the pulse with the earliest arrival time and an intensity value greater than the set value from the candidate pulses as the TOF value;
  • each field of view angle position of the point cloud is usually characterized by only one depth or distance value, it is necessary to select the most likely arrival time of the pulse as the measured TOF value. For this reason, in the above process, it is also necessary to calculate the arrival time of each pulse.
  • the arrival time of each candidate pulse can be calculated after obtaining each candidate pulse; the arrival time of each pulse can also be calculated first, and then filtered, for example, among the multiple pulses A certain number of pulses with the highest peak intensity, that is, the above-mentioned first screening pulse, calculate the arrival time of the first screening pulse, and then determine the candidate pulse.
  • the calculation of the arrival time T of each pulse can be combined with the signal detection to form a front-end and back-stage pipeline operation. There is no need to wait until all pulse detection is completed before starting the calculation. That is, when the characteristic value of pulse 0 is obtained, the arrival time T of pulse 0 can be estimated until the analysis of the Nth strongest pulse is completed.
  • FIG. 5 shows the calculation of each first filter in the embodiment of the present invention.
  • a flowchart of the arrival time of a pulse or each candidate pulse including the following steps:
  • step 501 the edge time of the pulse is determined based on the peak time of the pulse.
  • the high-frequency component within the limited time difference before the pulse peak time point can be selected for analysis.
  • the size of the limited time difference can be set as needed, which is equivalent to a swing threshold in the TOF direction. Taking into account the computing resources Limitation and calculation require a certain amount of time.
  • the threshold should not be too large or too small. It can be set according to the TOF error jitter requirements required by computing resources and equipment. For example, the threshold can be set to 12 statistical units. In addition, for different TOF distances, different thresholds can be set to refine the TOF error jitter swing to adapt to different distances and distances.
  • determine the peak time of the pulse and the start time of the pulse and determine the time that is a preset time difference before the peak time as the threshold time; determine whether the start time is between the threshold time and the peak time. time; if yes, determine the starting time to be the edge time of the pulse; if not, determine the threshold time to be the edge time of the pulse.
  • step 502 the arrival intensity is calculated according to the intensity of the pulse in the time period from the edge moment to the peak moment.
  • the signal between the start time and the edge time of the pulse is regarded as noise and does not participate in the calculation of the subsequent arrival time T.
  • the signal between the edge time and the peak time of the pulse is regarded as the pulse and does not participate in the subsequent arrival time T. Calculation of T.
  • the median value or mean value of the pulse intensity between the edge moment and the peak moment or a certain percentage (such as 40%) of the peak value can be calculated as the arrival intensity Ft.
  • step 503 the arrival time of the pulse is calculated based on the arrival intensity.
  • the two intensities closest to the arrival intensity are the first intensity and the second intensity
  • the time corresponding to the first intensity is determined to be the first time
  • the time corresponding to the second intensity is determined to be the second time.
  • (i-1) and (i) refer to the time value of the point, and the part represented by the absolute value is the weight of each time value.
  • the embodiments of the present invention are not limited to the TOF estimation interpolation operator of the above equation (5), and may also be linear interpolation, etc.
  • the laser radar echo signal processing method provided by the present invention decomposes the intensity value of each statistical unit in the received signal into a DC component and an AC component, and screens multiple pulses in the received signal based on this information to select the most interesting ones. Possible candidate pulses, and finally select the arrival time of a pulse from the candidate pulses as the TOF value.
  • the solution of the present invention has good performance in noise calculation and effective echo detection. By decomposing the intensity value of each statistical unit into a DC component and an AC component, the solution is further performed based on the DC component and AC component of each statistical unit.
  • Steps such as waveform reshaping, signal-to-noise ratio calculation, and candidate pulse screening can also filter out effective echoes for arrival time calculation in the case of low signal-to-noise ratio, such as long distance (weak echo signal) and low signal-to-noise ratio.
  • Effective echo detection; and the condition of filtering candidate pulses by pulse width has been added, which can further filter out effective echoes for arrival time calculation in the case of low signal-to-noise ratio, such as ultra-close distance (strong echo signal) ) wide saturation effective echo detection.
  • the solution of this application greatly improves the dTOF echo processing capability.
  • the processing flow of the solution of the present invention is a simple pipeline processing structure, which is easy to be implemented on a chip or FPGA (Field Programmable Gate Array, programmable array logic), and the hardened implementation logic is simple and does not require Excessive storage units reduce the power consumption and overhead of hardware implementation, reduce the complexity of applications, and can greatly improve product engineering and productization.
  • FPGA Field Programmable Gate Array, programmable array logic
  • the lidar echo signal processing method of the present invention can further calibrate the TOF value obtained in the above step 204.
  • the compensation value corresponding to each TOF value can be written into the corresponding In the list, the calibrated TOF value is obtained by looking up the table.
  • a TOF compensation value correspondence table is set in advance; the TOF compensation value correspondence table is searched according to the TOF value, and the compensation values corresponding to the two TOF values closest to the TOF value are obtained; and the compensation values corresponding to the two closest TOF values are obtained.
  • the compensation value corresponding to each TOF value is used to calibrate the TOF value.
  • the received signal described in step 201 above refers to a digital signal obtained by analog-to-digital conversion of the electrical signal generated by the lidar detection unit.
  • the detection unit will also be different, for example, it can be: SPAD (Single Photon Avalanche Diode, single photon avalanche diode), SiPM (Silicon photomultiplier, silicon photomultiplier tube), APD (Avalanche Photo Diode, Avalanche photodiode), etc., the embodiments of the present invention are not limited to this.
  • the electrical signal may specifically be an electrical signal generated by a detection unit in the associated time and associated space, that is, a digital signal obtained by converting the electrical signal generated by the detection unit in the associated time and associated space.
  • the digital signal is the received signal described in step 201.
  • the correlation space includes: the theoretical area corresponding to the echo signal in the detection unit array and the adjacent area, the adjacent area does not overlap or partially overlaps with the theoretical area, and the adjacent area is relatively close to the theoretical area.
  • the position is not limited, for example, it can be up, down, left, right, diagonal, etc.
  • the adjacent area and the theoretical area can be detector arrays of any shape, such as rectangular, circular, polygonal, etc., which are not limited in this embodiment of the present invention.
  • the echo signal is the echo signal corresponding to the laser radar at a certain field of view angle
  • the theoretical area is the corresponding echo signal in the detection unit array at the field of view angle when the aperture does not shift. area.
  • the correlation time corresponds to the statistical unit.
  • FIG. 6 it is a flow chart for converting electrical signals into digital signals using a time-to-digital conversion method in an embodiment of the present invention, which includes the following steps:
  • Step 601 Generate a set of histograms based on the correlation time and the number of illuminated detection units in the correlation space.
  • the statistics can be characterized by the single photon count activated within the statistical unit time period based on the single photon counting mechanism in the correlation time and correlation space.
  • the signal strengths received by several such statistical units are continuously correlated on the time axis to form a set of histograms. That is to say, the total number of detection units lit in the associated space in each statistical unit is calculated, and a set of histograms is generated based on the total number of each statistical unit.
  • Figure 7 it is a schematic diagram of a histogram formed on 2048 nanoseconds of continuous correlation time.
  • the correlation space is the theoretical area of the SPAD receiving array covered by the echo spot under ideal conditions. Considering that the light spot passing through the aperture may deviate from the correct receiving position on the SPAD array due to equipment assembly and adjustment process deviations, the correlation space also includes the SPAD array in adjacent areas adjacent to the theoretical area. Calculate the total number of detection units that are lit in the associated space in each statistical unit, that is, calculate the total number of detection units that are lit in the associated space within the theoretical area and adjacent areas during the drinking time of each statistical unit. , and generate a histogram.
  • the photon counting method of search area matching and accumulation can also be used to generate the histogram. Specifically, the number of lighted detection units in the theoretical area in each statistical unit is counted, and the number of lighted detection units in adjacent areas in each statistical unit is counted. According to the number of lighted detection units in the theoretical area and the number of illuminated detection units within the adjacent area to generate a set of histograms. That is, the statistical theory area and phase respectively The total number of lighted detection units in adjacent areas generates statistical data in corresponding statistical units in a set of histograms.
  • the number of lighted detection units in the theoretical area in each statistical unit is counted respectively, and the number of lighted detection units in adjacent areas in each statistical unit is counted.
  • the number of lighted detection units in the theoretical area forms an initial histogram, and a corresponding initial histogram is formed according to the number of lighted detection units in each of the adjacent areas, and each of the initial histograms is assigned Different weights are weighted to form a histogram of the theoretical area corresponding to the field of view, where the initial histogram weight corresponding to the theoretical area is higher than the initial histogram weight of the adjacent area.
  • the Macro-Pixel located in the center area of the SPAD array (the center of the nine-square grid in Figure 8, the area corresponding to the position of light spot C) is the theoretical area
  • the 8 Macro-Pixels surrounding it are adjacent areas.
  • the theoretical area and 8 adjacent areas each generate corresponding initial histograms, superimpose the initial histograms corresponding to the theoretical area and 8 adjacent areas to form a histogram of the theoretical area, and perform TOF calculation based on this histogram, where the theoretical area
  • the initial histogram weight is 50%
  • the initial histogram weight of each adjacent region is 6.25%.
  • the echo signal since the echo signal usually has a greater probability of falling in the theoretical area, it is the initial histogram of the theoretical area.
  • the map is assigned a higher weight ratio. This method can also offset the problem of increased ambient light noise caused by expanding the detector's area range (from the original theoretical area to the theoretical area plus adjacent areas), and can improve the echo. Signal processing robustness.
  • a macro pixel (MP, Macro Pixel) is composed of 20*12 SPADs, matching the size and shape of the aperture.
  • the echo spot covers exactly one MP, which is the theoretical area.
  • the spot coverage may shift.
  • the maximum left and right offset of the horizontal direction search is 3/4 of the macro pixel horizontal size
  • the vertical direction search has a maximum up and down offset of 1/2 of the macro pixel vertical size, that is, the adjacent area
  • the light spot search range is composed of 50*24 SPADs
  • the range of the array is the association space.
  • the embodiments of the present invention are not limited to the above search size and precision.
  • Step 602 Obtain the digital signal according to the histogram.
  • the above-mentioned method of expanding the search range can also be used.
  • the difference is that when using the analog-to-digital conversion method to convert the electrical signal into a digital signal, all detections in the associated area need to be The electrical signals output by the unit are combined and collected, and then the collected signals are converted into digital signals through analog-to-digital conversion.
  • the digital signal obtained above can also be filtered to smooth the signal, reduce noise, and improve the correlation of the signal waveform data.
  • the filtering is mainly to filter out the jitter caused by the noise on the peak value of the waveform.
  • the circled part in Figure 9a shows the jitter caused by noise (the peak of the pulse is split into two small peaks).
  • the above filtering can filter out the jitter and form a smooth pulse signal.
  • the filtering operation can be characterized as follows:
  • F1 represents the signal waveform of the signal I after being filtered by the FIR (Finite Impulse Response) filter.
  • the convolution coefficient C1 of the filter is all positive, which has a low-pass smoothing effect.
  • F2 represents the signal after the same-frequency half-phase resampling interpolation of the signal F1. For example, if the nth point is sampled, the center point of the sampling interpolation is at the n+0.5th position. If the order radius s2 is set to 1, then The points participating in the operation are F1(n-1), F1(n), F1(n+1), and F1(n+2). Among them, the FIR coefficient of the odd-numbered positions adjacent to the center point is positive, and the even-numbered position coefficient is Negative, and so on alternately with the number of orders.
  • the filtered histogram is shown in Figure 10. It can be seen that compared with the histogram before filtering ( Figure 7), the histogram after filtering ( Figure 10) is smoother, and the irregular jitter caused by noise is filtered out, which facilitates subsequent processing and calculation.
  • the lidar echo signal processing method provided by embodiments of the present invention has strong performance in terms of noise, false echo filtering, and effective echo detection.
  • the received signal includes the echo signal and interference signal reflected by the object
  • the received signal is processed, the intensity value of the received signal in each statistical unit is decomposed into a DC component and an AC component, and this information is used to filter out multiple
  • the TOF calculation is performed on the most likely echo, and the detection distance can be obtained.
  • the solution of the present invention can still filter out the effective echo in two situations with low signal-to-noise ratio: weak long-distance echo signal and ultra-short range wide saturated effective echo. Waves perform TOF calculations.
  • the solution of the present invention has strong robustness and simple operability, and the implementation power consumption, overhead and complexity are very low.
  • an embodiment of the present invention also provides a laser radar echo signal processing device, as shown in Figure 11, which is a schematic structural diagram of the device.
  • the signal acquisition module 111 is used to acquire received signals, where the received signals include multiple pulses and their corresponding time information and intensity information;
  • Orthogonal component calculation module 112 used to decompose the intensity values of the plurality of pulses in each statistical unit into DC components and AC components, and the statistical unit is a time point or a clock beat;
  • the pulse detection module 113 is configured to screen the plurality of pulses to obtain candidate pulses according to the pulse information of the received signal, where the pulse information includes the DC component and the AC component;
  • TOF analysis module 114 used to calculate the arrival time of each pulse
  • the TOF decision module 115 is used to select the arrival time of a pulse from the candidate pulses as the TOF value.
  • the received signal refers to a digital signal after analog-to-digital conversion of the electrical signal generated by the lidar detection unit.
  • the pulse detection module 113 may select a set number of pulses with the highest peak intensity from the plurality of pulses as the first screening pulses; calculate the pulse width of each first screening pulse, and calculate the pulse width according to the The DC component and the AC component calculate the signal-to-noise ratio of each first screening pulse; if the pulse width of the first screening pulse is less than the first threshold, and the signal-to-noise ratio of the first screening pulse is less than the second threshold, then delete the first screening pulse, otherwise retain the first screening pulse as a candidate pulse.
  • the TOF analysis module 114 can calculate each first screening pulse, or calculate the arrival time of each candidate pulse.
  • the specific calculation method can refer to the previous The description of the method embodiments of the present invention will not be repeated here. That is, the TOF analysis module can calculate the arrival time of each first screening pulse after screening out the first screening pulse, or calculate the arrival time of the candidate pulse after obtaining the candidate pulse through the second screening.
  • the TOF decision module 115 selects the arrival time of a pulse from the candidate pulses as the TOF value, it can make the selection based on different decision-making principles. For details, please refer to the description in the previous method embodiment of the present invention, which will not be described again here. .
  • FIG. 12 it is another structural schematic diagram of a laser radar echo signal processing device according to an embodiment of the present invention.
  • the laser radar echo signal processing device also includes: a resampling filter module 121, used to filter the received signal to smooth the received signal. Signal.
  • FIG. 13 it is another structural schematic diagram of a laser radar echo signal processing device according to an embodiment of the present invention.
  • the lidar echo signal processing device also includes: a waveform reshaping module 131, which is used to reshape the waveform of the pulse to obtain a reshaped received signal.
  • waveform reshaping The purpose of waveform reshaping is to return the starting and ending points of each pulse (ie, pulse intensity) to zero to facilitate subsequent calculations.
  • the traditional method is to subtract a uniform fixed DC level (i.e., the signal mean).
  • the statistical DC level is not necessarily ideal. Especially relative to each current time point, this will cause some of the DC levels subtracted at each time point to be too large and some to be too small, which will affect the accuracy of the signal pulse intensity during TOF interpolation calculation.
  • Unit is the unit, and the DC component of each statistical unit is retained, thus retaining the original characteristics of the statistical unit. It is only necessary to consider the AC component or high-frequency component of each statistical unit during TOF analysis and calculation.
  • the waveform reshaping algorithm in the embodiment of the present invention mainly has two points: one is to determine whether the intensity value of the current statistical unit is at the trough position, and the other is to determine whether the difference between the intensity value of the current statistical unit and the peak intensity of the previous pulse is high.
  • the associated swing of the AC component of the current statistical unit if these two conditions are met, the intensity value of the statistical unit is returned to zero.
  • the echo reshaping includes: determining whether the signal strength value of the current statistical unit is in the trough, and whether the difference between the signal strength of the current statistical unit and the peak intensity of the previous pulse is greater than the current statistical unit
  • the associated swing of the AC component, the associated swing of the AC component is the product of the AC component of the current statistical unit and the preset coefficient; if yes, then return the intensity value of the current statistical unit to zero; if not, then Keeps the intensity value of the current statistical unit unchanged.
  • the peak intensity of the previous pulse can be recorded by recursively comparing the extreme values of the before and after statistical units.
  • the signal strength value update of the statistical unit can be characterized as:
  • alpha is the preset coefficient
  • F3 is the reshaped waveform
  • FIG 14 it is another structural schematic diagram of a laser radar echo signal processing device according to an embodiment of the present invention.
  • the lidar echo signal processing device further includes: a TOF calibration module 141 for calibrating the TOF value.
  • TOF_after_calibration(d) TOF_before_calibration(d)+delta(d) (11)
  • delta(d) represents the calibration compensation value at the detection distance d.
  • the distance can be converted into a TOF value and quantified into a discrete TOF table.
  • the calibrated TOF value can be interpolated by looking up the table to find the calibration compensation value delta corresponding to the two closest TOF items. That’s it.
  • the embodiment of the present invention is not limited to which interpolation method is used, and the specific interpolation method can be defined during engineering implementation.
  • the size of the error in practical applications is related to the reflectivity of the measured object.
  • the corresponding calibration compensation value is different at the same distance. Therefore, it can be calculated by the distance before calibration and the strength of the received signal.
  • the reflectivity of the object is determined in the calibration table, and the compensation values corresponding to the two reflectivities closest to the calculated reflectivity at the distance before calibration are determined, and the calibration compensation value is calculated by interpolation.
  • TOF calibration module 141 can also be applied to the embodiments shown in FIG. 12 and FIG. 13 .
  • the device of the present invention may also include: a signal acquisition module (not shown), used to acquire the electrical signal generated by the detection unit in the associated time and associated space, and convert the electrical signal into a digital signal. signal, and the digital signal is the received signal obtained by the signal acquisition module.
  • a signal acquisition module (not shown), used to acquire the electrical signal generated by the detection unit in the associated time and associated space, and convert the electrical signal into a digital signal. signal, and the digital signal is the received signal obtained by the signal acquisition module.
  • the signal acquisition module can convert the electrical signal into a digital signal in various ways.
  • the signal acquisition module can include a TDC module or an ADC module. in:
  • the TDC module is used to convert the electrical signal into a digital signal using a time-to-digital conversion method
  • the ADC module is used to convert the electrical signal into a digital signal using an analog-to-digital conversion method. word signal.
  • the laser radar echo signal processing device decomposes the intensity value of each statistical unit into a DC component and an AC component for multiple pulses in the received signal including the echo signal reflected by the object and the interference signal. According to This information screens multiple pulses in the received signal to select the most likely candidate pulses, and finally selects the arrival time of a pulse from the candidate pulses as the TOF value.
  • the solution of the present invention not only ensures the dTOF echo processing capability, but also reduces the power consumption and overhead of hardware implementation, reduces the complexity of the application, and can greatly improve the engineering and commercialization of the product.
  • the lidar detection system 150 includes: a transmitting device 151, a receiving device 152, and the laser echo signal processing device 153 described in the previous embodiments. in:
  • the transmitting device 151 emits optical pulse signals
  • the receiving device 152 includes a plurality of detection units for receiving optical signals to generate receiving signals;
  • the laser echo signal processing device 153 is connected to the receiving device 152, and is used to determine the TOF time of emitting a light pulse signal based on the received signal, and determine the distance of the detection object based on the TOF time.
  • each module/unit included in each device and product described in the above embodiments may be a software module/unit or a hardware module/unit, or it may be partly a software module/unit and partly is a hardware module/unit.
  • each module/unit included therein can be implemented in the form of hardware such as circuits, or at least some of the modules/units can be implemented in the form of a software program.
  • the software program Running on the processor integrated inside the chip the remaining (if any) modules/units can use circuits and other hardware Implementation in the form of software; for each device or product applied or integrated in the chip module, each module/unit included in it can be implemented in the form of hardware such as circuits, and different modules/units can be located in the same component of the chip module ( For example, chips, circuit modules, etc.) or in different components, or at least some modules/units can be implemented in the form of software programs that run on the processor integrated inside the chip module, and the remaining (if any) partial modules/units can be implemented in the form of software programs.
  • the unit can be implemented in the form of hardware such as circuits; for each device or product applied or integrated in the terminal, each module/unit included in it can be implemented in the form of hardware such as circuits, and different modules/units can be located in the same component in the terminal (for example, chips, circuit modules, etc.) or in different components, or at least part of the modules/units can be implemented in the form of a software program that runs on the processor integrated inside the terminal, and the remaining (if any) part of the modules/units
  • the unit can be implemented in hardware such as circuits.
  • Embodiments of the present invention also provide a computer-readable storage medium.
  • the computer-readable storage medium is a non-volatile storage medium or a non-transitory storage medium, and a computer program is stored thereon.
  • the computer program is processed by a processor. During runtime, the steps of the method provided by the corresponding embodiment of FIG. 2 or FIG. 5 or FIG. 6 are executed.
  • An embodiment of the present invention also provides an electronic device, including a memory and a processor.
  • the memory stores a computer program that can be run on the processor.
  • the processor runs the computer program, it executes the above figure 2 Or Figure 5 or Figure 6 corresponds to the steps of the method provided by the embodiment.
  • the processor can be a central processing unit (CPU for short), and the processor can also be other general-purpose processors, digital signal processors (DSP for short), special-purpose Integrated circuits (application specific integrated circuits, ASICs for short), off-the-shelf programmable gate arrays (field programmable gate arrays, FPGAs for short) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
  • DSP digital signal processor
  • ASICs application specific integrated circuits
  • FPGAs field programmable gate arrays
  • a general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.
  • memory in embodiments of the present invention may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.
  • non-volatile Flexible memory can be read-only memory (ROM), programmable ROM (PROM), erasable programmable read-only memory (erasable PROM, EPROM), electrically erasable memory Programmable read-only memory (electrically EPROM, referred to as EEPROM) or flash memory.
  • Volatile memory may be random access memory (RAM), which is used as an external cache.
  • RAM random access memory
  • SRAM static RAM
  • DRAM dynamic random access memory
  • DRAM synchronous Dynamic random access memory
  • DDR SDRAM double data rate synchronous dynamic random access memory
  • ESDRAM enhanced synchronous dynamic random access memory
  • SLDRAM Synchronously connect dynamic random access memory
  • DDR RAM direct memory bus random access memory
  • DR RAM direct rambus RAM
  • Multiple appearing in the embodiment of the present invention refers to two or more than two.
  • connection appearing in the embodiments of the present invention refers to various connection methods such as direct connection or indirect connection to realize communication between devices, and the embodiments of the present invention do not limit this in any way.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Electromagnetism (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Optical Radar Systems And Details Thereof (AREA)

Abstract

一种激光雷达回波信号处理方法及装置、激光雷达探测系统，该方法包括：获取接收信号，接收信号包括多个脉冲及其对应的时间信息和强度信息；将接收信号在每个统计单元的强度值分解为直流分量和交流分量，统计单元为时刻点或时钟节拍；根据接收信号的脉冲信息筛选多个脉冲得到候选脉冲，脉冲信息包括直流分量和交流分量；从候选脉冲中选择一个脉冲的到达时间作为TOF值。本发明方案可以提高基于dTOF的回波处理能力，进而提升激光雷达系统的测距性能和精度。

Description

激光雷达回波信号处理方法及装置、激光雷达探测系统
本申请要求2022年8月26日提交中国专利局、申请号为202211035396.4、发明名称为“激光雷达回波信号处理方法及装置、激光雷达探测系统”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本发明涉及信号处理技术领域，具体涉及一种激光雷达回波信号处理方法及装置、激光雷达探测系统。
背景技术
随着高级驾驶辅助系统(Advanced Driving Assistance System，ADAS)到自动驾驶(Automated Driving，AD)技术的不断探索和发展，用于道路及障碍物的人工智能(Artificial Intelligence，AI)识别技术不断提高。对道路及障碍物的感知识别技术包括感知障碍物的精确距离、形状、以及道路类型等感知特征。这些特征感知获取的主流方法之一便是基于直接测量飞行时间(direct Time-Of-Flight，dTOF)原理的激光雷达探测，形成由场景深度信息描绘成的物体形状点云图。
dToF是通过发出短脉冲光然后测量发射的光返回所需的时间来检测与物体的距离，适用于工业机器人等需快速进行测距避障检测的应用、增强现实(Augmented Reality，AR)应用等多种领域。但现有的基于dToF原理的激光雷达探测装置还普遍存在测距能力不足、功耗开销较大的问题。
发明内容
本发明提供一种激光雷达回波信号处理方法及装置、激光雷达探测系统，以提高基于dTOF的回波处理能力，进而提升激光雷达系统 的测距性能和精度。
为此，本发明实施例提供如下技术方案：
一方面，本发明实施例提供一种激光雷达回波信号处理方法，所述方法包括：
获取接收信号，所述接收信号包括多个脉冲及其对应的时间信息和强度信息；
将所述接收信号在每个统计单元的强度值分解为直流分量和交流分量，所述统计单元为时刻点或时钟节拍；
根据所述接收信号的脉冲信息筛选所述多个脉冲得到候选脉冲，所述脉冲信息包括所述直流分量和所述交流分量；
从所述候选脉冲中选择一个脉冲的到达时间作为TOF值。
可选地，所述每个统计单元的直流分量为所述接收信号的当前统计单元之前的历史统计单元的强度统计值；所述每个统计单元的交流分量为所述当前统计单元的强度值减去所述直流分量。
可选地，所述方法还包括对所述脉冲进行波形重塑，得到重塑后的接收信号，所述波形重塑包括：判断当前统计单元的信号强度值是否处于波谷，且所述当前统计单元的信号强度与前一脉冲的峰值强度之差是否大于所述当前统计单元的交流分量的关联摆幅，所述交流分量的关联摆幅为所述当前统计单元的交流分量与预设系数的乘积；若是，则将当前统计单元的强度值归为零；若否，则保持当前统计单元的强度值不变。
可选地，所述根据所述接收信号的脉冲信息筛选所述多个脉冲得到候选脉冲包括：
从所述多个脉冲中选择峰值强度最高的设定数量的脉冲为第一筛选脉冲；
计算每个所述第一筛选脉冲的脉冲宽度，并根据所述直流分量和所述交流分量计算每个所述第一筛选脉冲的信噪比；
如果所述第一筛选脉冲的脉冲宽度小于第一阈值，且所述第一筛选脉冲的信噪比小于第二阈值，则删除所述第一筛选脉冲，否则保留所述第一筛选脉冲为候选脉冲。
可选地，所述方法还包括：计算每个所述第一筛选脉冲或每个所述候选脉冲的到达时间。
可选地，所述从所述候选脉冲中选择一个脉冲的到达时间作为TOF值包括：
从所述候选脉冲中选择强度值最大的脉冲的到达时间作为TOF值；或者
从所述候选脉冲中选择到达时间最早并且强度值大于设定值的脉冲的到达时间作为TOF值；或者
从所述候选脉冲中选择到达时间最晚并且强度值大于设定值的脉冲的到达时间作为TOF值。
可选地，所述计算每个所述第一筛选脉冲或每个所述候选脉冲的到达时间包括：
根据所述脉冲的峰值时刻确定所述脉冲的边沿时刻；
根据所述边沿时刻至所述峰值时刻的时间段内脉冲的强度计算到达强度；
根据所述到达强度计算所述脉冲的到达时间。
可选地，所述根据所述脉冲的峰值时刻确定所述脉冲的边沿时刻包括：
确定所述脉冲的峰值时刻和所述脉冲的开始时刻；
确定距离所述峰值时刻前预设时间差的时刻为阈值时刻；
判断所述开始时刻是否在所述阈值时刻至所述峰值时刻之间；
若是，则确定所述开始时刻为所述脉冲的边沿时刻；
若否，则确定所述阈值时刻为所述脉冲的边沿时刻。
可选地，所述根据所述边沿时刻至所述峰值时刻的时间段内脉冲的强度计算到达强度包括：计算所述边沿时刻至所述峰值时刻之间的脉冲强度的中位值或均值为到达强度。
可选地，所述根据所述到达强度计算所述脉冲的到达时间包括：
确定与所述到达强度最近的两个强度为第一强度和第二强度，确定所述第一强度对应的时刻为第一时刻，所述第二强度对应的时刻为第二时刻；
根据所述到达强度、所述第一强度和所述第二强度计算得到时间权重；
根据所述第一时刻、所述第二时刻和所述时间权重计算得到所述脉冲的到达时间。
可选地，所述方法还包括：对所述接收信号进行滤波，以平滑所述接收信号。
可选地，所述方法还包括：对所述TOF值进行校准。
可选地，所述方法还包括：获取关联时间及关联空间内探测单元产生的电信号；将所述电信号转换为数字信号，所述数字信号为所述接收信号。
可选地，所述关联空间包括：回波信号在探测单元阵列中对应的理论区域以及相邻区域，所述相邻区域与所述理论区域不重叠或部分重叠；所述关联时间与所述统计单元对应。
可选地，所述将所述电信号转换为数字信号包括：利用时间数字转换方法将所述电信号转换为数字信号；或者利用模数转换方法将所 述电信号转换为数字信号。
可选地，所述利用时间数字转换方法将所述电信号转换为数字信号包括：基于关联时间及关联空间内点亮的所述探测单元的数量，生成一组直方图；根据所述直方图得到所述数字信号。
可选地，所述基于关联时间及关联空间内点亮的所述探测单元的数量，生成一组直方图包括：计算每个所述统计单元中关联空间内点亮的所述探测单元总数，根据每个所述统计单元的所述总数，生成一组直方图；或者统计每个所述统计单元中所述理论区域内点亮的探测单元的数量，以及，统计每个所述统计单元中所述相邻区域内点亮的探测单元的数量，根据所述理论区域内点亮的探测单元的数量和所述相邻区域内点亮的探测单元的数量，生成一组直方图。
可选地，所述利用模数转换方法将所述电信号转换为数字信号包括：获取关联时间及关联空间内探测单元对应的电流信号或电压信号，将所述电流信号或电压信号转换为数字信号。
另一方面，本发明实施例还提供一种激光雷达回波信号处理装置，所述装置包括：
信号获取模块，用于获取接收信号，所述接收信号包括多个脉冲及其对应的时间信息和强度信息；
直交分量计算模块，用于将所述接收信号在每个统计单元的强度值分解为直流分量和交流分量，所述统计单元为时刻点或时钟节拍；
脉冲检测模块，用于根据所述接收信号的脉冲信息筛选所述多个脉冲得到候选脉冲，所述脉冲信息包括所述直流分量和所述交流分量；
TOF分析模块，用于计算各脉冲的到达时间；
TOF决策模块，用于从所述候选脉冲中选择一个脉冲的到达时间作为TOF值。
可选地，所述装置还包括：重采样滤波模块，用于对所述接收信号进行滤波，以平滑所述接收信号。
可选地，所述装置还包括：波形重塑模块，用于对所述脉冲进行波形重塑，得到重塑后的接收信号；所述波形重塑包括：判断当前统计单元的信号强度值是否处于波谷，且所述当前统计单元的信号强度与前一脉冲的峰值强度之差是否大于所述当前统计单元的交流分量的关联摆幅，所述交流分量的关联摆幅为所述当前统计单元的交流分量与预设系数的乘积；若是，则将当前统计单元的强度值归为零；若否，则保持当前统计单元的强度值不变；
可选地，所述脉冲检测模块，具体用于从所述多个脉冲中选择峰值强度最高的设定数量的脉冲为第一筛选脉冲；计算每个所述第一筛选脉冲的脉冲宽度，并根据所述直流分量和所述交流分量计算每个所述第一筛选脉冲的信噪比；如果所述第一筛选脉冲的脉冲宽度小于第一阈值，且所述第一筛选脉冲的信噪比小于第二阈值，则删除所述第一筛选脉冲，否则保留所述第一筛选脉冲为候选脉冲。
可选地，所述TOF分析模块，具体用于计算每个所述第一筛选脉冲，或者计算每个所述候选脉冲的到达时间。
可选地，所述装置还包括：TOF校准模块，用于对所述TOF值进行校准。
可选地，所述信号获取模块包括：信号采集模块，用于获取关联时间及关联空间内探测单元产生的电信号，将所述电信号转换为数字信号，所述数字信号为所述接收信号。
可选地，所述信号采集模块包括TDC模块或者ADC模块；
所述TDC模块，用于利用时间数字转换方法将所述电信号转换为数字信号；
所述ADC模块，用于利用模数转换方法将所述电信号转换为数字信号。
另一方面，本发明实施例还提供一种激光雷达探测系统，用于根据TOF时间确定探测物体的距离，所述系统包括：发射装置、接收装置以及前面所述的激光回波信号处理装置；
所述发射装置，用于发射光脉冲信号；
所述接收装置，包括多个探测单元，用于接收光信号生成接收信号；
所述激光回波信号处理装置，与所述接收装置相连，用于根据接收信号确定发射光脉冲信号的TOF时间，根据所述TOF时间确定探测物体的距离。
本发明提供的激光雷达回波信号处理方法及装置、激光雷达探测系统，获取接收信号，将接收信号中每个统计单元的强度值分解为直流分量和交流分量，根据这些信息对接收信号中的多个脉冲进行筛选，筛选出最有可能的候选脉冲，最后从候选脉冲中选择一个脉冲的到达时间作为TOF值。本发明方案在提高了dTOF回波处理能力的同时，降低了硬件实现的功耗及开销，降低了应用的复杂度，可以极大提升产品的工程化与产品化。
进一步地，在确定每个统计单元的直流分量时，只需采集前向历史值作为参考，即只考虑本统计单元和之前统计单元的数据，而非采集所有统计单元的数据后再进行计算，从而可以逐个统计单元计算统计值，不仅可以极大提高信号处理的实时性，而且还可节省额外的存储开销。本发明方案易于芯片化实现或FPGA实现，且硬化实现逻辑简单，不需要大量的片上RAM作缓存。
进一步地，在对多个脉冲进行筛选时，综合考虑了脉冲的峰值强度、脉冲宽度、信噪比，排除了由于干扰而产生的不符合要求的脉冲，使最终的TOF值更准确。
附图说明
图1是现有技术中激光雷达探测装置的工作原理示意图；
图2是本发明实施例激光雷达回波信号处理方法的一种流程图；
图3是本发明实施例中记录的最强脉冲的示意图；
图4是本发明实施例中近距离回波脉冲的示意图。
图5是本发明实施例中计算脉冲的到达时间的流程图；
图6是本发明实施例中利用时间数字转换方法将电信号转换为数字信号的流程图；
图7是本发明实施例中基于单光子计数机理形成的直方图示意图；
图8是本发明实施例中SPAD阵列中与光阑尺寸形状相匹配的宏像素示意图；
图9是本发明实施例中对波形峰值受噪声影响产生的抖动滤波前后对比示意图；
图10是对图7所示直方图滤波后的直方图示意图；
图11是本发明实施例激光雷达回波信号处理装置的一种结构示意图；
图12是本发明实施例激光雷达回波信号处理装置的另一种结构示意图；
图13是本发明实施例激光雷达回波信号处理装置的另一种结构示意图；
图14是本发明实施例激光雷达回波信号处理装置的另一种结构示意图；
图15是本发明实施例激光雷达探测系统的结构示意图。
具体实施方式
为使本发明的上述目的、特征和有益效果能够更为明显易懂，下 面结合附图对本发明的具体实施例做详细的说明。
下面首先结合图1对激光雷达探测装置的工作原理进行简要说明。
参照图1，现有的激光雷达测量装置主要由两部分组成，即发射装置TX和接收装置RX。对应视场中某一个角度(极坐标的水平角与垂直角)位置上的物体10，雷达发射一个很窄的红外光脉冲打到物体10表面，通过物体10反射回来的脉冲回波被探测器接收，例如，探测器为单光子雪崩二极管(single photon avalanche diode，SPAD)阵列，反射的脉冲回波轰击SPAD阵列，SPAD阵列中被点亮的点便可参与计数或者采集电流的变化，即反射回波返回的这段关联时间内形成单光子计数或采集电流变化。假设从激光发射时刻点开始计算，捕捉回波的最大时间片段为2048纳秒，那么最远探测距离即为光速乘上1024纳秒(激光往返时间的一半)。当雷达以一定的扫描方式(逐行、逐列、或其他特定的顺序)对视场中不同角度做完整遍历扫描时，便形成一帧完整的具有一定分辨率的场景深度点云图，该点云图正是后端自动驾驶主控计算平台所需的包含场景深度特征感知的数据。
一种dToF的核心组件包括：激光器、单光子雪崩二极管(Single Photon Avalanche Diode，SPAD)和时间数字转换器(Time to Digital Convert，TDC)。SPAD是一种具有单光子探测能力的光电探测雪崩二极管，只要有微弱的光信号就能产生电流。其中，激光器可以是垂直腔面发射激光器(vertical cavity surface emitting laser，VCSEL)、边发射激光器(edge emitting laser，EEL)，或者其他类型的激光器。dToF模组的激光器向场景中发射脉冲波，SPAD接收从目标物体反射回来的脉冲波。TDC能够记录每次接收到的光信号的飞行时间，也就是发射脉冲和接收脉冲之间的时间间隔。dToF会在一次测量时间内(即测量一个像素点对应的时间)发射和接收N次光信号，每次发射和接收光信号对应的时间为一个时间窗口，然后对记录的N次飞行时间做直方图统计，其中出现频率最高的飞行时间t用来计算待测物体 的深度。对于dToF而言，由于雪崩过程存在量子噪声和放大器噪声，以及dToF中TDC模块存在的固有噪声，导致目前dToF的实际精度(真实深度值和相机的测量值之间的差)只能达到cm级。
针对上述问题，本发明实施例提供一种激光雷达回波信号处理方法回波信号处理方法及装置、激光雷达探测系统，以提高基于dTOF的回波处理能力，进而提升激光雷达装置的测距性能和精度。
如图2所示，是本发明实施例激光雷达回波信号处理方法的一种流程图，包括以下步骤：
步骤201，获取接收信号，所述接收信号包括多个脉冲及其对应的时间信息和强度信息。
其中，所述时间信息和强度信息包括：脉冲的开始时间、结束时间、峰值对应的时间点和强度值，还包括每个统计单元的开始时间、结束时间、以及强度值。
需要说明的是，所述接收信号是指数字信号，或者对激光雷达探测单元产生的模拟信号经过模数转换后的数字信号。
步骤202，将所述接收信号在每个统计单元的强度值分解为直流分量和交流分量，所述统计单元为时刻点或时钟节拍。
为了提取出有效回波脉冲，需要进行噪声统计。为此，可以将每个统计单元的强度值分解为直流分量和交流分量。
在本发明实施例的接收信号是指，对应激光雷达视场中某一角度的探测器，其中所述某一角度的探测器可以包括至少一个探测单元(即接收信号是激光雷达视场中对应于一个像素点的探测器产生的信号)。所述统计单元可以以时间点为单位，即统计每个时间点的强度信息；或者所述统计单元以时钟节拍为单位，即统计每个时钟节拍内的强度信息，其中时钟节拍是指一个固定的时间间隔。
例如，当探测器选择SPAD阵列时(即对应于一个像素点的探测 器包括多个SPAD)，接收信号包括一次测量时间内(发射和接收N次光信号)探测器中点亮的SPAD数量，统计单元为时钟节拍，所述一次测量时间包括多个连续的时钟节拍，每个统计单元计算该时间段内探测器中点亮的SPAD数量作为该统计单元下接收到的光信号的强度值。又例如，当探测器选择SiPM时，探测器输出的接收信号为模拟信号，其中模拟信号的大小作为接收到的光信号的强度值(为便于后续计算需要将模拟信号转换为数字信号)，统计单元为时间点或者时间节拍；当以时间点为统计单元时，该时间点模拟信号的大小即为该统计单元的强度值；当以时钟节拍为统计单元时，该时钟节拍对应的时间段内模拟信号大小的统计值(如均值或者中位值或者其他统计方法)即为该统计单元的强度值。
与传统的计算均值、方差时的统计方式的区别在于，在本发明实施例中，只需采集前向历史值作为参考，即只考虑本统计单元和之前统计单元的数据，而非采集所有统计单元的数据后再进行计算。
具体地，对于每个统计单元，将所述接收信号的当前统计单元之前的历史统计单元的强度统计值作为该统计单元的直流分量，将当前统计单元的强度值减去所述直流分量作为该统计单元的交流分量。
统计单元的直流分量和交流分量可以表征为：
DC＝Funcdi＝[1，n](F2(i))  (1)
AC＝Funcai＝[1，n](DC(i))  (2)
上式(1)、(2)中DC、AC分别表示直流分量、交流分量当前统计单元的统计值，即直流分量和交流分量。其中，i表示统计单元的序号，F2(i)表示第i个统计单元的强度。Funcd()为自定义函数(例如平均、中位值、方差统计等均可)，Funca()为该统计单元的强度值减去直流分量。
需要说明的是，对于初始统计单元，可以认为其直流分量DC为0。
这样，可以逐个统计单元计算统计值，也就是说，在每个统计单元所属时间内，即可获得当前统计单元的统计值，这种方式不仅可以极大提高信号处理的实时性，而且还可节省额外的存储开销。
步骤203，根据所述接收信号的脉冲信息筛选所述多个脉冲得到候选脉冲，所述脉冲信息包括所述直流分量和所述交流分量。
在对所述多个脉冲进行筛选时，可以先选出一定数量峰值强度最高的脉冲，即所述第一筛选脉冲，然后再根据这些脉冲的脉冲宽度和信噪比确定候选脉冲。
在实际应用中，可以记录前N个最强脉冲，并记录其时间信息，若新脉冲的强度大于N个记录中强度最小的脉冲，则用新脉冲替换掉原最小强度的脉冲，以此获得前N个最强脉冲。
比如，记录4个最强脉冲，其对应的时间信息和强度信息如下：
PD0：PD0_TOF＝[1780(开始时间)，1806(峰值时刻)，1826(结束时间)]
PD1：PD1_TOF＝[1038(开始时间)，1049(峰值时刻)，1058(结束时间)]
PD2：PD2_TOF＝[1990(开始时间)，2007(峰值时刻)，2017(结束时间)]
PD3：PD3_TOF＝[1914(开始时间)，1924(峰值时刻)，1935(结束时间)]
PD0_peak＝[24.826171875(强度值)]
PD1_peak＝[23.866015625(强度值)]
PD2_peak＝[36.39296875(强度值)]
PD3_peak＝[25.03359375(强度值)]
上述4个最强脉冲对应的波形图如图3所示。
然后，再从记录的4个最强脉冲中选出所述候选脉冲。具体地，计算每个所述第一筛选脉冲的脉冲宽度和信噪比，如果第一筛选脉冲的脉冲宽度小于第一阈值，且所述第一筛选脉冲的信噪比小于第二阈值，则删除所述第一筛选脉冲，否则保留所述第一筛选脉冲为候选脉冲。
脉冲宽度可以根据脉冲对应的时间信息即脉冲的开始时间、结束时间计算得到，脉冲的信噪比可以根据脉冲强度值中的直流分量和交流分量计算得到。
脉冲的信噪比和脉冲宽度可表征如下：
SNR(n)＝(peak(n)-dc)/ac (3)
PW(n)＝tof_end(n)-tof_start(u) (4)
其中，SNR(n)表示第n个脉冲的信噪比，peak(n)表示该脉冲的强度，dc和ac分别表示该脉冲内包括的统计单元的平均直流分量和平均交流分量；
其中，PW(n)表示第n个脉冲的脉冲宽度，tof_end(n)表示该脉冲的结束时间，tof_start(n)表示该脉冲的开始时间。
在具体应用中，对所述第一筛选脉冲可以首先判断其脉冲宽度，若脉冲宽度大于等于第一阈值，则保留该第一筛选脉冲；如果脉冲的宽度小于第一阈值，则再次对其进行信噪比的判断，如果信噪比小于第二阈值，则删除该第一筛选脉冲(判断为噪声或假性回波)，否则保留该第一筛选脉冲。
上述对脉冲宽度和信噪比的判断主要是考虑到近距离(如10米以内)回波拖尾较长且接近饱和以及脉冲宽度较大导致的信噪比过低的情况。由于统计单元的直流分量是当前统计单元之前的所有单元的强度的平均值，因此在近距离的场景下，脉冲的强度很高且宽度很宽，计算得到的直流分量较大，交流分量较小，信噪比比较低。如图4所示，由于脉冲的宽度较宽，每个统计单元的强度基本差异很小，所以 计算得到的dc约等于该脉冲的强度值，即dc很大，ac会很小，进而计算得到的信噪比很小。如果仅依据信噪比筛选确定所述候选脉冲，则会漏掉这部分脉冲。通过增加对脉冲宽度的判断，则可避免这种情况，将有可能是近距离的回波对应的脉冲保留下来。
步骤204，从所述候选脉冲中选择一个脉冲的到达时间作为TOF值。
从候选脉冲中选择一个脉冲的到达时间作为TOF值时，可以基于不同的决策原则进行选择，比如可以是但不限于以下几种原则：
(1)最强回波原则：从所述候选脉冲中选择强度值最大的脉冲的到达时间作为TOF值；
(2)第一较强回波原则：从所述候选脉冲中选择到达时间最早并且强度值大于设定值的脉冲的到达时间作为TOF值；
(3)最远距离回波原则：从所述候选脉冲中选择到达时间最晚并且强度值大于设定值的脉冲的到达时间作为TOF值。
因为点云图每个视场角度位置上通常只用一个深度或距离值来表征，因此需要挑选出一个最可能的脉冲的到达时间作为测量得到的TOF值。为此，在上述过程中，还需要计算各脉冲的到达时间。需要说明的是，在具体应用中，可以在得到各候选脉冲后，再计算每个候选脉冲的到达时间；也可以先计算各脉冲的到达时间，再进行筛选，比如对所述多个脉冲中一定数量峰值强度最高的脉冲，即上述第一筛选脉冲，计算所述第一筛选脉冲的到达时间，然后再确定所述候选脉冲。
在本发明实施例中，每个脉冲的到达时间T的计算可与信号检测形成前后级流水运算，无需等到所有脉冲检测完成再启动该计算。即当脉冲0的特征值获取完，便可进行脉冲0的到达时间T的估算，直至第N个最强脉冲分析完。
具体地，如图5所示，示出了本发明实施例中计算每个第一筛选 脉冲或每个候选脉冲的到达时间的流程图，包括以下步骤：
在步骤501，根据脉冲的峰值时刻确定所述脉冲的边沿时刻。
在本发明实施例中，可以通过选取脉冲峰值时刻点之前有限时间差内的高频分量进行分析，该有限时间差的大小可以根据需要设定，相当于一个TOF方向的摆动阈值，考虑到计算资源的限制以及计算需要一定的时间，该阈值不宜过大或过小，具体可根据计算资源及设备所需的TOF误差抖动要求来设定，比如可以将该阈值设置为12个统计单元。另外，对于不同的TOF距离，可以设置不同的阈值，以细化适应不同远近距离的TOF误差抖动摆幅。
具体地，确定所述脉冲的峰值时刻和所述脉冲的开始时刻，并确定距离所述峰值时刻前预设时间差的时刻为阈值时刻；判断所述开始时刻是否在所述阈值时刻至所述峰值时刻之间；若是，则确定所述开始时刻为所述脉冲的边沿时刻；若否，则确定所述阈值时刻为所述脉冲的边沿时刻。
在步骤502，根据所述边沿时刻至所述峰值时刻的时间段内脉冲的强度计算到达强度。
也就是说，将脉冲的开始时刻至边沿时刻之间的信号看作噪声，不参与后续到达时间T的计算，将脉冲的边沿时刻至峰值时刻之间的信号看作该脉冲，参与后续到达时间T的计算。
具体地，可以计算所述边沿时刻至所述峰值时刻之间的脉冲强度的中位值或均值或峰值的某一百分比(比如40％)作为到达强度Ft。
在步骤503，根据所述到达强度计算所述脉冲的到达时间。
具体地，可以确定与所述到达强度最近的两个强度为第一强度和第二强度，确定所述第一强度对应的时刻为第一时刻，所述第二强度对应的时刻为第二时刻；根据所述到达强度、所述第一强度和所述第二强度计算得到时间权重；根据所述第一时刻、所述第二时刻和所述时间权重计算得到所述脉冲的到达时间。
比如，寻找与到达强度最近的前后时刻两个强度为F3(i-1)、F3(i)，通过计算到达强度Ft与F3(i-1)、F3(i)之间的强度差作为距离差获取权重，并插值得到强度Ft的对应对达时间T。
插值算式可以表征如下：
其中，(i-1)和(i)指该点的时间值，绝对值表示的部分为各时间值的权重。
当然，本发明实施例并不局限于上式(5)的TOF估算插值算子，也可是线性插值等。
本发明提供的激光雷达回波信号处理方法，将接收信号中的每个统计单元的强度值分解为直流分量和交流分量，根据这些信息对接收信号中的多个脉冲进行筛选，筛选出最有可能的候选脉冲，最后从候选脉冲中选择一个脉冲的到达时间作为TOF值。本发明方案在噪声计算、以及有效回波探测方面具有很好的性能表现，通过将每个统计单元的强度值分解成直流分量和交流分量，进而根据每个统计单元的直流分量和交流分量进行波形重塑、信噪比计算以及候选脉冲的筛选等步骤，在低信噪比的情况下也可以筛选出有效回波进行到达时间的计算，如远距离(回波信号弱)低信噪比有效回波探测；并且增加了通过脉宽筛选候选脉冲的条件，可以进一步的在低信噪比的情况下也可以筛选出有效回波进行到达时间的计算，如超近距离(回波信号强)宽饱和有效回波探测。相较于传统的通过固定强度筛选有效脉冲非方法，本申请的方案大大提高了dTOF回波处理能力。另外，与现有技术相比，本发明方案的处理流程处理为单纯的流水线处理结构，易于芯片化实现或FPGA(Field Programmable Gate Array，可编程阵列逻辑)实现，且硬化实现逻辑简单，不需要过多的存储单元，降低了硬件实现的功耗及开销，降低了应用的复杂度，可以极大提升产品的工程化与产品化。
在另一种非限制性实施例中，本发明激光雷达回波信号处理方法还可进一步对上述步骤204中得到的TOF值进行校准，比如，可以将对应各TOF值的补偿值写入相应的列表中，通过查表方式来得到校准后的TOF值。具体地，预先设置TOF补偿值对应表；根据所述TOF值查找所述TOF补偿值对应表，得到与所述TOF值最接近的两个TOF值对应的补偿值；根据所述最接近的两个TOF值对应的补偿值对所述TOF值进行校准。
前面提到，在上述步骤201中所述的接收信号是指对激光雷达探测单元产生的电信号经过模数转换后得到的数字信号。基于不同的探测器，所述探测单元也会有所不同，比如可以是：SPAD(Single Photon Avalanche Diode，单光子雪崩二极管)、SiPM(Silicon photomultiplier，硅光电倍增管)、APD(Avalanche Photo Diode，雪崩光电二极管)等，对此本发明实施例不做限定。
在本发明实施例中，所述电信号具体可以是关联时间及关联空间内探测单元产生的电信号，也就是说，将关联时间及关联空间内探测单元产生的电信号转换后得到的数字信号，所述数字信号即为步骤201中所述的接收信号。
其中，所述关联空间包括：回波信号在探测单元阵列中对应的理论区域以及相邻区域，所述相邻区域与所述理论区域不重叠或部分重叠，而且相邻区域相对于理论区域的位置不做限定，比如可以是上、下、左、右、斜角方向等均可。另外，相邻区域和理论区域可以是任意形状的探测器阵列，比如矩形、圆形、多边形等，对此本发明实施例不做限定。需要说明的是，所述回波信号为激光雷达在某一视场角度下对应的回波信号，所述理论区域为光阑未发生偏移时该视场角度下在探测单元阵列中对应的区域。
其中，所述关联时间与所述统计单元相对应。
在具体应用中，将上述电信号转换为数字信号可以采用多种方式来实现，比如：
(1)利用时间数字转换方法将所述电信号转换为数字信号；
(2)利用模数转换方法将所述电信号转换为数字信号。
下面对上述两种方式转换方式分别进行详细说明。
如图6所示，是本发明实施例中利用时间数字转换方法将电信号转换为数字信号的流程图，包括以下步骤：
步骤601，基于关联时间及关联空间内点亮的探测单元数量，生成一组直方图。
比如，在一种非限制性实现方式中，以探测器采用SPAD接收阵列为例，可以基于关联时间及关联空间内单光子计数机理，用统计单元时间段内激活的单光子计数来表征该统计单元接收的信号强度，若干个这样的统计单元接收的信号强度在时间轴上连续关联起来便形成一组直方图。也就是说，计算每个所述统计单元中关联空间内点亮的所述探测单元总数，根据每个所述统计单元的所述总数，生成一组直方图。如图7所示，为2048个纳秒连续关联时间上形成的直方图示意图。其中，关联空间为理想情况下回波光斑覆盖的SPAD接收阵列的理论区域。考虑到由于设备装调工艺偏差，可能会导致通过光阑的光斑偏离SPAD阵列上正确的接收位置，关联空间还包括与该理论区域相邻的相邻区域的SPAD阵列。计算每个所述统计单元中关联空间内点亮的所述探测单元总数，即计算每个统计单元对饮的时间内，包含理论区域和相邻区域内的关联空间中点亮的探测单元总数，并生成直方图。
考虑到前述光斑偏离SPAD阵列上正确的接收位置，在另一种非限制性实现方式中，还可以采用搜索区域匹配累加的光子计数方法生成所述直方图。具体地，统计每个统计单元中理论区域内点亮的探测单元的数量，以及统计每个统计单元中相邻区域内点亮的探测单元的数量，根据所述理论区域内点亮的探测单元的数量和所述相邻区域内点亮的探测单元的数量，生成一组直方图。即分别统计理论区域和相 邻区域内点亮的探测单元总数，分别将统计的数据生成在一组直方图中对应的统计单元中。
在另一种非限制性实现方式中，分别统计每个统计单元中理论区域内点亮的探测单元的数量，以及统计每个统计单元中相邻区域内点亮的探测单元的数量，根据所述理论区域内点亮的探测单元的数量形成一张初始直方图，根据每个所述相邻区域内点亮的探测单元的数量各自形成对应的初始直方图，将各个所述初始直方图分配不同的权重，加权形成理论区域对应视场的直方图，其中理论区域对应的初始直方图权重高于相邻区域的初始直方图权重。例如，如图8所示，假设位于SPAD阵列中心区域的Macro-Pixel(图8中九宫格中心，对应光斑C位置的区域)为理论区域，其环绕其周围的8个Macro-Pixel为相邻区域，理论区域和8个相邻区域各自生成对应初始直方图，将理论区域和8个相邻区域对应初始直方图叠加形成理论区域的直方图，并根据该直方图进行TOF计算，其中理论区域的初始直方图权重为50％，每个相邻区域的初始直方图的权重为6.25％。采用上述方法，可以在光阑发生偏移等情况下保证探测器端能准确获取到回波信号，此外由于通常情况下回波信号落在理论区域的概率更大，因此为理论区域的初始直方图分配更高的权重比，此方法还可以抵消因为扩大探测器的区域范围(由原始的理论区域扩大到理论区域加相邻区域)而造成的环境光噪声增大的问题，可以提高回波信号处理的鲁棒性。
比如图8所示，假设一个宏像素(MP，Macro Pixel)由20*12个SPAD组成，与光阑尺寸形状相匹配。理想情况下回波光斑正好覆盖一个MP，即所述理论区域，但由于接收偏差等因素，光斑覆盖范围可能产生偏移。若假设水平方向搜索最大左右偏移3/4宏像素水平尺寸，垂直方向搜索最大上下偏移1/2宏像素垂直尺寸，即所述相邻区域，则光斑搜索范围为50*24个SPAD组成的阵列范围，即所述关联空间。当然本发明实施例不局限于以上搜索尺寸与精度大小。
需要说明的是，本发明实施例并不局限参与搜索的区域块数，可由工程实现中做定义。
上述搜索区域匹配累加的光子计数方法，以某一空间搜索精度考虑多种偏移情况，可以有效实现关联时间内关联空间的单光子计数。步骤602，根据所述直方图得到所述数字信号。
利用上述第(2)种模数转换方法将所述电信号转换为数字信号时，需要获取关联时间及关联空间内探测单元对应的电流信号或电压信号，将所述电流信号或电压信号转换为数字信号。
需要说明的是，为了解决光斑偏移的问题，同样可以采用上述扩大搜索范围的方法，不同的是，采用模数转换方法将所述电信号转换为数字信号时，需要将关联区域内所有探测单元输出的电信号合并采集，再将采集的信号通过模数转换成数字信号。
进一步地，在本发明激光雷达回波信号处理方法另一非限制性实施例中，还可对上述得到的数字信号进行滤波，以平滑信号、消减噪声，同时改善信号波形数据的关联性。此处的滤波主要是滤除波形峰值受噪声影响抖动的情况。如图9a波形所示，图9a中圆圈部分所示为受噪声影响产生的抖动(该脉冲的峰值处被分裂为两个小峰)，上述滤波可以滤除该抖动，形成平滑的脉冲信号。
进一步地，还能够对急剧变化的信号脉冲边沿产生一个小高频过冲，从而有助于区分不同反射率的回波表征情况，尤其对于探测距离较近的脉冲回波，因为近距离脉冲回波信号强度接近饱和，脉冲高度几乎都相同，辨别的难度会增加。如图9b波形所示，图9a中圆圈部分所示为平滑后并增加了小高频过冲的脉冲信号。
因此，通过对数字信号进行滤波，可以进一步提高后续TOF分析计算的准确性。
需要说明的是，本发明实施例对上述滤波的具体运算公式不做限定，只要能起到平滑信号作用即可。
比如，在上述滤波的一种非限制性实现中，滤波运算可以表征如下：

其中，F1表示信号I经过FIR(Finite Impulse Response，有限冲激响应)滤波器滤波后的信号波形，该滤波器卷积系数C1皆为正值，起到低通平滑效果。
其中，F2表示信号F1经过同频半相位重采样插值后的信号，例如，采样第n个点，则采样插值的中心点在第n+0.5的位置，若阶数半径s2设为1，那么参与运算的点有F1(n-1)、F1(n)、F1(n+1)、F1(n+2)，其中与中心点相邻的奇数位置FIR系数为正，而偶数位置系数为负，如此随阶数多少交替下去。
为了提高滤波器的带宽与输出效果，可以使用多层重采样滤波。另外，对于上述公式(8)、(9)中的滤波器阶数、系数、层数(公式(8)和公式(9)重复的次数)不做限定。
比如，对于图7所示的直方图，滤波后的直方图如图10所示。可以看出，与滤波前的直方图(图7)相比，滤波后的直方图(图10)更加平滑，滤除了因噪声而导致的不规则抖动，便于后续的处理和计算。
本发明实施例提供的激光雷达回波信号处理方法，在噪声、假性回波滤除、以及有效回波探测方面都有强劲的性能表现。考虑到接收信号包括了经物体反射的回波信号和干扰信号，因此，对接收信号进行处理，将接收信号在每个统计单元的强度值分解为直流分量和交流分量，利用这些信息筛选出多个最有可能的回波进行TOF计算，进而可以得到探测距离。本发明方案在信噪比较低的两种情况：远距离回波信号弱、以及超近距离宽饱和有效回波，仍然可以筛选出有效回 波进行TOF计算。而且，本发明方案具有较强的鲁棒性与简易实操性，而且实现功耗及开销、复杂度都很低。
相应地，本发明实施例还提供一种激光雷达回波信号处理装置，如图11所示，是该装置的一种结构示意图。
该实施例的激光雷达回波信号处理装置包括以下各模块：
信号获取模块111，用于获取接收信号，所述接收信号包括多个脉冲及其对应的时间信息和强度信息；
直交分量计算模块112，用于将所述多个脉冲在每个统计单元的强度值分解为直流分量和交流分量，所述统计单元为时刻点或时钟节拍；
脉冲检测模块113，用于根据所述接收信号的脉冲信息筛选所述多个脉冲得到候选脉冲，所述脉冲信息包括所述直流分量和所述交流分量；
TOF分析模块114，用于计算各脉冲的到达时间；
TOF决策模块115，用于从所述候选脉冲中选择一个脉冲的到达时间作为TOF值。
需要说明的是，所述接收信号是指对激光雷达探测单元产生的电信号经过模数转换后的数字信号。
其中，所述脉冲检测模块113具体可以从所述多个脉冲中选择峰值强度最高的设定数量的脉冲为第一筛选脉冲；计算每个所述第一筛选脉冲的脉冲宽度，并根据所述直流分量和所述交流分量计算每个所述第一筛选脉冲的信噪比；如果所述第一筛选脉冲的脉冲宽度小于第一阈值，且所述第一筛选脉冲的信噪比小于第二阈值，则删除所述第一筛选脉冲，否则保留所述第一筛选脉冲为候选脉冲。
在具体应用中，所述TOF分析模块114可以计算每个第一筛选脉冲，或者计算每个候选脉冲的到达时间，具体计算方式可参照前面 本发明方法实施例中的描述，在此不再赘述。即TOF分析模块可以在筛选出第一筛选脉冲后对每个第一筛选脉冲的到达时间进行计算，或者在经过第二次筛选得到候选脉冲后再对候选脉冲的到达时间进行计算。
上述TOF决策模块115在从所述候选脉冲中选择一个脉冲的到达时间作为TOF值时，可以基于不同的决策原则进行选择，具体可参照前面本发明方法实施例中的描述，在此不再赘述。
如图12所示，是本发明实施例激光雷达回波信号处理装置的另一种结构示意图。
与图11所示实施例的区别在于，在该实施例中，所述激光雷达回波信号处理装置还包括：重采样滤波模块121，用于对所述接收信号进行滤波，以平滑所述接收信号。
需要说明的是，本发明实施例中对所述重采样滤波模块121具体采用何种滤波算法不做限定，只要能起到平滑信号的作用即可。
如图13所示，是本发明实施例激光雷达回波信号处理装置的另一种结构示意图。
与图12所示实施例相比，激光雷达回波信号处理装置还包括：波形重塑模块131，用于对所述脉冲进行波形重塑，得到重塑后的接收信号。
波形重塑的目的是将每个脉冲的起始与结束点(即脉冲强度)归到零点，以便于后续计算。
为了使脉冲的起始与结束点回归到零点(便于后续运算)，传统的做法是减去统一固定直流电平(即信号均值)，这种方法有一个缺陷是统计的直流电平不一定很理想，尤其相对于每一个当前时刻点，这样就会导致各时刻点减去的直流电平有些是过大、有些会过小，从而会影响TOF插值计算时的信号脉冲强度的准确性。
为此，在本发明实施例中，不需要将每个时刻点的信号强度减去直流电平(或信号均值)来使信号起始点和结束点趋于归零，而是以前面所述的统计单元为单位，保留每个统计单元的直流分量，从而保留了统计单元的原始特性，只需在TOF分析计算时考虑每个统计单元的交流分量或高频分量即可。
本发明实施例中的波形重塑算法主要有两点：一是通过判断当前统计单元的强度值是否处于波谷位置，二是判断当前统计单元的强度值与前一脉冲的峰值强度之差是否高于当前统计单元的交流分量的关联摆幅；若满足这两个条件，则将该统计单元的强度值归为零点。在具体实现时，所述回波重塑包括：判断当前统计单元的信号强度值是否处于波谷，且所述当前统计单元的信号强度与前一脉冲的峰值强度之差是否大于所述当前统计单元的交流分量的关联摆幅，所述交流分量的关联摆幅为所述当前统计单元的交流分量与预设系数的乘积；若是，则将当前统计单元的强度值归为零；若否，则保持当前统计单元的强度值不变。
需要说明的是，前一脉冲的峰值强度可通过前后统计单元作极值递归比较记录下来。统计单元的信号强度值更新可表征为：
其中，alpha为所述预设系数，F3为重塑后的波形。
如图14所示，是本发明实施例激光雷达回波信号处理装置的另一种结构示意图。
与图11所示实施例的区别在于，在该实施例中，所述激光雷达回波信号处理装置还包括：TOF校准模块141，用于对所述TOF值进行校准。
因为通过TOF决策模块，虽然每个视场角位置获得一个TOF精确估算值，但由于设备等固有因素导致该值与理论值之间往往存在一 定的误差，这就需要对每台雷达作一次出厂校准，即对不同距离上的TOF估算值作一次补偿，以使其逼近理论值。
对TOF值的补偿算式可表征如下：
TOF_after_calibration(d)＝TOF_before_calibration(d)+delta(d) (11)
其中，delta(d)表示在探测距离d上的校准补偿值。
在工程实现中，可将该距离换算成TOF值并量化成一个离散的TOF表，这样对待校准的TOF值，通过查表寻找与其最接近的两个TOF项所对应的校准补偿值delta作插值即可。当然，本发明实施例并不局限于使用何种插值方法，具体可由工程实现时定义。
具体的，实际应用中误差的大小与被测物体的反射率大小有关，当物体反射率不同时，在相同距离下对应的校准补偿值不同，因此可以通过校准前的距离以及接收信号的强度计算物体反射率，并在校准表中确定在校准前的距离时与计算的反射率最接近的两个反射率对应的补偿值，通过插值法计算出校准补偿值。
需要说明的是，上述TOF校准模块141同样可应用于图12和图13所示实施例。
进一步地，在本发明装置另一实施例中，还可包括：信号采集模块(未图示)，用于获取关联时间及关联空间内探测单元产生的电信号，将所述电信号转换为数字信号，所述数字信号即为所述信号获取模块获取的接收信号。
在实际应用中，所述信号采集模块可以采用多种方式将所述电信号转换为数字信号，比如，所述信号采集模块可以包括TDC模块或者ADC模块。其中：
所述TDC模块，用于利用时间数字转换方法将所述电信号转换为数字信号；
所述ADC模块，用于利用模数转换方法将所述电信号转换为数 字信号。
上述不同转换方法的具体实现方式在前面本发明方法实施例中有详细说明，在此不再赘述。
本发明提供的激光雷达回波信号处理装置，针对包括经物体反射的回波信号和干扰信号的接收信号中的多个脉冲，将每个统计单元的强度值分解为直流分量和交流分量，根据这些信息对接收信号中的多个脉冲进行筛选，筛选出最有可能的候选脉冲，最后从候选脉冲中选择一个脉冲的到达时间作为TOF值。本发明方案在保障dTOF回波处理能力的同时，降低了硬件实现的功耗及开销，降低了应用的复杂度，可以极大提升产品的工程化与产品化。
相应地，本发明实施例还提供一种激光雷达探测系统，用于根据所述TOF时间确定探测物体的距离。如图15所示，该激光雷达探测系统150包括：发射装置151、接收装置152以及前面各实施例所述的激光回波信号处理装置153。其中：
所述发射装置151，发射光脉冲信号；
所述接收装置152，包括多个探测单元，用于接收光信号生成接收信号；
所述激光回波信号处理装置153，与所述接收装置152相连，用于根据接收信号确定发射光脉冲信号的TOF时间，根据所述TOF时间确定探测物体的距离。
在具体实施中，关于上述实施例中描述的各个装置、产品包含的各个模块/单元，其可以是软件模块/单元，也可以是硬件模块/单元，或者也可以部分是软件模块/单元，部分是硬件模块/单元。
例如，对于应用于或集成于芯片的各个装置、产品，其包含的各个模块/单元可以都采用电路等硬件的方式实现，或者，至少部分模块/单元可以采用软件程序的方式实现，该软件程序运行于芯片内部集成的处理器，剩余的(如果有)部分模块/单元可以采用电路等硬 件方式实现；对于应用于或集成于芯片模组的各个装置、产品，其包含的各个模块/单元可以都采用电路等硬件的方式实现，不同的模块/单元可以位于芯片模组的同一组件(例如芯片、电路模块等)或者不同组件中，或者，至少部分模块/单元可以采用软件程序的方式实现，该软件程序运行于芯片模组内部集成的处理器，剩余的(如果有)部分模块/单元可以采用电路等硬件方式实现；对于应用于或集成于终端的各个装置、产品，其包含的各个模块/单元可以都采用电路等硬件的方式实现，不同的模块/单元可以位于终端内同一组件(例如，芯片、电路模块等)或者不同组件中，或者，至少部分模块/单元可以采用软件程序的方式实现，该软件程序运行于终端内部集成的处理器，剩余的(如果有)部分模块/单元可以采用电路等硬件方式实现。
本发明实施例还提供了一种计算机可读存储介质，所述计算机可读存储介质为非易失性存储介质或非瞬态存储介质，其上存储有计算机程序，所述计算机程序被处理器运行时执行上述图2或图5或图6对应实施例提供的方法的步骤。
本发明实施例还提供了一种电子设备，包括存储器和处理器，所述存储器上存储有可在所述处理器上运行的计算机程序，所述处理器运行所述计算机程序时执行上述图2或图5或图6对应实施例所提供的方法的步骤。
在本发明实施例中，所述处理器可以为中央处理单元(central processing unit，简称CPU)，该处理器还可以是其他通用处理器、数字信号处理器(digital signal processor，简称DSP)、专用集成电路(application specific integrated circuit，简称ASIC)、现成可编程门阵列(field programmable gate array，简称FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
还应理解，本发明实施例中的存储器可以是易失性存储器或非易失性存储器，或可包括易失性和非易失性存储器两者。其中，非易失 性存储器可以是只读存储器(read-only memory，简称ROM)、可编程只读存储器(programmable ROM，简称PROM)、可擦除可编程只读存储器(erasable PROM，简称EPROM)、电可擦除可编程只读存储器(electrically EPROM，简称EEPROM)或闪存。易失性存储器可以是随机存取存储器(random access memory，简称RAM)，其用作外部高速缓存。通过示例性但不是限制性说明，许多形式的随机存取存储器(random access memory，简称RAM)可用，例如静态随机存取存储器(static RAM，简称SRAM)、动态随机存取存储器(DRAM)、同步动态随机存取存储器(synchronous DRAM，简称SDRAM)、双倍数据速率同步动态随机存取存储器(double data rate SDRAM，简称DDR SDRAM)、增强型同步动态随机存取存储器(enhanced SDRAM，简称ESDRAM)、同步连接动态随机存取存储器(synchlink DRAM，简称SLDRAM)和直接内存总线随机存取存储器(direct rambus RAM，简称DR RAM)。
应理解，本文中术语“和/或”，仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。另外，本文中字符“/”，表示前后关联对象是一种“或”的关系。
本发明实施例中出现的“多个”是指两个或两个以上。
本发明实施例中出现的第一、第二等描述，仅作示意与区分描述对象之用，没有次序之分，也不表示本发明实施例中对设备个数的特别限定，不能构成对本发明实施例的任何限制。
本发明实施例中出现的“连接”是指直接连接或者间接连接等各种连接方式，以实现设备间的通信，本发明实施例对此不做任何限定。
虽然本发明披露如上，但本发明并非限定于此。任何本领域技术人员，在不脱离本发明的精神和范围内，均可作各种更动与修改，因此本发明的保护范围应当以权利要求所限定的范围为准。

Claims (27)

  1. 一种激光雷达回波信号处理方法，其特征在于，所述方法包括：
    获取接收信号，所述接收信号包括多个脉冲及其对应的时间信息和强度信息；
    将所述接收信号在每个统计单元的强度值分解为直流分量和交流分量，所述统计单元为时刻点或时钟节拍；
    根据所述接收信号的脉冲信息筛选所述多个脉冲得到候选脉冲，所述脉冲信息包括所述直流分量和所述交流分量；
    从所述候选脉冲中选择一个脉冲的到达时间作为TOF值。
  2. 根据权利要求1所述的方法，其特征在于：
    所述每个统计单元的直流分量为所述接收信号的当前统计单元之前的历史统计单元的强度统计值；
    所述每个统计单元的交流分量为所述当前统计单元的强度值减去所述直流分量。
  3. 根据权利要求2所述的方法，其特征在于，所述方法还包括对所述脉冲进行波形重塑，得到重塑后的接收信号，所述波形重塑包括：
    判断当前统计单元的信号强度值是否处于波谷，且所述当前统计单元的信号强度与前一脉冲的峰值强度之差是否大于所述当前统计单元的交流分量的关联摆幅，所述交流分量的关联摆幅为所述当前统计单元的交流分量与预设系数的乘积；
    若是，则将当前统计单元的强度值归为零；
    若否，则保持当前统计单元的强度值不变。
  4. 根据权利要求2所述的方法，其特征在于，所述根据所述接收信号的脉冲信息筛选所述多个脉冲得到候选脉冲包括：
    从所述多个脉冲中选择峰值强度最高的设定数量的脉冲为第一筛选脉冲；
    计算每个所述第一筛选脉冲的脉冲宽度，并根据所述直流分量和所述交流分量计算每个所述第一筛选脉冲的信噪比；
    如果所述第一筛选脉冲的脉冲宽度小于第一阈值，且所述第一筛选脉冲的信噪比小于第二阈值，则删除所述第一筛选脉冲，否则保留所述第一筛选脉冲为候选脉冲。
  5. 根据权利要求4所述的方法，其特征在于，所述方法还包括：
    计算每个所述第一筛选脉冲或每个所述候选脉冲的到达时间。
  6. 根据权利要求5所述的方法，其特征在于，所述从所述候选脉冲中选择一个脉冲的到达时间作为TOF值包括：
    从所述候选脉冲中选择强度值最大的脉冲的到达时间作为TOF值；或者
    从所述候选脉冲中选择到达时间最早并且强度值大于设定值的脉冲的到达时间作为TOF值；或者
    从所述候选脉冲中选择到达时间最晚并且强度值大于设定值的脉冲的到达时间作为TOF值。
  7. 根据权利要求5所述的方法，其特征在于，所述计算每个所述第一筛选脉冲或每个所述候选脉冲的到达时间包括：
    根据所述脉冲的峰值时刻确定所述脉冲的边沿时刻；
    根据所述边沿时刻至所述峰值时刻的时间段内脉冲的强度计算到达强度；
    根据所述到达强度计算所述脉冲的到达时间。
  8. 根据权利要求7所述的方法，其特征在于，所述根据所述脉冲的峰值时刻确定所述脉冲的边沿时刻包括：
    确定所述脉冲的峰值时刻和所述脉冲的开始时刻；
    确定距离所述峰值时刻前预设时间差的时刻为阈值时刻；
    判断所述开始时刻是否在所述阈值时刻至所述峰值时刻之间；
    若是，则确定所述开始时刻为所述脉冲的边沿时刻；
    若否，则确定所述阈值时刻为所述脉冲的边沿时刻。
  9. 根据权利要求7所述的方法，其特征在于，所述根据所述边沿时刻至所述峰值时刻的时间段内脉冲的强度计算到达强度包括：
    计算所述边沿时刻至所述峰值时刻之间的脉冲强度的中位值或均值为到达强度。
  10. 根据权利要求7所述的方法，其特征在于，所述根据所述到达强度计算所述脉冲的到达时间包括：
    确定与所述到达强度最近的两个强度为第一强度和第二强度，确定所述第一强度对应的时刻为第一时刻，所述第二强度对应的时刻为第二时刻；
    根据所述到达强度、所述第一强度和所述第二强度计算得到时间权重；
    根据所述第一时刻、所述第二时刻和所述时间权重计算得到所述脉冲的到达时间。
  11. 根据权利要求1所述的方法，其特征在于，所述方法还包括：
    对所述接收信号进行滤波，以平滑所述接收信号。
  12. 根据权利要求1所述的方法，其特征在于，所述方法还包括：
    对所述TOF值进行校准。
  13. 根据权利要求1至12任一项所述的方法，其特征在于，所述方法还包括：
    获取关联时间及关联空间内探测单元产生的电信号；
    将所述电信号转换为数字信号，所述数字信号为所述接收信号。
  14. 根据权利要求13所述的方法，其特征在于，所述关联空间包括：回波信号在探测单元阵列中对应的理论区域以及相邻区域，所述相邻区域与所述理论区域不重叠或部分重叠；
    所述关联时间与所述统计单元对应。
  15. 根据权利要求14所述的方法，其特征在于，所述将所述电信号转换为数字信号包括：
    利用时间数字转换方法将所述电信号转换为数字信号；或者
    利用模数转换方法将所述电信号转换为数字信号。
  16. 根据权利要求15所述的方法，其特征在于，所述利用时间数字转换方法将所述电信号转换为数字信号包括：
    基于关联时间及关联空间内点亮的所述探测单元的数量，生成一组直方图；
    根据所述直方图得到所述数字信号。
  17. 根据权利要求16所述的方法，其特征在于，所述基于关联时间及关联空间内点亮的所述探测单元的数量，生成一组直方图包括：
    计算每个所述统计单元中关联空间内点亮的所述探测单元总数，根据每个所述统计单元的所述总数，生成一组直方图；或者
    统计每个所述统计单元中所述理论区域内点亮的探测单元的数量，以及，统计每个所述统计单元中所述相邻区域内点亮的探测单元的数量，根据所述理论区域内点亮的探测单元的数量和所述相邻区域 内点亮的探测单元的数量，生成一组直方图。
  18. 根据权利要求15所述的方法，其特征在于，所述利用模数转换方法将所述电信号转换为数字信号包括：
    获取关联时间及关联空间内探测单元对应的电流信号或电压信号，将所述电流信号或电压信号转换为数字信号。
  19. 一种激光雷达回波信号处理装置，其特征在于，所述装置包括：
    信号获取模块，用于获取接收信号，所述接收信号包括多个脉冲及其对应的时间信息和强度信息；
    直交分量计算模块，用于将所述接收信号在每个统计单元的强度值分解为直流分量和交流分量，所述统计单元为时刻点或时钟节拍；
    脉冲检测模块，用于根据所述接收信号的脉冲信息筛选所述多个脉冲得到候选脉冲，所述脉冲信息包括所述直流分量和所述交流分量；
    TOF分析模块，用于计算各脉冲的到达时间；
    TOF决策模块，用于从所述候选脉冲中选择一个脉冲的到达时间作为TOF值。
  20. 根据权利要求19所述的装置，其特征在于，所述装置还包括：
    重采样滤波模块，用于对所述接收信号进行滤波，以平滑所述接收信号。
  21. 根据权利要求19所述的装置，其特征在于，所述装置还包括：
    波形重塑模块，用于对所述脉冲进行波形重塑，得到重塑后的接收信号；
    所述波形重塑包括：判断当前统计单元的信号强度值是否处于波谷，且所述当前统计单元的信号强度与前一脉冲的峰值强度之差是否大于所述当前统计单元的交流分量的关联摆幅，所述交流分量的关联摆幅为所述当前统计单元的交流分量与预设系数的乘积；若是，则将当前统计单元的强度值归为零；若否，则保持当前统计单元的强度值不变。
  22. 根据权利要求19所述的装置，其特征在于：
    所述脉冲检测模块，具体用于从所述多个脉冲中选择峰值强度最高的设定数量的脉冲为第一筛选脉冲；计算每个所述第一筛选脉冲的脉冲宽度，并根据所述直流分量和所述交流分量计算每个所述第一筛选脉冲的信噪比；如果所述第一筛选脉冲的脉冲宽度是否小于第一阈值，且所述第一筛选脉冲的信噪比小于第二阈值，则删除所述第一筛选脉冲，否则保留所述第一筛选脉冲为候选脉冲。
  23. 根据权利要求22所述的装置，其特征在于：
    所述TOF分析模块，具体用于计算每个所述第一筛选脉冲，或者计算每个所述候选脉冲的到达时间。
  24. 根据权利要求19所述的装置，其特征在于，所述装置还包括：
    TOF校准模块，用于对所述TOF值进行校准。
  25. 根据权利要求19至24任一项所述的装置，其特征在于，所述信号获取模块包括：
    信号采集模块，用于获取关联时间及关联空间内探测单元产生的电信号，将所述电信号转换为数字信号，所述数字信号为所述接收信号。
  26. 根据权利要求25所述的装置，其特征在于，所述信号采集模块包括TDC模块或者ADC模块；
    所述TDC模块，用于利用时间数字转换方法将所述电信号转换为数字信号；
    所述ADC模块，用于利用模数转换方法将所述电信号转换为数字信号。
  27. 一种激光雷达探测系统，用于根据TOF时间确定探测物体的距离，其特征在于，所述系统包括：发射装置、接收装置以及如权利要求19至26任一项所述的激光回波信号处理装置；
    所述发射装置，用于发射光脉冲信号；
    所述接收装置，包括多个探测单元，用于接收光信号生成接收信号；
    所述激光回波信号处理装置，与所述接收装置相连，用于根据接收信号确定发射光脉冲信号的TOF时间，根据所述TOF时间确定探测物体的距离。
PCT/CN2023/079470 2022-08-26 2023-03-03 激光雷达回波信号处理方法及装置、激光雷达探测系统 WO2024040912A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211035396.4A CN117665741A (zh) 2022-08-26 2022-08-26 激光雷达回波信号处理方法及装置、激光雷达探测系统
CN202211035396.4 2022-08-26

Publications (1)

Publication Number Publication Date
WO2024040912A1 true WO2024040912A1 (zh) 2024-02-29

Family

ID=90012268

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/079470 WO2024040912A1 (zh) 2022-08-26 2023-03-03 激光雷达回波信号处理方法及装置、激光雷达探测系统

Country Status (2)

Country Link
CN (1) CN117665741A (zh)
WO (1) WO2024040912A1 (zh)

Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005227167A (ja) * 2004-02-13 2005-08-25 Mitsubishi Electric Corp 測位計算精度の向上手段
CN102565781A (zh) * 2010-11-19 2012-07-11 株式会社电装 雷达设备
CN105640497A (zh) * 2014-11-28 2016-06-08 佳能株式会社 信号处理方法、声波处理装置和记录介质
EP3370079A1 (en) * 2017-03-01 2018-09-05 STMicroelectronics (Grenoble 2) SAS Range and parameter extraction using processed histograms generated from a time of flight sensor - pulse detection
WO2018181013A1 (ja) * 2017-03-29 2018-10-04 株式会社デンソー 光検出器
US20200271765A1 (en) * 2017-09-22 2020-08-27 Ams Ag Method for calibrating a time-of-flight system and time-of-flight system
CN111679290A (zh) * 2020-06-04 2020-09-18 上海禾赛光电科技有限公司 光子计数校正方法、激光雷达以及计算机可读介质
CN111868560A (zh) * 2018-03-20 2020-10-30 帕诺森斯有限公司 取决于脉冲类型选择lidar脉冲检测器
WO2021243612A1 (zh) * 2020-06-03 2021-12-09 深圳市大疆创新科技有限公司 测距方法、测距装置和可移动平台
CN114488175A (zh) * 2022-01-21 2022-05-13 深圳市灵明光子科技有限公司 一种提高信噪比的直方图构造方法及激光测距芯片
WO2022160610A1 (zh) * 2021-01-28 2022-08-04 深圳奥锐达科技有限公司 一种飞行时间测距方法、系统和设备
WO2022161481A1 (zh) * 2021-01-28 2022-08-04 深圳奥锐达科技有限公司 一种飞行时间测距方法、系统和设备

Patent Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005227167A (ja) * 2004-02-13 2005-08-25 Mitsubishi Electric Corp 測位計算精度の向上手段
CN102565781A (zh) * 2010-11-19 2012-07-11 株式会社电装 雷达设备
CN105640497A (zh) * 2014-11-28 2016-06-08 佳能株式会社 信号处理方法、声波处理装置和记录介质
EP3370079A1 (en) * 2017-03-01 2018-09-05 STMicroelectronics (Grenoble 2) SAS Range and parameter extraction using processed histograms generated from a time of flight sensor - pulse detection
WO2018181013A1 (ja) * 2017-03-29 2018-10-04 株式会社デンソー 光検出器
US20200271765A1 (en) * 2017-09-22 2020-08-27 Ams Ag Method for calibrating a time-of-flight system and time-of-flight system
CN111868560A (zh) * 2018-03-20 2020-10-30 帕诺森斯有限公司 取决于脉冲类型选择lidar脉冲检测器
WO2021243612A1 (zh) * 2020-06-03 2021-12-09 深圳市大疆创新科技有限公司 测距方法、测距装置和可移动平台
CN111679290A (zh) * 2020-06-04 2020-09-18 上海禾赛光电科技有限公司 光子计数校正方法、激光雷达以及计算机可读介质
WO2022160610A1 (zh) * 2021-01-28 2022-08-04 深圳奥锐达科技有限公司 一种飞行时间测距方法、系统和设备
WO2022161481A1 (zh) * 2021-01-28 2022-08-04 深圳奥锐达科技有限公司 一种飞行时间测距方法、系统和设备
CN114488175A (zh) * 2022-01-21 2022-05-13 深圳市灵明光子科技有限公司 一种提高信噪比的直方图构造方法及激光测距芯片

Also Published As

Publication number Publication date
CN117665741A (zh) 2024-03-08

Similar Documents

Publication Publication Date Title
US20240028667A1 (en) Method and apparatus for processing a histogram output from a detector sensor
US20210181317A1 (en) Time-of-flight-based distance measurement system and method
WO2021008209A1 (zh) 深度测量装置及距离测量方法
US20210116572A1 (en) Light ranging apparatus
EP2729826B1 (en) Improvements in or relating to the processing of time-of-flight signals
WO2022160611A1 (zh) 一种基于时间融合的距离测量方法、系统和设备
CN110221274B (zh) 时间飞行深度相机及多频调制解调的距离测量方法
CN110221273B (zh) 时间飞行深度相机及单频调制解调的距离测量方法
CA2650235A1 (en) Distance measuring method and distance measuring element for detecting the spatial dimension of a target
JP6261681B2 (ja) タイムオブフライト信号の処理における又はこれに関する改良
CN110320528A (zh) 时间深度相机及多频调制解调的降低噪声的距离测量方法
WO2020237448A1 (zh) 回波信号处理方法、装置、系统及存储介质
US20220043129A1 (en) Time flight depth camera and multi-frequency modulation and demodulation distance measuring method
JP2015219120A (ja) 距離測定装置
JP2020153799A (ja) 測距装置および測距方法
EP3721261B1 (en) Distance time-of-flight modules
US20220187430A1 (en) Time of flight calculation with inter-bin delta estimation
US20240248181A1 (en) Methods and devices for peak signal detection
US11561291B2 (en) High pulse repetition frequency lidar
WO2024040912A1 (zh) 激光雷达回波信号处理方法及装置、激光雷达探测系统
US20220206146A1 (en) A method of measuring a distance from a target
US20200355806A1 (en) Electronic apparatus and distance measuring method
US20230273321A1 (en) 3D Image Sensor Ranging System, and Ranging Method Using Same
US20240036208A1 (en) Time of flight sensing
US12032065B2 (en) System and method for histogram binning for depth detection

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23856009

Country of ref document: EP

Kind code of ref document: A1