A DTOF histogram ranging method and device for suppressing noise and peak searching

Abstract

This invention provides a method and apparatus for noise suppression and peak finding in DTOF histogram ranging. A first histogram containing ambient light noise is acquired during laser emission, and a second histogram containing pure noise is acquired during an equal-length period of laser non-emission; these two acquisitions are alternated. The first and second histograms are then differentially divided by time bins to obtain a noise-suppressed third histogram. A moving average filter with coefficients preset according to the laser pulse waveform is used to smooth and filter the third histogram, resulting in a fourth histogram. Super-resolution fitting is performed on the fourth histogram to find peaks. A local window is taken centered on the bin with the largest count value, and the peak position is calculated using the centroid method. The position corresponding to the peak with the highest intensity is selected as the time of flight. This invention utilizes the temporal correlation of noise for differential denoising, combined with weighted smoothing filtering of the matched waveform and centroid peak finding, to improve the noise resistance of ranging under strong light conditions without increasing hardware costs.

Description

Technical Field

[0001] This invention relates to the field of laser ranging technology, specifically to a method and apparatus for noise suppression and peak finding in DTOF histogram ranging. Background Technology

[0002] LiDAR, as a core navigation sensor for robots, senses the distance to target objects by measuring the time of flight of laser pulses. In Direct Time-of-Flight (DTOF) LiDAR based on single-photon avalanche diodes (SPADs), the working principle is as follows: a laser emits a laser pulse towards the target object, and a timer records the emission time; the returned light is received by the SPAD receiver, and a timer records the return time; the difference between the two times is the laser's time of flight, which is then used to calculate the target distance. To improve measurement reliability, DTOF LiDAR typically performs multiple repeated ranging measurements, statistically analyzing the time information from these multiple measurements to form a photon count histogram. The time point that appears most frequently in the histogram is selected as the laser's time of flight.

[0003] However, in practical applications, especially in strong outdoor lighting environments, ambient light noise such as sunlight can severely impact the ranging performance of DTOF lidar. Because ambient light noise photons are uniformly distributed throughout the measurement cycle, and earlier arriving noise photons are more likely to trigger the SPAD and be recorded, severe noise buildup occurs in the early stages of the measurement cycle. When the ambient light noise intensity is high, the signal-to-noise ratio of the effective signal peak in the histogram decreases significantly, and it may even be completely submerged by noise, making it extremely difficult to accurately extract the time of flight from the histogram, severely affecting the accuracy and reliability of ranging.

[0004] Existing methods for combating ambient light interference mainly rely on hardware measures such as increasing laser emission power or adding optical filters, but these methods are either costly or increase the size and complexity of the system. Summary of the Invention

[0005] The purpose of this invention is to provide a method and apparatus for noise suppression and peak finding in DTOF histogram ranging, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method and apparatus for noise suppression and peak finding in DTOF histogram ranging, comprising the following steps: S1. Control the lidar to perform multiple laser emission and receive echo signals within a preset measurement period, count the photon count in each time bin, and form a first histogram containing ambient light noise; S2. Within the same time interval as the measurement cycle in S1, control the lidar to not emit laser light and only receive signals, count the photon counts in each time bin, and form a second histogram characterizing pure ambient light noise. S3. Perform a difference operation on the first histogram in S1 and the second histogram in S2 according to the corresponding time bins to obtain a third histogram that initially suppresses ambient light noise. S4. Smooth the third histogram in S3 to obtain a fourth histogram that enhances the signal-to-noise ratio of the signal peak to the noise peak. S5. Perform super-resolution fitting on the fourth histogram in S4 to find the peaks and calculate the precise position of each peak. S6. Among all the peaks obtained by fitting S5, select the position corresponding to the peak with the largest peak intensity as the flight time of the laser.

[0007] Preferably, the acquisition methods of steps S1 and S2 are performed alternately: after each measurement cycle described in step S1, a time interval described in step S2 is immediately executed, and this process is repeated multiple times to generate a first histogram and a second histogram. This allows the acquisition of data with and without laser emission to be closely adjacent in time, effectively counteracting the time-varying fluctuations in ambient light intensity and ensuring a high correlation between the noise distributions in the two histograms, thereby improving the effect of subsequent differential denoising.

[0008] Preferably, the difference operation in step S3 is to subtract the count value of the second histogram from the count value of the first histogram for each bin according to the same time bin number, and the difference between each bin constitutes the third histogram, thus ensuring the physical rationality of the histogram data.

[0009] Preferably, the smoothing filter in step S4 uses a moving average filter. The shape of the filter coefficients matches the waveform of the laser pulse. This can effectively enhance the intensity of the real signal peak while smoothing noise, significantly improve the intensity ratio of the signal peak to the residual noise peak, and make subsequent peak finding more stable and reliable.

[0010] Preferably, in step S4, the count values ​​of each bin within the filtering window are multiplied by their corresponding coefficients and then summed. This summation is then divided by the total sum of the coefficients to obtain the initial count value of the time bin after filtering. The count values ​​of all time bins after filtering are then phase-corrected by shifting them forward by a phase offset corresponding to the filter order to obtain the final count value of the time bin. This ensures that the total energy of the histogram before and after filtering is conserved, avoids introducing amplitude distortion, and facilitates subsequent peak intensity comparison and threshold determination.

[0011] Preferably, the super-resolution fitting peak finding in step S5 includes: finding the bin with the largest count value in the fourth histogram as the target peak bin; taking N time bins on the left and right as the center of the target peak bin to form a local window of a preset width; and calculating the precise position of the target peak by weighting the count values ​​of each bin within the window; where N is a positive integer greater than or equal to 1, which allows for precise positioning of the target peak with the strongest signal and low computational load; at the same time, the local weighted averaging can effectively smooth single-bin noise, improving the peak finding resolution to sub-bin level accuracy.

[0012] Preferably, in step S5, the centroid-weighted average of the count values ​​of each bin within the window is performed to calculate the peak position; the formula for the weighted average is: ,in For the calculated peak position, The time bin number within the local window. The photon count value corresponding to the time bin is applicable to the centroid calculation for any N value. The peak position is calculated using this centroid method formula, and the count value is used as the weight for weighted averaging. This can accurately reflect the energy centroid position of the signal peak. It still has high positioning stability under low signal-to-noise ratio conditions and can obtain complete peak parameter information, providing sufficient data support for subsequent multi-target discrimination, false peak elimination, and comprehensive time-of-flight determination in complex scenarios.

[0013] Preferably, in step S6, multiple peaks are obtained by fitting, and each peak is recorded with at least the peak intensity, peak position, full width at half maximum (FWHM), and baseline information.

[0014] A device for noise suppression and peak finding in DTOF histogram ranging, comprising: A laser emitter is used to emit laser pulses toward a target object. SPAD receiver sensor, SPAD receiver sensor is used to receive echo photons reflected by the target object and record the trigger time; A processor, connected to the laser emitter and the SPAD receiving sensor, is configured to: within a preset measurement period during which the laser emitter emits laser light, count photons in each time bin based on the trigger time recorded by the SPAD receiving sensor, forming a first histogram; within a time interval of the same duration as the measurement period during which the laser emitter does not emit laser light, count photons in each time bin based on the trigger time recorded by the SPAD receiving sensor, forming a second histogram; perform a difference operation between the first histogram and the second histogram according to the corresponding time bins to obtain a third histogram; perform a smoothing filter on the third histogram to obtain a fourth histogram; perform super-resolution fitting on the fourth histogram to find peaks, and select the position corresponding to the peak with the largest peak intensity from the obtained peaks as the time of flight of the laser light.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This invention alternately acquires a first histogram with laser emission and a second histogram without laser emission. By utilizing the high correlation between ambient light noise in the time domain, the two histograms are differentially processed according to time bins. This process can remove the ambient light noise substrate from the histogram, improving the distinction between the true signal peak and the noise peak, and reducing the probability of selecting the wrong peak under strong light conditions.

[0016] Based on differential denoising and smoothing filtering, this invention uses the centroid method to calculate the peak position of the signal peak by local weighted averaging, and uses the count value as the weight to calculate the energy centroid position of the signal peak, which can smooth single-bin noise.

[0017] This invention employs a moving average filter with preset coefficients based on the characteristics of the laser pulse waveform for smoothing. The shape of the filter coefficients matches the laser pulse waveform, which can enhance the intensity of the real signal peak while suppressing noise, improve the signal-to-noise ratio of the signal peak relative to the residual noise peak, and make the subsequent peak selection logic more robust.

[0018] All processing steps of this invention can be implemented on the existing DTOF lidar hardware platform through software algorithms, without the need to add additional optical filters, increase laser emission power, or modify the SPAD device structure. Under the premise of achieving the same noise immunity performance, it has the advantages of low cost, small size, and easy integration. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the detection optical path of the device of the present invention.

[0020] Figure 2 This is the first histogram of the present invention.

[0021] Figure 3 This is the second histogram of the present invention.

[0022] Figure 4 This is the third histogram of the present invention.

[0023] Figure 5 This is the fourth histogram of the present invention.

[0024] Figure 6 This is a schematic diagram of the peak values ​​after filtering according to the present invention.

[0025] Figure 7 This is a schematic diagram of the peak values ​​before filtration in this invention.

[0026] Figure 8 This is a distribution diagram of the filter weights in this invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installed," "equipped with," "sleeved with," "connected," etc., should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.

[0029] Example: Please see Figures 1 to 8 The present invention provides a technical solution: A method and apparatus for noise suppression and peak finding in DTOF histogram ranging, comprising the following steps: S1. The lidar is controlled to perform multiple laser emission and echo signal reception within a preset measurement period, and the photon count in each time bin is counted to form a first histogram including ambient light noise. The lidar's laser emitter emits multiple laser pulses at a certain frequency towards the target object within the preset measurement period. Each laser pulse is reflected by the target object to form an echo signal, which is received by the SPAD receiver sensor and generates a trigger signal. The time-to-digital conversion circuit converts the time of each SPAD trigger into a corresponding time bin, and the histogram counting module accumulates the number of triggers within each time bin. After accumulating multiple laser emission and reception within this measurement period, a first histogram is finally formed with the time bin as the horizontal axis and the photon count value as the vertical axis. Since the SPAD receiver sensor also receives ambient light photons while receiving the echo signal, the first histogram includes both the laser signal peak reflected by the target object and the count distribution of ambient light noise in each time bin. The horizontal axis represents the time unit, reflecting the time point when the SPAD receives the signal, and the vertical axis represents the signal strength, reflecting the signal strength corresponding to the time point when the signal was received.

[0030] S2. After completing the acquisition of the measurement cycle described in S1, a time interval with the same duration as the S1 measurement cycle is entered. During this time interval, the laser emitter is kept off, and no laser pulses are emitted towards the target object. At this time, the SPAD receiving sensor only receives ambient light photons within the field of view and generates a trigger signal when triggered. The time-to-digital conversion circuit converts the time of each trigger into a corresponding time bin, and the histogram statistics module accumulates and counts the number of triggers within each time bin. After multiple accumulations within this time interval, a second histogram is formed with the time bin as the horizontal axis and the photon count value as the vertical axis. Since no laser is emitted during this time interval, all triggers recorded by the SPAD receiving sensor are caused by ambient light. Therefore, the second histogram represents the pure statistical distribution of ambient light noise within each time bin. This pure noise distribution corresponds to the ambient light noise component contained in the first histogram in S1 in terms of time statistical characteristics, providing a reference baseline for subsequent differential denoising.

[0031] A complete acquisition process consists of multiple acquisition cycles. In each acquisition cycle, the measurement cycle described in S1 is executed once, where the laser emitter emits laser pulses, the SPAD receiver receives echo signals and ambient light signals, and the data corresponding to the laser-containing histogram is generated. Immediately following, the time interval described in S2 is executed once, where the laser emitter is turned off, the SPAD receiver only receives ambient light signals, and the data corresponding to the laser-free histogram is generated. After the acquisition is completed, the data acquired in each measurement cycle of S1 are accumulated or averaged to form the final first histogram; the data acquired in each time interval of S2 are accumulated or averaged to form the final second histogram.

[0032] By employing the alternating acquisition method described above, acquisitions with and without laser emission are time-closely adjacent. In practical applications, ambient light intensity may change slowly or fluctuate instantaneously. If all acquisitions in S1 are concentrated in the same time period and all acquisitions in S2 are concentrated in another time period, the time interval between the two sets of acquisitions will be large, and the ambient light conditions may have changed, leading to a decrease in the consistency of noise distribution in the two histograms, thus affecting the subsequent differential denoising effect. However, through alternating acquisition, the acquisition times of each S1 and S2 set are close, and the statistical distribution of ambient light in the two sets of acquisitions maintains a high degree of consistency. When the ambient light intensity fluctuates, this fluctuation will be reflected approximately synchronously in S1 and S2 of the acquisition cycle, and can be canceled out after differential operation. Therefore, the alternating acquisition method ensures a high correlation between the ambient light noise distribution in the first and second histograms, providing a more accurate noise reference baseline for the subsequent differential denoising in S3.

[0033] S3 and the first histogram are acquired by S1 during laser emission. The photon counts in each time bin include both laser echo signal counts and ambient light noise counts. The second histogram is acquired by S2 during the laser off period within the same time interval. The photon counts in each time bin only include ambient light noise counts. Since S1 and S2 use an alternating acquisition method, the acquisitions during laser emission and those during non-laser emission are closely adjacent in time. The temporal distribution of ambient light noise in the two histograms is highly correlated, and the noise count values ​​corresponding to the same time bin in the two histograms are similar. During the difference operation, the first and second histograms are sorted by the same time bin number, and the count value of the first histogram is subtracted from the count value of the second histogram for each bin. The difference between each bin constitutes the third histogram. After this difference operation, the ambient light noise component that exists in both histograms is statistically canceled out, while the laser echo signal peak that exists only in the first histogram is retained. Since there are random fluctuations in the statistical process of photon counting, the difference between bins at some times may be negative. These negative value regions reflect the statistical fluctuation range of the noise, so that complete statistical information can be used for further processing during subsequent smoothing filtering.

[0034] The S4 and third histograms are obtained through the difference operation in S3. While the ambient light noise component has been largely canceled out statistically, noise fluctuations caused by statistical variations remain in the third histogram due to the random fluctuations in photon counting. These fluctuations manifest as small ups and downs in the count values ​​at each time bin. Directly searching for peaks in the third histogram could interfere with peak discrimination due to these residual fluctuations. Therefore, a smoothing filter is applied to the third histogram. A filter window slides across the histogram bin by bin, weighting the count values ​​within the window to smooth out random fluctuations and make the signal peaks more clearly defined. The fourth histogram obtained after smoothing filtering shows an improved intensity ratio of the signal peak to the residual noise peak, providing a more favorable data foundation for the subsequent super-resolution fitting peak finding in S5.

[0035] The essence of a moving average filter in the time domain is the convolution operation between the input signal and the filter coefficient array. The core calculation process of the weighted moving average in this invention is as follows: the histogram count sequence within the sliding window is multiplied element-wise with the preset filter coefficient sequence at corresponding positions, and the sum is then divided by the sum of all filter coefficients to obtain the filtered count value at that position. The formula for calculating the weighted moving average is: in, This represents the count value of bin at time n after filtering. It is the sum of all weighted coefficients of the filter. These are the weighting coefficients for the corresponding positions within the sliding window. This represents the original photon count value at time bin i before filtering.

[0036] In this invention, the coefficients of the moving average filter can be adjusted according to the actual characteristics of the laser pulse waveform used, and are not limited to a single value. As a preferred embodiment of this invention, a 16th-order moving average filter can be used, with its coefficients preset to [2,3,4,5,7,8,10,11,12,13,14,15,14,13,10,5]. First, the above weight vector is normalized so that its sum is 1, resulting in normalized weight coefficients to simplify subsequent calculations. The coefficient sequence is multiplied by the corresponding histogram count sequence within the sliding window, summed, and then divided by the total coefficient sum of 146 (i.e., the cumulative sum of the above 16 coefficients, corresponding to M in the formula), to obtain the initial count value of the corresponding time bin after filtering.

[0037] The filter coefficients described above comprise 16 orders, and their values ​​initially increase and then decrease, with larger values ​​in the middle and smaller values ​​at the ends. This coefficient pattern matches the waveform characteristics of the laser pulse. The pulse shape typically has a rising edge, a peak value, and a falling edge. The signal energy is mainly concentrated in the central region of the pulse and gradually attenuates towards the sides. When using this coefficient for moving average filtering, bins located in the center of the filter window receive higher weights, while bins located at the edges receive lower weights, consistent with the energy distribution characteristics of the laser pulse. When the filter window slides through the region where the real signal peak is located, the signal peak is weighted and enhanced; when the filter window slides through the pure noise region, the noise in each bin is weighted and averaged, tending to be smoothed. This filter can enhance the intensity of the real signal peak while smoothing noise, increasing the intensity ratio of the signal peak to the residual noise peak, making subsequent peak finding more stable and reliable.

[0038] For each time bin in the third histogram, using that bin as the center of the filtering window, the count value of each bin within the window is taken. The count value of each bin is then multiplied one by one by the corresponding coefficient in the filter coefficient array. All products are summed, and the result is divided by the sum of all coefficients in the coefficient array. The quotient is the count value of that time bin after filtering. Since moving average filtering (MA) introduces a fixed phase shift, causing the peak position of the filtered signal to shift backward relative to the original peak position, phase correction is needed for the filtered histogram. The count values ​​of all time bins are shifted forward by a phase shift corresponding to the filter order to ensure that the peak position of the filtered signal is consistent with the actual laser echo time. For example, when using the 16th-order moving average filter, a phase shift of 7.5 time bins occurs. In this case, the entire filtered histogram needs to be shifted forward by 7.5 time bins. The filtering window slides sequentially along the time bins, performing the above operation bin by bin, until all time bins of the third histogram have been traversed. The filtered count values ​​of each bin together constitute the fourth histogram. Dividing by the normalization of the sum of coefficients ensures that the total energy of the histogram before and after filtering remains constant, and the overall amplitude is not scaled due to the filtering operation. This facilitates direct comparison of peak intensities and threshold determination in subsequent S5.

[0039] S5. The fourth histogram, after smoothing filtering in S4, shows that noise fluctuations have been effectively suppressed and the signal peak outlines are clear. To obtain a distance resolution higher than the original bin width of the histogram, super-resolution fitting is required to find the peaks.

[0040] In the fourth histogram, all time bins are traversed to find the bin with the largest count value, which is then taken as the target peak bin. This target peak bin corresponds to the position where the laser echo signal peak energy is most concentrated. A local window of a preset width is taken with this target peak bin as the center. The preset width of the local window is N time bins to the left and right of the center bin, where N is a positive integer greater than or equal to 2, and the window range is [target peak bin - N, target peak bin + N]. In a preferred embodiment of the present invention, N is 4, corresponding to the 9-point centroid method, and the window covers 9 time bins. Within this local window, the count values ​​of each bin are calculated using a centroid method weighted average to obtain the precise position of the target peak.

[0041] The formula for the weighted average is: , in For the calculated peak position, The time bin number within the local window. The photon count value corresponding to the time bin is used to calculate the energy centroid position of the signal peak within the local window, which can smooth single-bin noise. The weighted average with the count value can accurately reflect the energy centroid position of the signal peak. It still has high positioning stability under low signal-to-noise ratio conditions and can obtain complete peak parameter information, providing sufficient data support for subsequent multi-target discrimination, false peak elimination, and comprehensive time-of-flight determination in complex scenarios.

[0042] The 9-point centroid method sets the time coordinate of the target peak bin to T0, i.e., the reference time Ti = T0. Centered on T0, four bins are taken forward and four bins backward, forming a local window of nine consecutive time bins. The time offset weighting coefficients for each bin within the window are set sequentially as follows: B0=0, B1=1, B2=2, B3=3, B4=4, B5=5, B6=6, B7=7, B8=8. The photon count values ​​corresponding to each bin are C0, C1, C2, C3, C4, C5, C6, C7, C8, where C4 is the photon count value of the target peak bin.

[0043] Substituting the above parameters into the general centroid method formula, we obtain the simplified calculation formula for the 9-point centroid method: in, For the final precise flight time, Let i be the photon count value within the i-th bin. This represents the time coordinate corresponding to this bin. Expanded, it can be represented as: in, For the final calculated laser flight time, C0~C8 are the photon count values ​​corresponding to each time bin within the window, A0~A8 are the amplitude weighting coefficients of each bin, and B0~B8 are the time offset weighting coefficients of each bin relative to the reference time coordinate.

[0044] The SPAD receiver sensor performs multiple statistical analyses on the echo signal to obtain the photon count within each time bin, forming a fourth histogram after differential denoising and smoothing filtering. The fourth histogram iterates through all time bins to find the bin with the highest photon count as the target peak bin, and its time coordinate T0 is determined. Using the target peak bin as the center, nine data points are extracted, four bins before and four bins after it, and substituted into the aforementioned nine-point centroid method formula to calculate the energy centroid position of the signal peak, i.e., the precise laser flight time.

[0045] Table 1: Peak Analysis Results After Filtering Table 2: Peak Analysis Results Before Filtering Comparing the two tables, we can see that: in terms of peak finding accuracy, the peak position of the target signal peak was corrected from 52.337 bins to 52.666 bins, improving the positioning accuracy by 0.329 bins. Regarding signal-to-noise ratio, in Table 2, the intensity ratio of the strongest peak (peak 1, intensity 24.054) to the second strongest peak (peak 14, intensity 13.312) is 1.8, while in Table 1, the intensity ratio of the strongest peak (peak 1, intensity 16.725) to the second strongest peak (peak 8, intensity 6.6477) increases to 2.5, an increase of approximately 39%. This improvement in intensity ratio makes the strongest peak more prominent among all peak values.

[0046] S6. After the super-resolution fitting peak finding in S5, the signal peaks in the fourth histogram are identified one by one. Each identified peak is recorded with peak intensity, peak position, full width at half maximum (FWHM), and baseline information. The peak intensity reflects the cumulative intensity of photon counts at the corresponding time bin; the peak position is the precise coordinate of the peak on the time bin axis; the FWHM reflects the broadening degree of the peak; and the baseline is the noise floor level of the local region where the peak is located. Among all the peaks identified in S5, the peak of the true laser echo signal, due to its concentrated energy and large cumulative photon count, usually has a higher peak intensity than other peaks caused by residual ambient light noise or circuit dark counting. Therefore, the peak with the highest peak intensity is selected from all peaks, and its corresponding peak position is taken as the laser's time of flight. Combining the data shown in Table 1 above, the peak intensity of peak 1 is 16.725, the highest among all 10 peaks, and its peak position is 52.666. Therefore, 52.666 is selected as the laser time of flight for this ranging measurement, and the distance to the target object can be calculated based on the time of flight.

[0047] A device for noise suppression and peak finding in DTOF histogram ranging includes a laser emitter, a SPAD receiver sensor, and a processor. Within a preset measurement period when the laser emitter emits laser light, the processor counts the photons in each time bin based on the trigger times recorded by the SPAD receiver sensor, forming a first histogram containing ambient light noise. During a time interval of the same length as the measurement period where the laser emitter does not emit laser light, the processor counts the photons in each time bin based on the trigger times recorded by the SPAD receiver sensor, forming a second histogram characterizing pure ambient light noise.

[0048] After generating the first and second histograms, the processor performs differential operations on the first and second histograms bin by bin according to the corresponding time bins to obtain a third histogram that initially suppresses ambient light noise. Then, the processor performs smoothing filtering on the third histogram to obtain a fourth histogram with enhanced signal-to-noise ratio. Finally, the processor performs super-resolution fitting on the fourth histogram to find peaks, selecting the position corresponding to the peak with the highest peak intensity from the identified peaks as the laser's time of flight, and then calculates the distance to the target object based on the time of flight.

[0049] The laser emitter and SPAD receiver sensor constitute the optical front end, while the processor performs histogram statistics and signal processing. All processing steps can be implemented on existing DTOF lidar hardware platforms without the need for additional optical filters or increased laser emission power. The processor can be a microcontroller, digital signal processor, or field-programmable gate array (FPGA) or other device with computing capabilities. Through the coordinated operation of these modules, this device can achieve stable and reliable distance measurement under strong ambient light conditions.

[0050] All other parts of this invention not described herein are the same as existing technologies, or are known technologies, or can be implemented using existing technologies, and will not be described in detail here.

[0051] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method of DTOF histogram ranging to suppress noise and peak finding, characterized in that, Includes the following steps: S1. Control the lidar to perform multiple laser emission and receive echo signals within a preset measurement period, count the photon count in each time bin, and form a first histogram containing ambient light noise; S2. Within the same time interval as the measurement cycle in S1, control the lidar to not emit laser light and only receive signals, count the photon counts in each time bin, and form a second histogram characterizing pure ambient light noise. S3. Perform a difference operation on the first histogram in S1 and the second histogram in S2 according to the corresponding time bins to obtain a third histogram that initially suppresses ambient light noise. S4. Smooth the third histogram in S3 to obtain a fourth histogram that enhances the signal-to-noise ratio of the signal peak to the noise peak. S5. Perform super-resolution fitting on the fourth histogram in S4 to find the peaks and calculate the precise position of each peak. S6. Among all the peaks obtained by fitting S5, select the position corresponding to the peak with the largest peak intensity as the flight time of the laser.

2. The DTOF histogram ranging method of claim 1, wherein, The acquisition method of steps S1 and S2 is to perform them alternately: after each measurement cycle described in step S1, a time interval described in step S2 is immediately executed, and this process is repeated multiple times to generate the first histogram and the second histogram.

3. The DTOF histogram ranging method of claim 1, wherein, The difference operation in step S3 is to subtract the count value of the second histogram from the count value of the first histogram for each bin, according to the same time bin number, and the difference between each bin constitutes the third histogram.

4. The DTOF histogram ranging method of claim 1, wherein, The smoothing filter described in step S4 uses a moving average filter.

5. The DTOF histogram ranging method of claim 4, wherein, Step S4 involves multiplying the count values ​​of each bin within the filtering window by their corresponding coefficients, summing the results, and then dividing by the sum of the coefficients to obtain the initial count values ​​of the bin at that time after filtering. Phase correction is performed on the count values ​​of all time bins after filtering, and the entire count value is shifted forward by a phase shift amount corresponding to the filter order to obtain the final count value of that time bin.

6. The method of DTOF histogram ranging to suppress noise and peak finding of claim 1, wherein, The super-resolution fitting peak finding in step S5 includes: finding the bin with the largest count value in the fourth histogram as the target peak bin; taking the target peak bin as the center, taking N time bins on the left and right to form a local window of a preset width, and performing a weighted average of the count values ​​of each bin within the window to calculate the precise position of the target peak; where N is a positive integer greater than or equal to 1.

7. The method of DTOF histogram ranging to suppress noise and peak finding of claim 6, wherein, The peak position is calculated by the weighted average of the count values of each bin in the window in step S5. The formula of the weighted average is: wherein is the calculated peak position, is the time bin number in the local window, is the photon count value of the corresponding time bin, which is suitable for the weighted average calculation of any N value.

8. The method of DTOF histogram ranging to suppress noise and peak finding of claim 1, wherein, In step S6, multiple peaks are obtained through fitting, and each peak is recorded with at least the peak intensity, peak position, full width at half maximum (FWHM), and baseline information.

9. A DTOF histogram ranging device to suppress noise and find peaks as claimed in any one of claims 1-8, wherein, include: A laser emitter is used to emit laser pulses toward a target object. SPAD receiver sensor, SPAD receiver sensor is used to receive echo photons reflected by the target object and record the trigger time; A processor, connected to the laser emitter and the SPAD receiving sensor, is configured to: within a preset measurement period during which the laser emitter emits laser light, count photons in each time bin based on the trigger time recorded by the SPAD receiving sensor, forming a first histogram; within a time interval of the same duration as the measurement period during which the laser emitter does not emit laser light, count photons in each time bin based on the trigger time recorded by the SPAD receiving sensor, forming a second histogram; perform a difference operation between the first histogram and the second histogram according to the corresponding time bins to obtain a third histogram; perform a smoothing filter on the third histogram to obtain a fourth histogram; perform super-resolution fitting on the fourth histogram to find peaks, and select the position corresponding to the peak with the largest peak intensity from the obtained peaks as the time of flight of the laser light.

Images