A sensor image processing method and processing device based on multi-peak detection

Abstract

The embodiment of the present application discloses a sensor image processing method and data processing device based on multi-peak detection, which are used to remove invalid peaks. The method of the embodiment of the present application includes: finding N1 peaks according to histogram data; screening out N2 peaks to determine as candidate peaks, and calculating the matching value of the candidate peak based on the characteristic vector of the candidate peak and a preset matching function, wherein the characteristic vector includes at least a time box corresponding to the peak and a photon count value corresponding to the peak; screening according to the matching value of the candidate peak to obtain the target peak. The method provided in the present application can effectively screen out signal peaks.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of signal processing, and in particular to a sensor image processing method and a processing device based on multi-peak detection. Background Art

[0002] In the process of array ranging and imaging, it is necessary to perform signal processing on the image data, and then obtain the distance information to be measured based on the processing results. In the image data, there are often interference signals caused by glare, dust, module and cover reflection, which will cause some pixels to introduce additional small peaks other than the target object into the histogram during the ranging process, resulting in distance detection errors. When measuring distance, if there is only one target object, there should be only one peak in the histogram, and the time bin corresponding to the maximum value of the peak corresponds to the distance from the target object to the detector. However, if there are multiple peaks in the histogram, it may be misread and the small peak formed by interference may be taken as the target peak, resulting in distance detection errors.

[0003] The first existing technology is to remove noise in the histogram only through median filtering, Gaussian filtering, etc., but the glare removal effect is not good for large films, and the effective depth information of high frequencies will be lost;

[0004] The second prior art is to select a closer or farther peak according to the farthest and closest distances that can be detected in advance, but this method requires the approximate distance of the target object to be known in advance, which increases the limitation of the measurement;

[0005] The third prior art is to calculate the signal-to-noise ratio of each peak, remove the peaks with low signal-to-noise ratio, and leave the peaks with high signal-to-noise ratio. This method is not effective in eliminating dust at close range and glare caused by highly reflective objects, and cannot optimize the removal of small peaks.

[0006] The fourth prior art is to select the farther peak output according to the peak position, but when there are multiple target objects, the peaks near and far are all valid peaks, so this method cannot be simply used for screening.

[0007] Therefore, there is no method in the prior art that can effectively remove the interference peaks and avoid the defects caused by the above four methods. Summary of the invention

[0008] The technical problem to be solved by this application is: how to effectively remove interference peaks.

[0009] The present application provides a sensor image processing method based on multi-peak detection, comprising:

[0010] Find N1 peaks based on the histogram data;

[0011] N2 peaks were screened out as candidate peaks, where N1>N2;

[0012] Calculating a matching value of the candidate peak based on a characteristic vector of the candidate peak and a preset matching function, wherein the characteristic vector includes at least a time bin corresponding to the peak and a photon count value corresponding to the peak, the preset matching function is a function related to the characteristic vector, and the matching value of the candidate peak reflects the reflectivity of an actual object causing the candidate peak;

[0013] Screening is performed according to the matching values ​​of the candidate peaks to obtain the target peak.

[0014] Preferably, the screening out N2 peaks to determine as candidate peaks specifically includes:

[0015] Setting a peak threshold and peak characteristic conditions, wherein the peak characteristic conditions include peak signal-to-noise ratio, peak width, and peak distribution;

[0016] The N1 peaks are screened, and the N2 peaks having peak values ​​greater than a threshold value and satisfying peak characteristic conditions are determined as the candidate peaks.

[0017] Preferably, the characteristic vector also includes: pulse width, rising edge time, and falling edge time.

[0018] Preferably, when the characteristic vector is: the time bin corresponding to the peak value, and the photon count value corresponding to the peak value, the matching function is:

[0019] M i =k×Ci×(Ti-n) 2 ; wherein the M i is the matching value of the i-th peak among the candidate peaks (i=1, 2, 3...); the C i is the photon count value corresponding to the peak value of the ith peak in the candidate peak; i is the time bin corresponding to the peak value of the ith peak in the candidate peak; k and n are correction parameters, and c is the speed of light.

[0020] Preferably, when the characteristic vector is: the time bin corresponding to the peak, the photon count value corresponding to the peak, the pulse width, the rising edge time, and the falling edge time, the matching function is:

[0021]

[0022] Where K(τi) is a parameter related to the pulse width.

[0023] Preferably, 1) if Ci×(Ti-n) 2 Greater than the first threshold, k(τi)=k1*τk / τi; where τk is a value related to the light waveform;

[0024] 2) If Ci×(Ti-n)2 When it is greater than the second threshold and less than the first threshold, then k(τi)=k2;

[0025] 3) If Ci×(Ti-n) 2 is less than the second threshold, then k(τi)=k3 / exp(abs(τi-τk) / τk); wherein k2 and k3 are coefficients.

[0026] Preferably, the screening according to the matching values ​​of the candidate peaks to obtain the target peak comprises: sorting the calculated matching values; and screening the target peak from the candidate peaks according to the sorting result.

[0027] Preferably, the screening according to the matching values ​​of the candidate peaks to obtain the target peak includes: judging whether the calculated matching value is greater than a preset threshold; if so, determining the peak whose matching value is greater than the preset threshold as the target peak.

[0028] In addition, the present application also provides a data processing device, including:

[0029] Laser, used for emitting laser light;

[0030] A splitter, used for splitting the emitted laser beam and emitting it;

[0031] A transmitting lens is used to focus the light beam transmitted from the separator onto the target object;

[0032] A receiving lens, used for receiving the light beam reflected by the target object;

[0033] A SPAD sensor for sensing the light beam;

[0034] A time-to-digital conversion module, used for converting the detected optical analog signal into a digital signal;

[0035] A DSP module, used for generating a histogram according to the digital signal;

[0036] The control module is used to control the laser and process the histogram; and is also used to execute a sensor image processing method based on multi-peak detection.

[0037] Preferably, the control module is further used to sort the calculated matching values; and select the target peak from the candidate peaks according to the sorting result.

[0038] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0039] For a histogram with multiple peaks, the peak feature vectors are used to construct a matching function, and the multiple peaks are screened according to the matching values.

[0040] If the peak is composed of glare, the matching value of the glare is usually lower than that of a normal object. Through screening, the invalid peak composed of the glare can be quickly removed; compared with the prior art one, the present application does not perform filtering in the frequency domain, so the effective depth information of the high frequency will not be lost, and large areas of glare can also be removed well.

[0041] If the peak is made of glass, the matching value of the glass varies greatly. Combined with other factors, such as the presence of image points, it can be quickly identified. Compared with the third prior art, the present application can better identify the glass using the signal-to-noise ratio method.

[0042] If the peak is composed of dust, the dust has a large matching value and a large difference in matching with surrounding objects. Combined with the uncertainty of its position between multiple frames, the dust can be identified by comparing the peak positions under multiple frames, thereby eliminating the invalid peak. Compared with the prior art three, the present application can better identify dust by using the signal-to-noise ratio method.

[0043] If you want to eliminate the noise contained in the peak, you can also use the matching value to do so.

[0044] Therefore, by calculating the matching values ​​of multiple peaks and combining the methods of the prior art, such as calculating the signal-to-noise ratio, etc., the defects brought by the four methods used in the prior art can be avoided. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a schematic diagram of distance measurement in an embodiment of the present application;

[0046] Figure 2 This is a flow chart of the image processing method according to an embodiment of the present application;

[0047] Figure 3 A schematic diagram of a histogram in the process of processing changes in an embodiment of the present application;

[0048] Figure 4 This is a schematic diagram of a histogram before processing in an embodiment of the present application;

[0049] Figure 5 The figure is a schematic diagram of the internal structure of the ranging module according to an embodiment of the present application. DETAILED DESCRIPTION

[0050] The embodiment of the present application provides an image processing method for effectively removing invalid peaks among all peaks.

[0051] In experimental research, people often need to obtain the distance information of the target object. Figure 1 In the process of array ranging and imaging, the image data acquired by the sensor 101 is subjected to signal processing to obtain the distance information of the target object 102 to meet the needs of research and application.

[0052] The principles of constructing the matching function and using the feature vector are as follows:

[0053] Laser receiving power: P rx =P tx *(σ / (α*β*R 2 ))*η reflct *(Arec / (pi*R 2 ))*η atm 2 *η Rx formula(*)

[0054] Among them, P rx： Received power; P tX : Transmit power;

[0055] σ is the target light receiving area, R is the distance between the target object and the SPAD sensor, α*β is the FOI emission angle,

[0056] α*β*R 2 is the light receiving area corresponding to the ROI solid angle;

[0057] η reflct is the reflectivity of the target object, Arec is the photon collection efficiency (usually the lens area)

[0058] η atm Atmospheric propagation efficiency, η Rx is the photoelectric conversion efficiency of the SPAD sensor

[0059] It can be seen that, removing η reflct In addition to R, other parameters that affect the return light energy are generally related to the device module, that is, the parameters of the same module are basically the same, so the received power P rx Mainly related to η reflct Related to R.

[0060] Observation formula (*), P rx ∝η reflct / R 2 ~η reflct / R 4 ,

[0061] When the target light receiving area σ is larger than the light receiving area corresponding to the ROI solid angle (α*β*R 2 ), we can assume that Prx∝η reflct / R 2

[0062] The peak number of photons in the histogram, peak_count, can be simply expressed by the relation:

[0063] peak_count∝Prx∝η reflct / R2

[0064] The reflectivity η of the target object can be solved by peak_count and R(Ti*c / 2) reflct , and then get the scale factor k*η reflct , where k is constant in the same measurement, so the reflectivity η reflct To sort.

[0065] Principle of reflectivity compensation:

[0066] Taking into account the nonlinearity of the spad sensor, when the reflectivity of the target object is high, the received pulse will become narrower, the rising edge will become smaller, and the falling edge will become wider (the jitter effect of the spad). With these three characteristics, the matching function of the waveform can be constructed to further compensate for the reflectivity scaling factor.

[0067] When the received pulse energy is particularly high (greater than the first threshold), it can be basically confirmed that it is a target object with high reflectivity, so the estimation of the reflectivity coefficient is increased;

[0068] When the received pulse energy is normal (less than the first threshold but greater than the second threshold), the estimation of the reflectivity coefficient is not compensated;

[0069] When the received pulse energy is very small (less than a second threshold), pulse width matching is used.

[0070] 1) Environmental noise elimination:

[0071] Considering the characteristics of noise: the width of the received pulse is inconsistent with the width of the transmitted reference pulse. Therefore, the more consistent the width of the received pulse is with the transmitted reference pulse, the more likely it is a valid echo. signal =1 / exp(abs(τi-τk) / τk);P signal is the power of the effective signal, which is related to the pulse width τi

[0072] 2) Glare elimination:

[0073] Since glare is caused by the reflection of light in the lens, the equivalent reflectivity of the glare point is generally between 0.01% and 1% of the reflectivity of the glare source point (related to the lens used), so the reflectivity of the glare point is much lower than that of a normal object. Therefore, the probability that the target point with low reflectivity is glare is very high (the correlation of flat scenes is 92%, the correlation of general scenes is 96%, and the correlation of special scenes is 83%). By constructing a distance normalization function to obtain the target reflectivity scale coefficient, we can obtain a target point that is less likely to be glare.

[0074] 3) Identify glass:

[0075] The reflectivity of glass varies greatly. When the multi-frame reflectivity scale coefficient is generally small and there is an image point, it can be determined that this is a glass point.

[0076] 4) Identify dust:

[0077] The reflectivity of dust is very large, but it is quite different from the reflectivity of surrounding objects. There is position uncertainty between multiple frames, which can be used to distinguish dust.

[0078] The following describes an embodiment of the image processing method of the present application in combination with the above information. Figure 2 , an embodiment of the image processing method of the embodiment of the present application includes:

[0079] 301. Obtain N1 peaks according to the histogram data;

[0080] After obtaining the histogram data, a search is then performed in the histogram data to obtain N1 peaks associated with the histogram data as a prerequisite for screening.

[0081] 302. Screen out N2 peaks as candidate peaks, where N1>N2;

[0082] The screening out of N2 peaks to determine as candidate peaks specifically includes:

[0083] Setting peak thresholds and peak characteristic conditions;

[0084] The N1 peaks are screened, and the N2 peaks having peak values ​​greater than a threshold value and satisfying peak characteristic conditions are determined as the candidate peaks.

[0085] Preferably, after screening out N2 peaks as candidate peaks, the method further includes: constructing a characteristic vector of the candidate peak, and constructing a corresponding matching function based on the characteristic vector, wherein the characteristic vector includes: a time bin corresponding to the peak, a photon count value corresponding to the peak, a pulse width, a rising edge time, and a falling edge time.

[0086] 303. Calculate the matching value of the candidate peak based on the characteristic vector of the candidate peak and a preset matching function, wherein the characteristic vector is a vector related to the shape and size of the peak;

[0087] Specifically, interference can be glare. The reflectivity of glare points is much lower than that of normal objects. Interference can also be dust. Dust particles are small, but the reflectivity is large, and the corresponding peak stability is poor. Interference can also be glass, the reflectivity of glass varies greatly, or there are other interfering factors, which are not limited here.

[0088] 1. When the characteristic vector is: the time bin corresponding to the peak value, the photon count value corresponding to the peak value, the matching function is:

[0089] M i =k×Ci×(Ti-n) 2 Formula 1

[0090] Among them, the M i is the matching value of the i-th peak among the candidate peaks (i=1, 2, 3...); the C i is the photon count value corresponding to the peak value of the ith peak in the candidate peak; i is the time bin corresponding to the peak value of the ith peak in the candidate peak; k and n are correction parameters, and c is the speed of light.

[0091] 2. When the characteristic vector is: the time bin corresponding to the peak, the photon count value corresponding to the peak, and the pulse width, the matching function is

[0092]

[0093] Wherein τi is the pulse width of the i-th peak among the alternative peaks, and τk is a value related to the light emitting waveform.

[0094] 3. When the characteristic vector is: the time bin corresponding to the peak, the photon count value corresponding to the peak, the rising edge time, and the falling edge time, the matching function is

[0095]

[0096] Among them, τi_up is the rising edge time and τi_dn is the falling edge time.

[0097] 4. When the characteristic vector is: the time bin corresponding to the peak, the photon count value corresponding to the peak, the pulse width, the rising edge time, and the falling edge time, the matching function is:

[0098]

[0099] Where k(τi) is a parameter related to the pulse width.

[0100] k(τi) can be determined as follows:

[0101] If Ci×(Ti-n) 2 When it is greater than the first threshold, k(τi)=k1*τk / τi, where τk is a value related to the transmission waveform;

[0102] If Ci×(Ti-n) 2 When it is greater than the second threshold value and less than the first threshold value, k(τi)=k2;

[0103] If Ci×(Ti-n) 2When it is less than the second threshold, k(τi)=k3 / e abs(τi-τk) / τk , where τk is a value related to the transmitted waveform;

[0104] The first threshold is much larger than the second threshold.

[0105] In summary, the construction of the matching function is related to the problem to be solved. Different matching functions can be constructed according to the required scenarios, or the same matching function can be constructed for several scenarios at the same time. In actual engineering, the parameters of the required matching function need to be calibrated according to experiments.

[0106] 304. Screening is performed according to the matching values ​​of the candidate peaks to obtain a target peak.

[0107] The screening conditions are pre-set according to the purpose, and the matching values ​​of the candidate peaks are screened. If a matching value meets the screening conditions, the peak corresponding to the matching value is determined as the target peak.

[0108] Specifically, the matching values ​​are screened and the target peak can be determined by sorting. The matching values ​​of the calculated candidate peaks are sorted, and then the peak with a higher matching value is determined as the target peak according to the sorting result. Among them, the target peak can be one or two or more, and the number of target peaks is less than the number of candidate peaks. The specific number of target peaks can be determined according to the purpose.

[0109] Specifically, the matching values ​​can be screened, and the target peak can also be determined by the preset threshold method. First, a specific threshold is preset according to the purpose, and then the calculated matching value of the candidate peak is compared with the preset threshold. If the matching value of a peak among the candidate peaks is greater than the preset threshold, the peak is determined as the target peak. Here, the target peak can be one or more, which are determined according to the purpose.

[0110] It is understandable that the screening may be by sorting, or by setting a preset threshold or other methods, which are not specifically limited here.

[0111] To facilitate the understanding of the target peak screening process, Figure 3 A schematic diagram of the process of screening out the target peak from all peaks is given. All peaks are screened for the first time, and the peak that meets the conditions is the N1 peak. Then the N1 peak is screened, that is, the second screening, to obtain the N2 peak. The N2 peak is determined as the candidate peak, and the candidate peak is screened by the matching function, photon count value and time box, that is, the third screening, to obtain the target peak.

[0112] The following example illustrates how to use the matching value to screen valid peaks.

[0113] like Figure 4As shown, there are 2 peaks in the histogram, and the eigenvectors of the 2 peaks are as follows:

[0114] The peak on the left The peak on the right Time bin corresponding to the peak value: Ti 14 18 Photon count value corresponding to the peak value: Ci 33 4 Pulse width: τi 1.1049 1.5 Rising edge time: τi_up 0.5156 0.5 Falling edge time: τi_dn 0.5893 1

[0115] The peak on the left may be caused by glare. If conventional technology is used, that is, the highest peak is selected as the effective peak of the target object, the peak caused by glare will be mistakenly used for calculating the distance of the target object.

[0116] Existing technology 1 cannot be eliminated by Gaussian filtering; Existing technology 2 can only select the small peak at the back, but it requires knowing the approximate position of the target peak in advance, which has limitations; Existing technology 3 has a larger signal-to-noise ratio of the left peak and cannot be eliminated by the signal-to-noise ratio; Existing technology 4, if there are multiple target objects, it cannot be determined that the left peak is definitely caused by glare, and it will be mistakenly screened out.

[0117] This application can be screened by calculating the matching value, as follows:

[0118]

[0119] Because the first threshold is 180 and the second threshold is 1,

[0120] When the Ci of the two peaks are both greater than the second threshold and less than the first threshold, k(τi)=k2; k(τi)=1 can be set;

[0121] The matching value of the left peak is

[0122] M1=k(τi)×(τi_dn / τi_up)×Ci×(Ti-n) 2 =1×(0.5893 / 0.5156)×33×(14-11) 2 =339

[0123] Matching value of the right peak

[0124] M2=k(τi)×(τi_dn / τi_up)×Ci×(Ti-n) 2 =1×(1 / 0.5)×4×(18-11) 2 =392

[0125] Then we compare the sizes of M1 and M2, and we know that M2>M1.

[0126] Because the matching value of the glare point is lower than that of the normal object, the left peak with small M1 is caused by glare and can be screened out. The right peak is the peak reflected by the target object and can be used to calculate the actual distance of the target object.

[0127] The construction of the matching degree function is related to the problems to be solved. Different matching degree functions can be constructed according to the required scenarios (such as glare, dust, glass, etc.), or the same matching degree function that can be applied to several scenarios at the same time can be constructed. In actual engineering, the parameters of the required function need to be calibrated according to experiments.

[0128] For example:

[0129] In the case of glare, use Formulas 1 - 4 to calculate respectively, as follows;

[0130] 1) Calculation by Formula 1: M i = k × Ci × (Ti - n) 2

[0131] M1 = 1 * 33 * (14 - 11)^2 = 297,

[0132] M2 = 1 * 4 * (18 - 11)^2 = 198

[0133] Because M1 > M2, so when calculating the matching degree values of the two peaks using Formula 1 and then comparing and sorting, it is impossible to filter out the left peak caused by glare.

[0134] 2) Calculation by Formula 2:

[0135]

[0136] M1 = 1 / e ^ (1.1049 - 1) * 33 * (14 - 11)^2 = 0.9 * 297 = 267

[0137] M2 = 1 / e ^ (1.5 - 1) * 4 * (18 - 11)^2 = 7.4 * 198 = 9.5, where τ K = 1

[0138] Because M1 > M2, so when calculating the matching degree values of the two peaks using Formula 1 and then comparing and sorting, it is impossible to filter out the left peak caused by glare.

[0139] 3) Calculation by Formula 3:

[0140]

[0141] M1 = 1 * (0.5893 / 0.5156) * (14 - 11)^2 * 33 = 1.14 * 297 = 338

[0142] M2 = 1 * (1 / 0.5) * (18 - 11)^2 * 4 = 2 * 198 = 392

[0143] Because M1 < M2, so when calculating the matching degree values of the two peaks using Formula 3 and then comparing and sorting, it is possible to filter out the left peak caused by glare.

[0144] 4) Formula 4 calculation:

[0145]

[0146] It is necessary to determine whether Ci*(Ti - n) 2 is between the first threshold and the second threshold; the specific calculation process has been described above and will not be elaborated here.

[0147] Because M1 < M2, so use Formula 4 to calculate the matching degree values of the two peaks, and then compare and sort, the left peak caused by glare can be screened out.

[0148] Therefore, the construction of the specific matching degree function is related to the problem to be solved. If it is necessary to eliminate glare, it is necessary to use the matching degree function constructed by Formula 3 or 4 to calculate the matching degree value. In actual engineering, it is mainly judged according to the measured histogram waveform. Because there are many influencing factors, it is impossible to use the same formula to calculate in different scenarios for different noise sources. Therefore, different formulas need to be used for verification, and the formula may need to be modified, such as adding feature vectors to adjust the relationship of the formula. But the core is to calculate the reflectivity of the target object, and find the noise through the difference in reflectivity between the object and the noise, so as to eliminate the noise, including environmental noise, glare, dust, glass, etc.

[0149] Technical effects of this application:

[0150] 1. Solve the problem of eliminating non-target multiple peaks and the problem of selecting small targets through peak characteristics and matching degree function;

[0151] 2. The glare area can be identified by the sensor described in this application. By subtracting the SPAD photosensitive area after elimination from the SPAD photosensitive area before elimination, the difference part is the glare area. Through this method, the result of the glare area is sent to other sensors (RGB) to enhance the glare elimination ability of other sensors.

[0152] 3. If the method of eliminating glare in this application is combined with spatial filtering, the remaining 10% of glare points can be further eliminated.

[0153] In this embodiment, all peaks can be obtained according to the histogram data, the signal peaks are obtained through one screening, and then the target peaks are obtained through screening by the matching degree function. Through the two screenings, the invalid peaks can be removed, and the removal rate of the invalid peaks is improved compared with one screening.

[0154] The data processing device of the embodiment of this application will be described in detail below. Please refer to Figure 5 , another embodiment of the data processing device 700 of the embodiment of this application, includes:

[0155] Laser 710, transmitting lens 720, receiving lens 721, separator 722, control module 730, single photon avalanche diode (SPAD) sensor 740, digital-to-analog conversion module 750 and digital signal processing (DSP) module 760;

[0156] For ease of understanding, Figure 5 Target object 770 is provided for illustration;

[0157] The control module 730 has a microprocessor that can be used for timing control of input and output and power control of the laser 710;

[0158] The laser 710 is used to emit infrared light with a wavelength of 800nm ​​to 950nm, wherein the data processing device of the embodiment of the present application has good detection performance for the peak with a wavelength of 905nm. It is understandable that in other embodiments, the laser 710 can also emit infrared light with a wavelength of 1550nm, which is not specifically limited here;

[0159] A splitter 722, used to split the emitted laser beam and emit it;

[0160] An emission lens 720, used to focus the light beam transmitted from the separator 722 onto the target object 770, wherein the emission lens 720 may include a plurality of lens elements;

[0161] The shape and size of the target object 770 will affect the laser pattern generated subsequently, wherein the medium between the target object 770 and the data processing device 700 is air and dust, which will also affect the result;

[0162] A receiving lens 721 is used to receive the light beam reflected by the target object 770, wherein the receiving efficiency is an important indicator. The receiving lens 721 may include multiple lens elements and be coated with an anti-reflection coating to solve the glare problem generated by a fully exposed or dispersed light source position;

[0163] SPAD sensor 740 for photoelectric detection;

[0164] A digital-to-analog conversion module 750 is used to convert the analog optical signal into a digital signal to facilitate subsequent processing;

[0165] DSP module 760 is used to implement the above Figure 2 The method can generate a histogram, wherein in other embodiments, the DSP module 760 can be integrated with the control module 730 and controlled by the control module 730.

[0166] The control module 730 controls the laser 710 and emits infrared light to the separator 722. The light beam processed by the separator 722 passes through the transmitting lens 720 to the target object 770, and the reflected light beam is reflected back through the receiving lens 721 to the SPAD sensor 740. After photoelectric detection, the digital-to-analog conversion module 750 converts the analog signal obtained into a digital signal, which is then transmitted to the DSP module 760 for processing.

[0167] The control module 730 is also used to execute the above-mentioned sensor image processing method based on multi-peak detection, including: finding N1 peaks according to histogram data; screening out N2 peaks to be determined as candidate peaks, where N1>N2; calculating the matching value of the candidate peak based on the characteristic vector of the candidate peak and a preset matching function, the characteristic vector at least including the time box corresponding to the peak and the photon count value corresponding to the peak, the preset matching function is a function related to the characteristic vector, and the matching value of the candidate peak reflects the reflectivity of the actual object causing the candidate peak; screening is performed according to the matching value of the candidate peak to obtain the target peak.

[0168] according to:

[0169] Laser receiving power: P rx =P tx *(σ / (α*β*R 2 ))*η reflct *(Arec / (pi*R 2 ))*η atm 2 *η Rx formula(*)

[0170] Among them, P rx： Received power; P tX : Transmit power;

[0171] σ is the target light receiving area, R is the distance between the target object and the SPAD sensor, α*β is the FOI emission angle,

[0172] α*β*R 2 is the light receiving area corresponding to the ROI solid angle;

[0173] η reflct is the reflectivity of the target object, Arec is the photon collection efficiency (usually the lens area)

[0174] η atm Atmospheric propagation efficiency, η Rx is the photoelectric conversion efficiency of the SPAD sensor

[0175] It can be seen that, removing η reflctIn addition to R, other parameters that affect the return light energy are generally related to the device module, that is, the parameters of the same module are basically the same, so the received power P rx Mainly related to η reflct Related to R.

[0176] It can be seen that the laser receiving power P rx It is related to the parameters of the above hardware modules.

[0177] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0178] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0179] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0180] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0181] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, read-only memory), random access memory (RAM, random access memory), disk or optical disk and other media that can store program code.

Claims

1. A sensor image processing method based on multi-peak detection, characterized in that: include: Find N1 peaks based on the histogram data; N2 peaks were screened out as candidate peaks, where N1>N2; Calculating a matching value of the candidate peak based on a characteristic vector of the candidate peak and a preset matching function, wherein the characteristic vector includes at least a time bin corresponding to the peak and a photon count value corresponding to the peak, the preset matching function is a function related to the characteristic vector, and the matching value of the candidate peak reflects the reflectivity of an actual object causing the candidate peak; Screening is performed according to the matching values ​​of the candidate peaks to obtain the target peak.

2. The image processing method according to claim 1, characterized in that: The screening out of N2 peaks to determine as candidate peaks specifically includes: Setting a peak threshold and peak characteristic conditions, wherein the peak characteristic conditions include peak signal-to-noise ratio, peak width, and peak distribution; The N1 peaks are screened, and the N2 peaks having peak values ​​greater than a threshold value and satisfying peak characteristic conditions are determined as the candidate peaks.

3. The image processing method according to claim 1, characterized in that: The characteristic vector also includes: pulse width, rising edge time, and falling edge time.

4. The image processing method according to claim 3, characterized in that: When the characteristic vector is: the time bin corresponding to the peak value and the photon count value corresponding to the peak value, the matching function is: M i =k×Ci×(Ti-n) 2 ; wherein the M i is the matching value of the i-th peak among the candidate peaks, i=1, 2, 3...; Ci is the photon count value corresponding to the i-th peak value among the candidate peaks; Ti is the time box corresponding to the i-th peak value among the candidate peaks; k and n are correction parameters.

5. The image processing method according to claim 4, characterized in that: When the characteristic vector is: the time bin corresponding to the peak, the photon count value corresponding to the peak, the pulse width, the rising edge time, and the falling edge time, the matching function is: Among them, k(τi) is a parameter related to the pulse width, τi_up is the rising edge time, and τi_dn is the falling edge time.

6. The image processing method according to claim 5, characterized in that: 1) If Ci×(Ti-n) 2 Greater than the first threshold, k(τi)=k1*τk / τi; where τk is a value related to the light waveform; 2) If Ci×(Ti-n) 2 When it is greater than the second threshold and less than the first threshold, then k(τi)=k2; 3) If Ci×(Ti-n) 2 is less than the second threshold, then k(τi)=k3 / exp(abs(τi-τk) / τk); wherein k2 and k3 are coefficients.

7. The image processing method according to claim 1, characterized in that: The screening according to the matching values ​​of the candidate peaks to obtain the target peaks includes: Sort the calculated matching values; The target peak is screened out from the candidate peaks according to the sorting result.

8. The image processing method according to claim 1, characterized in that: The screening according to the matching values ​​of the candidate peaks to obtain the target peaks includes: Determine whether the calculated matching value is greater than a preset threshold; If so, the peak whose matching value is greater than the preset threshold is determined as the target peak.

9. A data processing device, characterized in that: include: Laser, used for emitting laser light; A splitter, used for splitting the emitted laser beam and emitting it; A transmitting lens is used to focus the light beam transmitted from the separator onto the target object; A receiving lens, used for receiving the light beam reflected by the target object; A SPAD sensor for sensing the light beam; A time-to-digital conversion module, used for converting the detected optical analog signal into a digital signal; A DSP module, used for generating a histogram according to the digital signal; A control module, used to control the laser and process the histogram; and also used to execute the method according to any one of claims 1 to 8.

10. The processing device according to claim 9, characterized in that: The control module is further used to sort the calculated matching values; and select the target peak from the candidate peaks according to the sorting result.

Images