Radiation source signal sorting method and device based on point cloud detection, equipment and medium

Abstract

The application discloses a radiation source signal sorting method and device based on point cloud detection, equipment and a medium, the method comprises the following steps: in response to the received radar pulse sequence, selecting a plurality of preset value range transformation length, using different transformation length transformation to obtain a two-dimensional scatter diagram; using random sampling to extract a feature curve from the obtained two-dimensional scatter diagram; merging the pulse sequences with consistent slopes in the two-dimensional scatter diagram; extracting the pulse repetition interval of the pulse sequence and determining the radiation source type. The application can realize high-precision sorting in the multi-radiation source scene of fixed, jitter and uneven PRI.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of radar signal processing, and particularly relates to a radiation source signal sorting method and device based on point cloud detection, equipment and a medium. BACKGROUND

[0002] The current main radar signal sorting methods are mainly divided into sorting algorithms based on inter-pulse parameter characteristics and sorting algorithms based on intra-pulse modulation characteristics.

[0003] The sorting method based on intra-pulse modulation characteristics is to extract high-discrimination features through intra-pulse feature extraction means to complete signal sorting. For example, time-frequency analysis methods such as wavelet transform and short-time Fourier transform can analyze and extract the time-frequency distribution and the convergence of wavelet coefficients of the signal to realize the recognition of different modulation signals. However, this kind of method has problems such as large amount of original data and algorithm operation, and difficulty in obtaining intra-pulse data, so in actual engineering application, the inter-pulse parameter feature sorting algorithm based on pulse description word is more used.

[0004] The radar signal sorting algorithm based on inter-pulse parameter characteristics generally uses one or more inter-pulse parameters for sorting. This kind of sorting method is mainly divided into two levels. The first level sorting is pre-sorting using pulse description word, that is, multi-parameter sorting. The pulse signal density after pre-processing is diluted, and further main sorting is performed on pulse repetition interval (PRI). After completing the PRI sorting, the third level is entered to analyze special signals such as frequency agile and PRI staggered signals. For example, the sequence difference histogram (SDIF) method estimates the potential PRI by setting a reasonable detection threshold and only needs to perform histogram statistics on the difference values of pulse arrival time. The histogram method has a serious performance decline when encountering PRI sliding and PRI jitter. The plane transformation technology is to transform the mixed signal to a specific two-dimensional plane for processing, to find the relationship between the sub-signal in the plane figure and the signal parameter size, and finally to sort out the sub-signal. The above method uses a single reconnaissance receiver to receive radar countermeasure information, analyzes the parameters of the signal, and finds the similarity of the same radar signal and the difference of different radar signals to complete the sorting. With the increasing complexity of the electromagnetic environment, the serious overlap of pulse parameters and the weak correlation of inter-pulse information make this kind of method have insufficient robustness in the case of multiple radiation sources, and it is difficult to accurately extract the points on the feature curve. SUMMARY

[0005] The application aims to overcome the defects of the prior art and provide a radiation source signal sorting method and device based on point cloud detection, equipment and a medium, which are designed to quickly, accurately and adaptively extract target signal feature curves and realize high-precision sorting of radiation sources under a single station system.

[0006] The application aims to overcome the defects of the prior art and provide a radiation source signal sorting method and device based on point cloud detection, equipment and a medium, which are designed to quickly, accurately and adaptively extract target signal feature curves and realize high-precision sorting of radiation sources under a single station system.

[0007] A radiation source signal sorting method based on point cloud detection, the method comprising:

[0008] In response to the received radar pulse sequence, a transform length in a plurality of preset value ranges is selected, and a two-dimensional scatter plot is obtained by using different transform lengths for transformation;

[0009] The obtained two-dimensional scatter plot is used for random sampling to extract a feature curve;

[0010] The pulse sequences with consistent slopes in the two-dimensional scatter plot are merged;

[0011] The pulse repetition interval of the pulse sequence is extracted, and the type of the radiation source is determined.

[0012] Further, the pulse sequences with consistent slopes include parallel pulse sequences and pulse sequences with slopes differing by no more than a certain threshold value.

[0013] Further, the random sampling of the obtained two-dimensional scatter plot to extract a feature curve specifically comprises:

[0014] The maximum number of iterations is calculated;

[0015] A plurality of sample points are randomly selected in the current two-dimensional scatter plot, and the slope and intercept of a straight line are calculated by selecting any two of the sample points;

[0016] The distances of the sample points to the calculated straight line are calculated, and the sample points with distances less than a preset threshold value are counted as inliers of the current iteration;

[0017] When the number of inliers is the largest and the proportion of the inliers to the total sample points is greater than a certain threshold value, the current model is saved as an optimal model;

[0018] The iteration is stopped when the number of iterations reaches the maximum number of iterations.

[0019] Further, the calculation of the maximum number of iterations comprises:

[0020] When the least number of sample points is used for each calculation of the model, the probability that at least one of the selected points is an outlier is

[0021] In the case of k iterations, The probability that the model is calculated for k iterations is at least sampled to one outer point, and the probability that the correct N points are sampled to calculate the correct model is:

[0022]

[0023] The maximum iteration number k is calculated as:

[0024]

[0025] The minimum sample point number N min , the inlier ratio inlierRatio, the maximum distance of the inlier distance curve maxDistance, and the confidence P.

[0026] Further, the method comprises the following steps of:

[0027] The pulse repetition interval of the extracted sequence is obtained by histogram statistics.

[0028] Further, the method comprises the following steps of:

[0029] First-order difference operation is performed on the pulse arrival time of the obtained sequence, and the difference result is a potential pulse repetition interval value;

[0030] The frequency of each difference value is counted by histogram statistics, and peak value extraction is performed;

[0031] If one peak value appears, it indicates that the corresponding radiation source is a radiation source with fixed pulse repetition interval, if multiple peak values within a certain range appear, it indicates a radiation source with jittered pulse repetition interval, and if the difference values are dispersed, it indicates different sub-pulse repetition intervals of a radiation source.

[0032] Further, the method determines the type and pulse repetition interval of the radiation source by setting the threshold value for peak value extraction to 0.7 times the maximum peak value of the histogram.

[0033] On the other hand, the application also provides a radiation source signal sorting device based on point cloud detection, which comprises:

[0034] A scatter plot construction module, which selects a plurality of transformation lengths within a plurality of preset value ranges in response to the received radar pulse sequence, and uses different transformation lengths to obtain two-dimensional scatter plots;

[0035] A feature curve extraction module, which extracts feature curves from the obtained two-dimensional scatter plots using random sampling;

[0036] a pulse merging module, which merges pulse sequences with consistent slopes in a two-dimensional scatter plot;

[0037] a radiation source judging module, which extracts a pulse repetition interval of the pulse sequence and determines a radiation source type.

[0038] In another aspect, the present application also provides a computer device, which comprises a processor and a memory, and the memory stores a computer program, which is loaded and executed by the processor to implement any one of the above-mentioned radiation source signal sorting methods based on point cloud detection.

[0039] In another aspect, the present application also provides a computer readable storage medium, which stores a computer program, which is loaded and executed by a processor to implement any one of the above-mentioned radiation source signal sorting methods based on point cloud detection.

[0040] The present application has the following beneficial effects:

[0041] (1) The present application introduces the point cloud detection method of random sampling consensus on the basis of the feature curve obtained by plane transformation, and realizes high-precision and adaptive sorting of radiation source signals under the pulse sequence system received by a single reconnaissance receiver, so as to overcome the problem of weak pulse interval information correlation of single station compared with multi-station sorting.

[0042] (2) Compared with the SDIF algorithm, the present application can more intuitively show the PRI modulation characteristics of the radiation source on the image, enhance the semantic features of the data, and is beneficial to establish the existence of potential radiation sources and identify the PRI modulation type. Compared with the traditional plane transformation method, the present application can more robustly and adaptively extract points on the feature curve in the low signal-to-noise ratio mixed scene of uneven PRI, jittered PRI and fixed PRI, so as to more accurately extract the pulse sequence and improve the PRI estimation precision. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 is a flowchart of the radiation source signal sorting method based on point cloud detection provided by the embodiment of the present application;

[0044] Figure 2 is a histogram statistical result of the first-order difference of the TOA sequence of the uneven radiation source with the PRI uneven frame [1404, 1333];

[0045] Figure 3 is a precision statistical diagram of 20 radiation source sorting of the embodiment of the present application;

[0046] Figure 4 is a structure block diagram of the radiation source signal sorting device based on point cloud detection provided by the embodiment of the present application. DETAILED DESCRIPTION

[0047] Other advantages and effects of the present application can be easily understood by those skilled in the art from the description of the embodiments of the present application. The present application can also be implemented or applied in other different embodiments, and various modifications or changes can be made to the details of the description based on different views and applications without departing from the spirit of the present application. It should be noted that the following examples and features in the examples can be combined with each other without conflict.

[0048] All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative labor are within the scope of protection of the present application.

[0049] The traditional radar signal sorting method has the problems of serious pulse parameter overlap and weak inter-pulse information correlation, which makes it difficult to accurately extract the points on the feature curve.

[0050] To solve the above technical problems, the present application provides the following embodiments of the radiation source signal sorting method, device, equipment and medium based on point cloud detection.

[0051] Embodiment 1

[0052] Reference Figure 1 As Figure 1 shown is a flowchart of the radiation source signal sorting method provided by the present embodiment, which specifically includes the following steps:

[0053] Step 1: Given a sequence of received radar pulse arrival time (TOA) {t1, t2,..., tn} of a string of pulses, the sequence length is n. The transform length W is selected in the range [Wmin, Wmax], and the two-dimensional scatter plot A is obtained by using different transform lengths. Each transform length W corresponds to a scatter plot A. n} sequence length is n. The transform length W is selected in the range [W min ,W max ] and the two-dimensional scatter plot A is obtained by using different transform lengths. Each transform length W j corresponds to a scatter plot A. j The pulse sequence with pulse arrival time in t1 to W is intercepted and displayed in the first row of the two-dimensional plane, the pulse in t1+W to t1+2W is intercepted and displayed in the second row, and so on. The pulse in t1+(n-1)W to t1+nW is displayed in the nth row of the two-dimensional plane. Therefore, the coordinates of A can be obtained as:

[0054]

[0055] Where x i is the horizontal coordinate, y i is the vertical coordinate, and ti is the arrival time value of the i-th pulse corresponding to the point, W j is the current transform length. mod(·) is the modulo operation, and floor(·) is the floor function. The pulse sequence of the same radiation source appears as a characteristic curve on the two-dimensional scatter plot.

[0056] Step 2: Extract the characteristic curve from the obtained two-dimensional scatter plot using the random sample consensus algorithm. The specific steps are as follows:

[0057] Step 2-1: Calculate the number of iterations k. Let N min be the minimum number of sample points of the model, inlierRatio be the proportion of inliers, maxDistance be the maximum distance of the inlier distance curve, and P be the confidence. When N min points are used to calculate the model each time, the probability that at least one outlier is selected is In the case of k iterations, the probability that at least one outlier is sampled in k iterations of the calculation model is

[0058]

[0059] Then it can be deduced that the number of iterations is:

[0060]

[0061] Step 2-2: Randomly select N j sample points from the current two-dimensional scatter plot A min , and randomly select two points to calculate the parameters a and b of the straight line passing through the two points. Parameter a is the slope of the straight line, and parameter b is the intercept of the straight line.

[0062] Step 2-3: Calculate the distance from the N min sample points obtained in step 2-2 to the straight line obtained in step 2-2, and count the sample points whose distance values are less than maxDistance as the inliers obtained in this iteration.

[0063] Step 2-4: When the number of inliers is maximum and the proportion of inliers to total sample points is greater than inlierRatio, save the current model as the optimal model, that is, save the current straight line parameters a best , b best and the inlier index inlierIdx, otherwise do nothing.

[0064] Step 2-5: When the number of iterations meets k, stop iteration, otherwise go back to step 2-1 and continue iteration.

[0065] Step 3: Merge the scatter plots A jThe sequence with consistent slope is a straight line. The sequence with inconsistent slope is a curve. In one aspect, the fixed radiation source is transformed into a plurality of equidistant parallel characteristic curves in the two-dimensional plane. The staggered radiation source is transformed into a plurality of parallel characteristic curves in the two-dimensional plane, and the transformation length is equal to the skeleton period of the staggered radiation source. When the transformation length is an integer multiple of the fixed radiation source, a plurality of equidistant parallel straight lines appear. On the other hand, due to the influence of the radar scanning mode, the same radiation source can only be irradiated by the radar beam within a certain time period, so the period of the received PDW is also related to the beam scanning period, which is specifically shown as a plurality of parallel characteristic curves distributed on the vertical axis in the image. Therefore, the inner points corresponding to the straight lines with similar slopes need to be merged.

[0066] Step 4: Extract the PRI of the sequence by histogram statistics. Perform a first-order difference operation on the pulse arrival time of the sequence obtained by merging in step 3, and the difference result is the potential PRI (pulse repetition interval) value. After histogram statistics of the frequency of each difference value, peak extraction is performed. If a peak value appears, it indicates that the radiation source is a fixed PRI radiation source. If a plurality of peak values within a certain range appear, it is a dithering PRI. If a radiation source with a relatively dispersed interval appears, it is a different sub-PRI of the staggered radiation source. Set maxpeak as the maximum peak value of the histogram, and set the threshold value peakThresh for peak extraction as maxpeak*0.7 to determine the type and PRI of the radiation source.

[0067] The embodiment introduces the point cloud detection method of the random sample consensus method on the basis of the characteristic curve obtained by plane transformation, realizes high-precision and adaptive sorting of the radiation source signal under the pulse sequence system received by a single reconnaissance receiver, and overcomes the problem of weak pulse interval information correlation of single station compared with multiple stations. Compared with the SDIF algorithm, the PRI modulation characteristics of the radiation source can be more intuitively represented on the image, the semantic features of the data are enhanced, the existence of the potential radiation source is established, and the PRI modulation type is identified. Compared with the traditional plane transformation method, the points on the characteristic curve can be more robust and adaptively extracted in the mixed scene of staggered PRI, dithering PRI and fixed PRI with low signal-to-noise ratio, so that the pulse sequence can be more accurately extracted, and the PRI estimation accuracy is improved.

[0068] Embodiment 2

[0069] The embodiment sets up a simulation scene containing 20 radiation sources, each radiation source has different PRI size and different PRI modulation type. The PRI range is 500-5000 μs, and the PRI modulation type includes fixed PRI (TOA measurement accuracy 1%), dithering PRI (dithering rate 10%-20%), and staggered PRI (one staggered frame contains 2-4 sub-PRI). The sequence contains 10% random noise.

[0070] Step 1: Set the pulse arrival time of the received radar pulse sequence as {t1, t2,..., tn} and the sequence length as n = 24039. Select the transform length range as [400, 10000], and use different transform lengths to transform the sorting sequence to obtain a two-dimensional scatter plot. n} sequence length n = 24039. Select the transform length range as [400, 10000], and use different transform lengths to transform the sorting sequence to obtain a two-dimensional scatter plot.

[0071] Set the current transform length as W j = 1021, and intercept the pulse sequence within t1 to W j , which represents the first row of the two-dimensional plane, and intercept the pulse from (t1 + W j ) to (t1 + 2·W j ) in the second row, and so on, and display the pulse from t1 + (n-1)W j to t + nW j in the nth row of the two-dimensional plane. Thus, the coordinates of the scatter points of the two-dimensional scatter plot can be obtained as:

[0072]

[0073] where x i is the horizontal coordinate, y i is the vertical coordinate, t i is the arrival time value of the i-th pulse corresponding to the point, and W j is the current transform length. mod(·) is the modulo operation, and floor(·) is the floor operation. The pulse sequence of the same radiation source appears as a characteristic curve on the two-dimensional scatter plot.

[0074] Step 2: Extract the characteristic curve from the obtained two-dimensional scatter plot using the random sample consensus algorithm. The random sample consensus algorithm is an iterative algorithm for estimating mathematical parameters from a set of data containing outliers. Inliers refer to the data that form the estimated model, and outliers refer to the noise in the data, such as mismatches in matching and outliers in the estimated curve. The specific steps are as follows:

[0075] Step 2-1: Calculate the number of iterations k. Set the minimum number of model sample points N min = 2, the inlier ratio inlierRatio = 3%, the maximum distance of inliers from the curve maxDistance = 1, and the confidence P = 0.99. When N min points are used to calculate the model each time, the probability that at least one outlier is selected is (1-0.03) 2 . In the case of k iterations, the probability that at least one outlier is sampled in the k iterations is (1-0.03 2 ) k . It can be deduced that the number of iterations k is:

[0076]

[0077] Step 2-2: In the current two-dimensional scatter plot A j Two sample points are randomly selected from the data. The parameters a and b of the line passing through these two points are then calculated.

[0078] Step 2-3: Calculate the distance from the two sample points taken in Step 2-2 to the line obtained in Step 2-2, and count the sample points with a distance value less than 1, which are the interior points obtained in this iteration.

[0079] Steps 2-4: When the number of interior points is at its maximum and its proportion to the total number of sample points is greater than 3%, save the current model as the optimal model, that is, save the current line parameter a. best =-400.0658,b best =12687 and the internal point index inlierIdx, otherwise do nothing.

[0080] Step 2-5: Stop iterating when the number of iterations reaches 5114; otherwise, return to step 2-1 to continue iterating.

[0081] Step 3: Merge scatter plot A j The sequence has a consistent slope. On one hand, after planar transformation, the staggered radiation source appears as multiple parallel characteristic curves on a two-dimensional plane, and the transformation length is equal to the skeleton period of the staggered radiation source. On the other hand, due to the influence of the radar scanning mode, the same radiation source can only be illuminated by the radar beam within a specific time period. Therefore, the period of the received PDW is also related to the beam scanning period, which is specifically represented in the image as multiple characteristic curves distributed parallel to each other on the vertical axis. Therefore, it is necessary to merge the interior points corresponding to lines with similar slopes.

[0082] Step 4: Utilize histogram statistics to obtain the extracted PRI values ​​from the sequence. Perform a first-order difference operation on the pulse arrival times of the merged sequence obtained in Step 3. The difference results are the potential PRI values. After statistically analyzing the frequency of each difference value using a histogram, peak extraction is performed. If a single peak appears, it indicates a radiation source with a fixed PRI. If multiple peaks appear within a certain range, it indicates a jittery PRI. If radiation sources are spaced far apart, they indicate different sub-PRIs of a staggered radiation source. (Refer to...) Figure 2 ,like Figure 2 The figure shows the histogram statistics of the TOA sequence of the staggered radiation source with PRI staggered frames [1404, 1333] after first-order difference in this embodiment. Let maxpeak be the maximum peak value of the histogram. Figure 2The maximum peak is 57. The threshold value peakThresh = 57*0.7 = 40 is set by setting the peak extraction threshold value to determine the type of radiation source and the PRI. The two largest peaks are the two sub-PRI of this uneven radiation source, that is, 1404 and 1333, and the peak corresponding to 2737 is the skeleton period thereof, which is caused by the missed sorting. However, considering the subsequent positioning accuracy, the sorted sequence should be accurately from the same radiation source as a priority, and the method meets this demand.

[0083] With reference to Figure 3 As shown in Figure 3 FIG. 20 is a precision statistical diagram of 20 radiation source sorting according to the embodiment. As can be seen, in the multi-radiation source scene of fixed, jitter and uneven PRI, the method provided in the embodiment can realize high-precision sorting.

[0084] Embodiment 3

[0085] With reference to Figure 4 As shown in Figure 4 FIG. 20 is a precision statistical diagram of 20 radiation source sorting according to the embodiment. As can be seen, in the multi-radiation source scene of fixed, jitter and uneven PRI, the method provided in the embodiment can realize high-precision sorting.

[0086] A scatter plot construction module, the scatter plot construction module is responsive to the received radar pulse sequence, selects a plurality of preset value ranges of transformation length, and uses different transformation lengths to transform to obtain a two-dimensional scatter plot;

[0087] A feature curve extraction module, the feature curve extraction module extracts a feature curve from the obtained two-dimensional scatter plot using random sampling;

[0088] A pulse merging module, the pulse merging module merges pulse sequences with consistent slopes in the two-dimensional scatter plot;

[0089] A radiation source judgment module, the radiation source judgment module extracts a pulse repetition interval of the pulse sequence and determines a radiation source type.

[0090] Embodiment 4

[0091] The preferred embodiment provides a computer device, which can implement the steps in any embodiment of the radiation source signal sorting method based on point cloud detection provided in the embodiments of the application, and thus can implement the beneficial effects of the radiation source signal sorting method based on point cloud detection provided in the embodiments of the application. Details are described in the foregoing embodiments, which will not be described here.

[0092] Embodiment 5

[0093] Those skilled in the art can understand that all or part of the steps of various methods of the above embodiments can be completed by instructions or by controlling relevant hardware by the instructions, which can be stored in a computer readable storage medium and loaded and executed by a processor. To this end, the embodiments of the present application provide a storage medium, wherein a plurality of instructions are stored, which can be loaded by a processor to execute the steps of any embodiment of the radiation source signal sorting method based on point cloud detection provided by the embodiments of the present application.

[0094] The storage medium can include a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0095] Since the instructions stored in the storage medium can execute the steps in any embodiment of the radiation source signal sorting method based on point cloud detection provided by the embodiments of the present application, the beneficial effects that can be achieved by any radiation source signal sorting method based on point cloud detection provided by the embodiments of the present application can be achieved, which are described in detail in the foregoing embodiments and will not be described here.

[0096] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for sorting signals of radiation sources based on point cloud detection, characterized in that, The method comprises: in response to the received radar pulse sequence, selecting a transform length in a plurality of preset value ranges, and using different transform lengths to obtain a two-dimensional scatter plot; extracting a feature curve from the obtained two-dimensional scatter plot using random sampling; merging pulse sequences with consistent slopes in the two-dimensional scatter plot; extracting a pulse repetition interval of the pulse sequence and determining a radiation source type; the extracting a feature curve from the obtained two-dimensional scatter plot using random sampling specifically comprises: calculating a maximum iteration number; randomly selecting a plurality of sample points in the current two-dimensional scatter plot, and calculating a slope and an intercept of a corresponding straight line by selecting any two points; calculating distances of the plurality of sample points to the calculated straight line, and counting sample points with distances less than a preset threshold as inliers of the current iteration; when the number of inliers is maximum and the proportion of the inliers to total sample points is greater than a certain threshold, saving the current model as an optimal model; stopping iteration when the iteration number reaches the maximum iteration number.

2. The point cloud detection based radiation source signal sorting method of claim 1, wherein, The pulse sequences with consistent slopes include parallel pulse sequences and pulse sequences with slopes differing by no more than a certain threshold.

3. The point cloud detection based radiation source signal sorting method of claim 1, wherein, The calculating a maximum iteration number comprises: When each time the model uses the minimum number of sample points, the probability that the selected points have at least one outlier is ; In In the case of the second iteration, For The probability that the model calculated in the second iteration is correct is: The probability that the model calculated in the second iteration is correct is: ； the maximum iteration number k is calculated in the following manner: ； Wherein, the minimum sample point number , the inner point proportion , the inner point distance curve maximum distance , the confidence .

4. The point cloud detection based radiation source signal sorting method of claim 1, wherein, The extracting a pulse repetition interval of the pulse sequence and determining a radiation source type comprises: obtaining the extracted pulse repetition interval of the sequence by histogram statistics.

5. The point cloud detection based radiation source signal sorting method of claim 4, wherein, The obtaining the extracted pulse repetition interval of the sequence by histogram statistics specifically comprises: performing a first-order difference operation on pulse arrival times of the merged sequence, and the difference result is a potential pulse repetition interval value; performing peak value extraction after histogram statistics of frequencies of each difference value; if one peak value appears, it indicates that the corresponding radiation source is a radiation source with a fixed pulse repetition interval, if a plurality of peak values within a certain range appear, it is a radiation source with a jittered pulse repetition interval, and if interval-dispersed difference values appear, it is different sub-pulse repetition intervals of a staggered radiation source.

6. The point cloud detection based radiation source signal sorting method of claim 5, wherein, The method determines the type and pulse repetition interval of the radiation source by setting a threshold value for peak value extraction to be 0.7 times the maximum peak value of the histogram.

7. A radiation source signal sorting device based on point cloud detection, characterized in that, The device comprises: a scatter plot construction module, which, in response to a received radar pulse sequence, selects a transform length in a plurality of preset value ranges, and uses different transform lengths to obtain a two-dimensional scatter plot; a feature curve extraction module, which extracts a feature curve from the obtained two-dimensional scatter plot using random sampling; a pulse merging module, which merges pulse sequences with consistent slopes in the two-dimensional scatter plot; a radiation source judgment module, which extracts a pulse repetition interval of the pulse sequence and determines a radiation source type; the extracting a feature curve from the obtained two-dimensional scatter plot using random sampling specifically comprises: calculating a maximum iteration number; randomly selecting a plurality of sample points in the current two-dimensional scatter plot, and calculating a slope and an intercept of a corresponding straight line by selecting any two points; calculating distances of the plurality of sample points to the calculated straight line, and counting sample points with distances less than a preset threshold as inliers of the current iteration; when the number of inliers is maximum and the proportion of the inliers to total sample points is greater than a certain threshold, saving the current model as an optimal model; stopping iteration when the iteration number reaches the maximum iteration number.

8. A computer device, comprising: The computer device comprises a processor and a memory, and the memory stores a computer program, which is loaded and executed by the processor to implement the point cloud detection based radiation source signal sorting method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which is loaded and executed by the processor to implement the point cloud detection based radiation source signal sorting method according to any one of claims 1 to 6.

Images