Laser active imaging noise processing method

Abstract

The invention relates to a laser active imaging noise processing method, and belongs to the technical field of image processing and denoising. The single-photon denoising imaging processing method provided by the invention is designed based on the imaging characteristics of the single-photon array detector, the time characteristics of the echo data and the spatial correlation, and has a relatively good inhibition effect on background noise; the accurate target scene echo distance image and intensity image can be rapidly and accurately extracted from single-photon detection original data, and the method is extremely important for active imaging application of high-sensitivity single-photon array laser imaging.

Description

Technical Field

[0001] The invention belongs to the technical field of image processing denoising, and in particular relates to a laser active imaging noise processing method. Background Art

[0002] In recent years, under the continuous demand for the detection of non-cooperative targets with weak echoes at long distances, single-photon laser imaging (also known as photon counting laser imaging) has been applied and can meet the detection needs of tens to hundreds of kilometers. Compared with traditional linear PIN and APD laser ranging radars, single-photon laser imaging mainly uses high-sensitivity single-photon detectors (usually including single-photon avalanche diodes SPAD, photomultiplier tubes PMT and superconducting nanowire single-photon detectors SNSPD) and time-correlated single-photon counting technology. It has the ability to detect single photons and can realize the detection of non-cooperative targets under long-distance weak echoes. It has broad application prospects in remote sensing detection, target recognition and other fields.

[0003] Laser imaging based on single-photon array detectors has high sensitivity, spatial resolution and ranging accuracy, and also has the characteristics of rapid imaging acquisition. Compared with scanning single-photon laser imaging, it is a more efficient detection method and is a hot topic of research at home and abroad. However, laser imaging based on single-photon array detectors usually works in gated mode and is easily interfered by strong background noise. Its imaging effect drops sharply as the noise increases. The commonly used peak method and sliding window method are difficult to effectively and quickly extract the target echo signal, and it is difficult to achieve strong noise suppression and improve the imaging quality. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] The technical problem to be solved by the present invention is: in order to solve the problem that single-photon array laser imaging is easily affected by background noise, resulting in reduced imaging quality, and some existing processing algorithms are difficult to quickly and effectively suppress its strong background noise, making it difficult to quickly and accurately extract target information, a laser active imaging noise processing method is provided.

[0006] (II) Technical solution

[0007] In order to solve the above technical problems, the present invention provides a laser active imaging noise processing method, comprising the following steps:

[0008] Obtain the noise photoelectron count rate using imported single-photon laser imaging data or known noise parameters of the detection scene;

[0009] According to the laser imaging image data collected by multiple frames, a coarse time resolution histogram of pixels and its displacement histogram are constructed;

[0010] The target scene recognition threshold distribution is obtained by combining the coarse time resolution, the number of pulse accumulations, the noise photoelectron count rate, and the required pulse accumulation false alarm probability;

[0011] Compare the statistical counts of the coarse time resolution histogram and its misaligned histogram with the target scene recognition threshold distribution; determine whether the statistical counts of the coarse time resolution histogram and its misaligned histogram meet the recognition threshold requirement;

[0012] When the recognition threshold requirement is met, the effective echo rough position of the pixel is quickly extracted, and the precise time and maximum count value of the pixel are quickly extracted by using the echo rough position and the fine time resolution histogram of the pixel through the sliding window weighting method;

[0013] When the recognition threshold requirement is not met, it is determined whether the number of spatial correlation field correlations that have been executed meets the requirement. If so, it is assumed that this pixel has no echo. At this time, the time and maximum count are set to empty or 0, and the corresponding distance value and intensity value are also empty or 0. If it is not met, the spatial neighborhood correlation characteristics are used to select the neighborhood data around the pixel for supplementation to form new echo data of the pixel, and the new echo data is processed again according to the initial steps;

[0014] Finally, all pixel times and maximum count values ​​are counted, and distance and intensity images are calculated, and median filtering is performed on the image.

[0015] Preferably, the method of obtaining the noise photoelectron count rate by using the imported single-photon laser imaging data or the known noise parameters of the detection scene includes: the imported single-photon laser imaging data is based on multi-frame laser imaging data collected by a single-photon array detector, and the single-photon array detector performs a statistical histogram on all data or a sampled portion of the data by working in a gated mode, and fits the statistical histogram using a single detection probability model in the gated mode, and the fitting formula is as follows:

[0016]

[0017] Where P SFA_gate is the false alarm probability of single pulse detection, which is the ratio of the statistical count of the nth time grid in the statistical histogram to the total pulse accumulation number. is the noise photoelectron count rate, and Δt is the time resolution of the statistical histogram. Here, it is necessary to replace the maximum value with the relevant neighborhood count to prevent the target signal echo from affecting the accurate extraction of the noise photoelectron count rate. Then, the noise photoelectron count rate is extracted using the fitting formula: or based on a known noise parameter, the background noise photon count rate and dark count rate The noise photoelectron count rate is calculated by combining the detector detection efficiency η, and the formula is as follows:

[0018]

[0019] Preferably, the method of constructing a coarse time resolution histogram of pixels and its misalignment histogram based on the laser imaging image data collected from multiple frames includes: using the multi-frame accumulated data of the pixels, combined with the set coarse time resolution (the setting of the coarse time resolution is determined according to the laser pulse width), to construct a coarse time resolution histogram of pixels and its misalignment histogram, and the construction principle is as follows: Figure 2 As shown, the coarse time resolution is 6t fwhm / 2.355·2, where t fwhm is the full width at half maximum of Gaussian laser, t fwhm / 2.355 is the standard deviation of Gaussian laser, 6t fwhm The size of / 2.355 contains 99.7% of the energy of the laser. Its purpose is to quickly extract the approximate echo position of the target scene and avoid misjudgment caused by the target echo statistics being evenly distributed in two adjacent time grids. In addition, when the noise is large, a smaller coarse time resolution size is usually set, such as 6t fwhm / 2.355, to construct the misalignment histogram. Although some signal echo counts are lost, the influence of noise can be better suppressed.

[0020] Preferably, the target scene recognition threshold distribution is calculated and obtained by combining the coarse time resolution, the number of pulse accumulation, the noise photoelectron count rate and the pulse accumulation false alarm probability that meets the requirements, including: when the coarse time resolution and the noise photoelectron count rate are known, the single pulse detection false alarm probability distributed over the time grid is obtained by formula (1), and then the false alarm probability after multiple pulse accumulation can be obtained by combining the number of multiple pulse accumulations and the recognition threshold, and the relationship is as follows:

[0021]

[0022] In the formula, k th is the threshold for identifying the target within the time gate, 1≤k th ≤N, the index of k′ ranges from 0 to k th -1, N is the pulse accumulation number, It represents the number of combinations of k′ different elements taken out from N different elements each time. Therefore, by setting the false alarm probability after multiple pulses and the false alarm probability of single pulse detection distributed over the time grid, the target scene recognition threshold distribution can be obtained.

[0023] Preferably, the comparing of the statistical counts of the coarse time resolution histogram and its staggered histogram with the target scene recognition threshold distribution includes: comparing the generated time-distributed coarse time resolution histogram and its staggered histogram statistical counts with the acquired time-distributed recognition threshold distribution, where both constructed histogram statistical counts are required to be compared with them to prevent the situation where the target scene echo statistical counts are evenly distributed in two adjacent time grids and cause misjudgment.

[0024] Preferably, the determination of whether the statistical counts of the coarse time resolution histogram and its misaligned histogram meet the recognition threshold requirement includes: after comparing the statistical counts of the coarse time resolution histogram and its misaligned histogram with the recognition threshold distribution, if any one of the two constructed histogram statistical counts meets the recognition threshold requirement, it can be determined that the echo of the pixel is a valid echo, otherwise it is determined whether the number of spatial-related field-related data reorganizations that have been executed meets the requirement.

[0025] Preferably, when the recognition threshold requirement is met, the effective echo rough position of the pixel is quickly extracted, and the precise time and maximum count of the pixel are quickly extracted by using the echo rough position and the fine time resolution histogram of the pixel through a sliding window weighting method.

[0026] Preferably, when the recognition threshold requirement is not met, the number of spatial correlation field data reorganizations is performed to see if the requirement is met. If so, it is assumed that this pixel has no echo. At this time, the time and the maximum count are set to empty or 0, and the corresponding distance value and intensity value are also empty or 0. If it is not met, the spatial neighborhood correlation characteristics are used to select the neighborhood data around the pixel for supplementation to form new echo data for the pixel, and the new echo data is processed again according to the initial steps (return to the first step, that is, the step of obtaining the noise photoelectron count rate using the imported single-photon laser imaging data or the known noise parameters of the detection scene).

[0027] Preferably, the counting of all pixel times and maximum count values, and calculation to form a distance image and an intensity image, includes: using the above-mentioned processing method to obtain the effective time and maximum count value of the pixel, counting all pixels and calculating to form a distance image and an intensity image, performing median filtering on the distance image and the intensity image, and further completing image noise suppression.

[0028] The present invention also provides a system for implementing the method.

[0029] (III) Beneficial effects

[0030] The advantages of the present invention are:

[0031] 1) All or part of the multi-frame single-photon laser imaging data is used to perform statistical histograms, and the statistical histograms are fitted with the single detection probability model in the gated mode. The noise photoelectron count rate is extracted using the fitting formula, which provides important parameter support for obtaining the target scene recognition threshold distribution.

[0032] 2) Using the multi-frame accumulated data of pixels, combined with the set coarse time resolution, the coarse time resolution histogram of pixels and its dislocation histogram are constructed. The purpose is to quickly extract the approximate echo position of the target scene and avoid the situation where the target echo statistical counts are evenly distributed in two adjacent time grids and cause misjudgment. In addition, when the noise is greater, a smaller coarse time resolution size is usually set to construct the dislocation histogram. Although some signal echo counts are lost, the influence of noise can be better suppressed.

[0033] 3) In combination with the coarse time resolution, the number of pulse accumulation, the noise photoelectron counting rate and the pulse accumulation false alarm probability that meets the requirements, the target scene recognition threshold distribution is calculated and obtained, wherein the pulse accumulation false alarm probability can be set according to the actual situation. If the noise filtering is more strict, it can be set lower, such as 0.1%. If the requirements are not strict, it can be set to a relatively high point, such as 5%. When the set parameters are different, the target scene recognition threshold distribution will be different.

[0034] 4) When the recognition threshold requirement is met, the effective echo coarse position of the pixel can be quickly extracted. The coarse echo position and the fine time resolution histogram of the pixel are used to quickly extract the precise time and maximum count of the pixel through a sliding window weighted method. This process is based on constructing a coarse time resolution histogram of the pixel and its misalignment histogram. Combined with the fine time resolution histogram and the sliding window weighted method, the time and maximum count of the target scene can be extracted faster and more accurately.

[0035] 5) When the recognition threshold requirement is not met, the spatial neighborhood correlation characteristics are utilized to select the neighborhood data around the pixel for supplementation to form new echo data of the pixel, and the designed method is executed again for processing to further accurately improve the time corresponding to the pixel and the maximum count.

[0036] 6) A laser active imaging noise processing method is proposed based on the imaging characteristics of single-photon array detectors, the temporal characteristics of echo data and the spatial correlation. By flexibly setting the parameters, it can have a good suppression effect on background noise, and can quickly and accurately extract more accurate distance images and intensity images from the single-photon detection raw data, which is crucial for the application of high-sensitivity single-photon laser imaging active imaging. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1: A schematic diagram of a method for constructing a single-photon laser imaging echo model;

[0038] Figure 2 :Schematic diagram of the construction of coarse time resolution histogram and its offset histogram;

[0039] Figure 3 : The true value distance map and intensity map corresponding to the generated single photon array laser imaging data; (a) is the true value distance map, (b) is the true value intensity map;

[0040] Figure 4 : The constructed coarse time resolution statistical histogram and the fitted green curve;

[0041] Figure 5 : Construct the coarse time resolution histogram and its misalignment histogram of pixel (1,1); (a) is the coarse time resolution histogram, (b) is the misalignment histogram of the coarse time resolution;

[0042] Figure 6 : The coarse time resolution histogram constructed by pixel (1,1), the fitted green curve and the recognition threshold distribution under 500 pulse accumulation and the false alarm probability not exceeding 5%;

[0043] Figure 7 : The relationship between the coarse time resolution histogram constructed by pixel (1,1) and its misalignment histogram statistical counts and the recognition threshold distribution; (a) is the coarse time resolution histogram and the recognition threshold distribution, (b) is the coarse time resolution misalignment histogram and the recognition threshold distribution;

[0044] Figure 8 : The range image and intensity image processed by the proposed processing method; (a) is the range image processed by the processing method of the present invention, and (b) is the intensity image processed by the processing method of the present invention;

[0045] Fig. 9 : The distance image and intensity image processed based on the peak method, (a) is the distance image processed based on the peak method, (b) is the intensity image processed based on the peak method. DETAILED DESCRIPTION

[0046] In order to make the purpose, content and advantages of the present invention more clear, the specific implementation methods of the present invention are further described in detail below in conjunction with the drawings and examples.

[0047] Aiming at the needs of noise suppression and rapid processing and reconstruction of detection data, the present invention proposes a laser active imaging noise processing method, which specifically includes: obtaining the noise photoelectron count rate by using imported single-photon laser imaging data or known noise parameters of the detection scene; constructing a coarse time resolution histogram and its misalignment histogram of pixels according to laser imaging image data collected by multiple frames; combining the coarse time resolution, the number of pulse accumulations, the noise photoelectron count rate and the pulse accumulation false alarm probability that meets the requirements, and then obtaining the target scene recognition threshold distribution; comparing the coarse time resolution histogram and its misalignment histogram statistical counts with the recognition threshold distribution; judging whether the histogram statistical counts meet the recognition threshold requirements; and when the recognition threshold requirements are met, , quickly extract the effective echo rough position of the pixel, use the echo rough position and the fine time resolution histogram of the pixel, and use the sliding window weighting method to quickly extract the precise time and maximum count of the pixel; when the requirements are not met, determine whether the number of spatial correlation field correlations that have been executed meets the requirements. If so, it is assumed that this pixel has no echo. At this time, the time and maximum count are set to empty or 0, and the corresponding distance value and intensity value are also empty or 0. If it is not satisfied, the spatial neighborhood correlation characteristics are used to select the neighborhood data around the pixel for supplementation to form new echo data for the pixel, and the new echo data is processed again according to the initial steps; finally, all pixel times and maximum count values ​​are counted and calculated to form distance images and intensity images, and median filtering is performed on the image.

[0048] The single-photon denoising imaging processing method proposed in the present invention is designed based on the imaging characteristics of single-photon array detectors, the temporal characteristics of echo data and the spatial correlation. It can have a good suppression effect on background noise, and can quickly and accurately extract more accurate target scene echo distance images and intensity images from single-photon detection raw data, which is extremely important for high-sensitivity single-photon array laser imaging active imaging applications.

[0049] Figure 1 This is a schematic diagram of a laser active imaging noise processing method of the present invention. Please refer to Figure 1 , the method specifically includes the following steps:

[0050] Step 1, start importing the collected single-photon array laser imaging data. Here, the multi-frame single-photon array laser imaging data is simulated and generated by the laser imaging true value intensity map and distance map, noise photoelectron count rate, echo photon rate equation, Poisson distribution and Monte Carlo method. The true value distance map and intensity map are as follows: Figure 3 shown.

[0051] Step 2, use the imported single-photon laser imaging data or the known noise parameters of the detection scene to obtain the noise photoelectron count rate. The imported single-photon laser imaging data is based on the multi-frame laser imaging data collected by the single-photon array detector. Here, a statistical histogram is performed on the sampled data, and the statistical histogram is fitted using the single detection probability model in the gated mode. The noise photoelectron count rate can be extracted using the fitting formula, where the coarse time resolution histogram constructed by the first pixel (1,1) extracted and the fitted green curve are shown in Figure 2. Figure 4 shown.

[0052] Step 3, constructing a coarse time resolution histogram of pixels and its misalignment histogram according to the laser imaging image data collected by multiple frames, including: using the multi-frame accumulated data of the pixels, combined with the set coarse time resolution, the setting of the coarse time resolution is determined according to the laser pulse width, and then constructing a coarse time resolution histogram of pixels and its misalignment histogram, wherein the coarse time resolution histogram of pixels (1,1) and its misalignment histogram are as follows: Figure 5 As shown in the figure, in order to include the echo counts corresponding to all the energy contained in the laser pulse width and avoid the situation where the target echo statistical counts are evenly distributed in two adjacent time grids and cause misjudgment, the coarse time resolution is set to 6t fwhm / 2.355·2, where t fwhm / 2.355 is the standard deviation of the laser pulse.

[0053] Step 4, combining the coarse time resolution, the number of pulse accumulation, the noise photoelectron count rate and the pulse accumulation false alarm probability that meets the requirements, calculate and obtain the target scene recognition threshold distribution, including: when the coarse time resolution and the noise photoelectron count rate are known, the single pulse detection false alarm probability distributed over the time grid can be obtained, and then the false alarm probability after multiple pulse accumulation can be obtained by combining the number of multiple pulse accumulations and the recognition threshold. The coarse time resolution histogram constructed by pixel (1,1), the fitted green curve and the recognition threshold distribution (represented by the black curve) under 500 pulse accumulations and the false alarm probability not exceeding 5% are shown as follows Figure 6 shown.

[0054] Step 5, comparing the coarse time resolution histogram and its misalignment histogram statistics with the recognition threshold distribution, including: comparing the generated coarse time resolution histogram and its misalignment histogram statistics with the time distribution with the already acquired recognition threshold distribution, such as Figure 7 As shown, the coarse time resolution histogram constructed by pixel (1,1) and its misalignment histogram statistical counts are respectively related to the recognition threshold distribution.

[0055] Step 6, judging whether the histogram statistical counts meet the recognition threshold requirements, including: comparing the coarse time resolution histogram and its offset histogram statistical counts with the recognition threshold distribution. If any one of the two constructed histogram statistical counts meets the recognition threshold requirements, the pixel echo can be judged as a valid echo. Otherwise, it is judged whether the number of spatial-related field-related data reorganizations that have been executed meets the requirements.

[0056] Steps 7 and 8, when the recognition threshold requirement is met, the effective echo rough position of the pixel is quickly extracted, and the precise time and maximum count of the pixel are quickly extracted by using the echo rough position and the fine time resolution histogram of the pixel through the sliding window weighted method.

[0057] Steps 11 to 13, when the recognition threshold requirement is not met, the number of spatial correlation field data reorganizations is performed to see if the requirement is met. If so, it is assumed that this pixel has no echo, i.e., step 13. At this time, the time and maximum count are set to empty or 0, and the corresponding distance value and intensity value are also empty or 0. If it is not met, the spatial neighborhood correlation characteristics are used to select the neighborhood data around the pixel for supplementation to form new echo data for the pixel, i.e., step 12. The new echo data is processed again according to the initial steps, i.e., steps 2 to 6.

[0058] Steps 9 to 10, finally counting all pixel times and maximum count values ​​and calculating the distance image and intensity image, include: using the above-mentioned processing method to process and obtain the effective time and maximum count value of the pixel, counting all pixels and calculating the distance image and intensity image, performing median filtering on the distance image and intensity image, and further completing image noise suppression. The distance image and intensity image are as follows: Figure 8 As shown, the entire laser active imaging noise processing method steps are then ended.

[0059] Finally, in order to show the denoising imaging effect, the range image and intensity image extracted by the peak method are shown, such as Fig. 9 As shown in the figure, it can be obviously found that the proposed processing method has a better imaging effect.

[0060] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A laser active imaging noise processing method, characterized in that: The following steps are involved: Obtain the noise photoelectron count rate using imported single-photon laser imaging data or known noise parameters of the detection scene; According to the laser imaging image data collected by multiple frames, a coarse time resolution histogram of pixels and its displacement histogram are constructed; The target scene recognition threshold distribution is obtained by combining the coarse time resolution, the number of pulse accumulations, the noise photoelectron count rate, and the required pulse accumulation false alarm probability; Compare the statistical counts of the coarse time resolution histogram and its misaligned histogram with the target scene recognition threshold distribution; determine whether the statistical counts of the coarse time resolution histogram and its misaligned histogram meet the recognition threshold requirement; When the recognition threshold requirement is met, the effective echo rough position of the pixel is quickly extracted, and the precise time and maximum count value of the pixel are quickly extracted by using the echo rough position and the fine time resolution histogram of the pixel through the sliding window weighting method; When the recognition threshold requirement is not met, it is determined whether the number of spatial correlation field correlations that have been executed meets the requirement. If so, it is assumed that this pixel has no echo. At this time, the time and maximum count are set to empty or 0, and the corresponding distance value and intensity value are also empty or 0. If it is not met, the spatial neighborhood correlation characteristics are used to select the neighborhood data around the pixel for supplementation to form new echo data of the pixel, and the new echo data is processed again according to the initial steps; Finally, all pixel times and maximum count values ​​are counted, and distance and intensity images are calculated, and median filtering is performed on the image.

2. The method according to claim 1, characterized in that The method of obtaining the noise photoelectron count rate by using the imported single-photon laser imaging data or the known noise parameters of the detection scene includes: the imported single-photon laser imaging data is based on multiple frames of laser imaging data collected by a single-photon array detector, and the single-photon array detector performs a statistical histogram on all data or a sampled portion of data by working in a gated mode, and fits the statistical histogram using a single detection probability model in the gated mode, and the fitting formula is as follows: Where P SFA_gate is the false alarm probability of single pulse detection, which is the ratio of the statistical count of the nth time grid in the statistical histogram to the total pulse accumulation number. is the noise photoelectron counting rate, Δt is the time of the statistical histogram, and then the noise photoelectron counting rate is extracted using the fitting formula or based on a known noise parameter, the background noise photon count rate and dark count rate The noise photoelectron count rate is calculated by combining the detector detection efficiency η, and the formula is as follows:

3. The method according to claim 2, characterized in that The method of constructing a coarse time resolution histogram of pixels and its misalignment histogram based on the laser imaging image data collected by multiple frames includes: using the multi-frame accumulated data of the pixels, combined with the set coarse time resolution, to construct a coarse time resolution histogram of pixels and its misalignment histogram, wherein the coarse time resolution is 6t fwhm / 2.355·2, where t fwhm is the full width at half maximum of Gaussian laser, t fwhm / 2.355 is the standard deviation of Gaussian laser, 6t fwhm / 2.355 contains 99.7% of the energy of the laser.

4. The method according to claim 3, characterized in that The method combines the coarse time resolution, the number of pulse accumulation, the noise photoelectron count rate and the pulse accumulation false alarm probability that meets the requirements to calculate and obtain the target scene recognition threshold distribution, including: when the coarse time resolution and the noise photoelectron count rate are known, the single pulse detection false alarm probability distributed along the time grid is obtained by formula (1), and then the false alarm probability after multiple pulse accumulation is obtained by combining the multiple pulse accumulation numbers and the recognition threshold, and the relationship between them is as follows: In the formula, k th is the threshold for identifying the target within the time gate, 1≤k th ≤N, the index of k′ ranges from 0 to k th -1, N is the pulse accumulation number, It represents the number of combinations of k′ different elements taken out from N different elements each time. By setting the false alarm probability after multiple pulses and the false alarm probability of single pulse detection distributed over the time grid, the target scene recognition threshold distribution is obtained.

5. The method according to claim 4, characterized in that The comparing the statistical counts of the coarse time resolution histogram and its misalignment histogram with the target scene recognition threshold distribution includes: comparing the generated coarse time resolution histogram and its misalignment histogram statistical counts distributed over time with the already acquired recognition threshold distribution distributed over time.

6. The method according to claim 5, characterized in that The method of judging whether the statistical counts of the coarse time resolution histogram and its misaligned histogram meet the recognition threshold requirement includes: comparing the statistical counts of the coarse time resolution histogram and its misaligned histogram with the recognition threshold distribution, if any one of the two constructed histogram statistical counts meets the recognition threshold requirement, it can be judged that the echo of the pixel is a valid echo, otherwise it is judged whether the number of spatial correlation field related data reorganizations that have been executed meets the requirement.

7. The method according to claim 6, characterized in that When the recognition threshold requirement is met, the effective echo rough position of the pixel is quickly extracted, and the precise time and maximum count of the pixel are quickly extracted by using the echo rough position and the fine time resolution histogram of the pixel through a sliding window weighting method.

8. The method according to claim 7, characterized in that When the recognition threshold requirement is not met, the number of spatial correlation field data reorganizations is performed to see if the requirement is met. If so, it is assumed that this pixel has no echo. At this time, the time and the maximum count are set to empty or 0, and the corresponding distance value and intensity value are also empty or 0. If it is not met, the spatial neighborhood correlation characteristics are used to select the neighborhood data around the pixel for supplementation to form new echo data for the pixel, and the new echo data is processed again according to the initial steps.

9. The method according to claim 8, characterized in that The counting of all pixel times and maximum count values, and calculation to form a distance image and an intensity image, includes: obtaining the effective time and maximum count value of the pixel, counting all pixels and calculating to form a distance image and an intensity image, performing median filtering on the distance image and the intensity image, and further completing image noise suppression.

10. A system for implementing the method according to any one of claims 1 to 9.

Images