A point extraction and judgment method based on multi-dimensional target features

Abstract

The present invention provides a point trace extraction and judgment method based on multi-dimensional target features, including filtering the input video signal by distance and azimuth, and secondary filtering by number; setting a distance\azimuth size frame and an amplitude threshold frame; sampling and quantizing the filtered signal to extract its multi-dimensional feature information; making a preliminary joint judgment based on the quantized distance and azimuth feature information in combination with the distance\azimuth size frame, identifying the points with quality and the points without quality, and marking them; combining with the actual target, statistically forming the multi-dimensional feature information of each type of target when the target point trace is judged to have quality with high probability; thereby forming the logic of point trace extraction and judgment, extracting and judging the input point trace data; according to the results of extraction and judgment, adding a credibility characteristic mark to the point trace data, and outputting it to the back-end reference. The scheme of the present invention can improve the processing performance of the system to a certain extent while improving the correct probability of point trace extraction and judgment.

Description

Technical Field

[0001] The invention relates to the field of scene surveillance radar, and in particular to a point trace extraction and judgment method based on multi-dimensional target features. Background Art

[0002] With the development of the aviation transportation industry, the scale of airlines is growing, the annual throughput of airports is increasing year by year, the increase in flight business volume has led to an increase in ground service vehicles, and the airport surface traffic and operating environment are becoming increasingly complex. Traditional visual surveillance methods can no longer meet the requirements of airport surface surveillance. In addition, due to adverse weather conditions such as fog, haze, rain, snow, and dust, airports urgently need new surveillance methods to achieve surface surveillance and provide controllers with airport surface traffic situation awareness under various adverse weather conditions, night vision, and poor visibility.

[0003] Europe has proposed an important technical means for airport surface monitoring: surface surveillance radar. As an indispensable independent means of monitoring the airport surface, surface surveillance radar can provide controllers with status awareness of airport traffic conditions around the clock and in all weather conditions.

[0004] The principle of surface surveillance radar is equivalent to that of primary radar, which extracts and determines the target by emitting electromagnetic waves and receiving the reflected signal from the target. Among them, the extraction and determination of point traces are very critical and important processing links. Point trace data is the basic data for target determination and the main data input for subsequent track establishment, update and maintenance. The wrong extraction or loss of point traces greatly affects the determination of the target and the formation and maintenance of the track. Point trace processing is the basis of radar data processing, and the extraction and determination of point traces occupies a very important position in the radar data processing process.

[0005] Most of the existing point track extraction and determination methods are completed in the link of track establishment and maintenance, which is completed through the following steps:

[0006] Step 1: Extract according to the input data amplitude, and output the data that reaches the amplitude threshold as point traces to the point-to-point navigation association link;

[0007] Step 2: Establish a point track association gate based on track prediction, and use the association gate as the condition for point track filtering and selection. When the input data falls within the association gate, it is initially determined to be a possible valid point track. If it is outside the association gate, it is not determined to be a valid point track of this track.

[0008] Step 3: Select and judge according to the number of points in the associated gate. If there is only one point in the associated gate, it is judged as a valid point. If there are multiple points in the associated gate, the one closest to the predicted track or with the largest amplitude is selected as the valid point.

[0009] The existing methods have the following problems:

[0010] 1. Before associating point tracks with tracks, the input data is simply extracted based on the threshold, resulting in some input data that is not within the system requirement range but has a large amplitude, or data such as clutter, noise, multipath interference, etc. that is within the system requirement range but is not the target. Due to the large amplitude, they also enter the point-to-track association judgment process, resulting in a large amount of data processed by the backend, consuming more processing resources and time for data that is not within the system requirement range. The processing of useless clutter, noise, multipath interference and other data has reduced the system processing performance to a certain extent;

[0011] 2. The extraction and determination of points only rely on whether they fall into the correlation gate and the amplitude of the points. The establishment of the correlation gate is usually based on target features such as distance and orientation. The target characteristics based on it are very limited, and there is a high probability that false points cannot be filtered out, resulting in false points being determined as valid points.

[0012] 3. When false traces caused by clutter, noise, multipath interference, etc. fall into the association gate, if there are multiple traces in the association gate, it may cause trace misjudgment, especially when the false traces caused by the multipath effect are close to the real traces, there is a high probability of trace misjudgment; when the amplitude of the false traces is large, the probability of trace misjudgment is high;

[0013] 4. The extraction and judgment of point traces rely on the correlation gate established by track prediction, but the track prediction is not absolute. There are errors caused by measurement errors and different prediction algorithms. That is, there are errors in establishing the correlation gate itself. In addition, the input point traces are only extracted and judged based on the amplitude, which further increases the probability of false point traces and erroneous correlation. Summary of the invention

[0014] In view of the problems existing in the prior art, a point trace extraction and judgment method based on multi-dimensional target features is provided. Before establishing the point navigation association gate, a point trace extraction and judgment processing link based on multi-dimensional target features is added.

[0015] The technical solution adopted by the present invention is as follows: a point trace extraction and judgment method based on multi-dimensional target features, characterized in that it includes the following steps:

[0016] Step 1: filter the input video signal once according to the distance range and azimuth range that the system needs to focus on, and filter out the signals outside the range;

[0017] Step 2, performing a second filtering on the video signal after the first filtering according to the number of times the video signal is received and accumulated;

[0018] Step 3: According to the input video signal amplitude and in combination with the system processing model, an amplitude threshold frame based on the amplitude feature is set;

[0019] Step 4: According to the characteristics of the video signal reflected by the actual target, set the distance\azimuth size frame based on the distance and azimuth characteristics;

[0020] Step 5: Sampling and quantizing the video signal after secondary filtering to extract its multi-dimensional feature determination information;

[0021] Step 6: Based on the distance and range feature information of the video signal and the distance\azimuth size frame, the traces are divided into quality traces and non-quality traces, and marked;

[0022] Step 7, combining the video signals of multiple known real targets, executing the process of step 1 to step 6, obtaining the marked traces, and counting the multi-dimensional feature information corresponding to the real target when the proportion of the traces with quality is greater than 90%, to form a multi-dimensional feature information set;

[0023] Step 8: Based on the multi-dimensional feature information set obtained in step 7, the amplitude threshold frame set in step 3 and the distance / azimuth size set in step 4 are fed back and updated to obtain the logic of point extraction and determination;

[0024] Step 9: Compare the sampled and quantized input trace data with the logic of trace extraction and determination, and take the trace data that meets the logic of trace extraction and determination as the extraction and determination result;

[0025] Step 10: According to the results of extraction and determination, a credibility characteristic mark is added to the point trace data and output to the back-end for reference.

[0026] Furthermore, the step 2 includes the following sub-steps:

[0027] Step 21, according to the beam dwell time and the system transmission and reception signal cycle, calculate the number of video signals that should be received at the corresponding position of the target when the target exists during the beam dwell time, which is M;

[0028] Step 22: according to the system processing tolerance requirement, set the minimum number of video signals to be received, N, N≤M;

[0029] Step 23: Perform reception accumulation statistics within the beam dwell time for the input video signal after the first filtering, and the number of times the target actually receives the video signal at the corresponding position is L;

[0030] Step 24, compare the number of times L of actually received video signals with the set number of times N of video signals that should be received at least, if L≥N, retain the input video signal received this time, if L<N, filter the input video signal.

[0031] Furthermore, the step 4 includes the following sub-steps:

[0032] Step 41: According to the characteristics of the video signal reflected by the actual target, combined with the system's minimum distance resolution unit, minimum azimuth resolution unit, target type, actual size of each type of target, radiation angle, and target motion posture, set the minimum distance resolution unit number Z occupied by the video signal reflected by the target in the distance dimension. min and the maximum number of distance resolution units Z max , set the minimum number of azimuth resolution units and γ occupied by the video signal reflected by the target in azimuth angle min Maximum number of azimuth resolution units γ max , set the minimum number of distance resolution units occupied by the video signal reflected by the target on the minimum azimuth size resolution unit D min and the maximum number of distance resolution units D max ;

[0033] Step 42: According to the set target distance size range Z min ~Z max , target azimuth angle range γ min ~γ max , the distance dimension range D of the target azimuth angle min ~D max , create a distance\azimuth dimension box.

[0034] Furthermore, the multidimensional feature information in step 5 includes: distance feature information, azimuth feature information, and the number of sampling points within the amplitude threshold frame, the proportion of the sampling points within the amplitude threshold frame in the total number of sampling points, the number of sampling points of the signal within the distance\azimuth size frame, the proportion of the sampling points within the distance\azimuth size frame in the total number of sampling points, the number of sampling points of the signal within the distance\azimuth size frame and within the amplitude threshold frame, and the proportion of the sampling points within the distance\azimuth size frame and within the amplitude threshold frame to the total number of sampling points.

[0035] Furthermore, in step 6, the judgment method is: when the distance dimension of the input video signal is within the distance dimension range, and the azimuth feature of the input video signal is greater than or equal to the minimum azimuth angle range, and the distance dimension of the input video signal at the azimuth angle is greater than or equal to the distance dimension at the minimum azimuth angle, and the azimuth dimension of the input video signal is less than or equal to the maximum azimuth angle range or the distance dimension per unit angle is less than or equal to the distance dimension at the maximum azimuth angle, it is judged as a quality point trace, otherwise it is judged as a qualityless point trace.

[0036] Furthermore, in step 7, the multidimensional feature information set is composed of multidimensional feature information of multiple targets, including: the distance size range, azimuth angle range, amplitude range, the minimum number of sampling points within the amplitude threshold frame, the proportion of the minimum number of sampling points within the amplitude threshold frame to the total number of sampling points, the proportion of the minimum number of sampling points within the distance\azimuth size frame to the total number of sampling points, and the proportion of the minimum number of sampling points within the distance\azimuth size frame and within the amplitude threshold frame to the total number of sampling points.

[0037] Furthermore, in step 8, the logic acquisition method for point trace extraction and determination is:

[0038] The distance\azimuth size frame is adjusted according to the distance size range and azimuth angle range of the target; the amplitude threshold frame is adjusted according to the amplitude range of the target; and the minimum number of sampling points in the amplitude threshold frame, the proportion of the minimum number of sampling points in the amplitude threshold frame to the total number of sampling points, the proportion of the minimum number of sampling points in the distance\azimuth size frame to the total number of sampling points, and the proportion of the minimum number of sampling points in the distance\azimuth size frame and in the amplitude threshold frame to the total number of sampling points are used as judgment parameters; the adjusted distance\azimuth size frame, amplitude threshold frame and judgment parameters are combined to form the logic of point trace extraction and judgment.

[0039] Furthermore, in step 9, the input trace and the trace extraction are compared with all conditions in the logic of judgment. If all conditions are met, it is judged as a valid trace and the feature information (quantized information) is extracted; otherwise, it is judged as an invalid trace.

[0040] Compared with the prior art, the beneficial effects of adopting the above technical solution are:

[0041] 1. It can filter out input signals that are not within the system requirements, effectively reduce additional system inputs, reduce back-end processing loads, save processing resources and time, and improve the system's processing performance to a certain extent;

[0042] 2. Utilize the randomness and suddenness of clutter, noise and multipath interference, and effectively filter out interference data input by cumulative number of times;

[0043] 3. Collect, quantify and extract multi-dimensional target features of input data, providing more comprehensive and multi-dimensional data support for subsequent judgment. It can set filtering, identification and judgment conditions according to multiple dimensions, filter out noise and interference that cannot be filtered out by conventional methods, and improve the correct probability of point extraction and judgment;

[0044] 4. Through multi-dimensional target feature filtering, threshold measurement, collection quantification, preliminary judgment, measured statistics, feedback adjustment, logic judgment, credibility identification and other operations, before point navigation association, the input data is extracted and judged, and both quality point traces and non-quality point traces are output to the back end, and the point traces are marked with credibility, which not only retains the output of the point traces to the maximum extent, but also provides more reference target characteristics for point navigation association, making point navigation association more reliable and reducing the probability of false point traces and error correlation;

[0045] 5. The multi-dimensional target feature information obtained based on actual use cases and test statistics is more real, reliable, effective, and has strong verifiability. Its reliability and credibility are greatly improved, which greatly increases the probability of extracting valid points and reduces the probability of misjudgment of false points;

[0046] 6. It has a processing mechanism with continuous feedback adjustment, which can further improve the accuracy probability of point extraction and judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a flow chart of a point trace extraction and judgment method based on multi-dimensional target features proposed by the present invention. DETAILED DESCRIPTION

[0048] The present invention is further described below in conjunction with the accompanying drawings.

[0049] like Figure 1 As shown, the embodiment proposes a point trace extraction and judgment method based on multi-dimensional target features, including the following process:

[0050] Step 1: filter the input video signal once according to the distance range and azimuth range that the system needs to focus on, and filter out the signals outside the range;

[0051] Step 2, performing a second filtering on the video signal after the first filtering according to the number of times the video signal is received and accumulated;

[0052] Step 3: According to the input video signal amplitude and in combination with the system processing model, an amplitude threshold frame based on the amplitude feature is set;

[0053] Step 4: According to the characteristics of the video signal reflected by the actual target, set the distance\azimuth size frame based on the distance and azimuth characteristics;

[0054] Step 5: Sampling and quantizing the video signal after secondary filtering to extract its multi-dimensional feature determination information;

[0055] Step 6: Based on the distance and range feature information of the video signal and the distance\azimuth size frame, the traces are divided into quality traces and non-quality traces, and marked;

[0056] Step 7, combining the video signals of multiple known real targets, executing the process of step 1 to step 6, obtaining the marked traces, and counting the multi-dimensional feature information corresponding to the real target when the proportion of the traces with quality is greater than 90%, to form a multi-dimensional feature information set;

[0057] Step 8: Based on the multi-dimensional feature information set obtained in step 7, the amplitude threshold frame set in step 3 and the distance / azimuth size set in step 4 are fed back and updated to obtain the logic of point extraction and determination;

[0058] Step 9: Compare the sampled and quantized input trace data with the logic of trace extraction and determination, and take the trace data that meets the logic of trace extraction and determination as the extraction and determination result;

[0059] Step 10: According to the results of extraction and determination, a credibility characteristic mark is added to the point trace data and output to the back-end for reference.

[0060] Specifically, step 1 includes:

[0061] Step 11: According to the system requirements, for the input video signal, according to the distance range of interest of the system requirements, with the minimum action distance and the maximum action distance of interest as the filtering threshold, filter out the input video signal that is less than the minimum action distance of interest and greater than the maximum action distance of interest, and only retain the input video signal within the distance range of interest;

[0062] Step 12: According to system requirements, for the input video signal, according to the azimuth range of interest of the system requirements, use the azimuth range of interest (which may be one or more azimuth ranges) as the filtering threshold, filter out the input video signals that are not within the azimuth range of interest, and only retain the input video signals within the azimuth range of interest.

[0063] Step 2 includes:

[0064] Step 21: According to the beam dwell time and the period of transmitting and receiving signals of the system, calculate the number of video signals that should be received at the corresponding position of the target during the beam dwell time if the target exists, which is M;

[0065] Step 22: according to the system processing tolerance requirement, set the number of video signal reception accumulation times N (N≤M), that is, if the target exists, the number of video signals that should be received at least at the position corresponding to the target is N;

[0066] Step 23: Perform cumulative statistics of reception during the beam dwell time based on the input video signal filtered by distance and azimuth, and the number of times the video signal is actually received at the corresponding position of the target is L;

[0067] Step 24: Compare the number of times L that the video signal is actually received with the set number of times N that the video signal should be received at least. If L≥N, retain the input video signal received this time. If L<N, determine that the input video signal received this time may be an accidental reflection signal caused by random / burst multipath effects or random / burst strong interference, and filter this input video signal.

[0068] In view of the problem that the prior art simply extracts data based on the amplitude threshold without filtering and judging the input data in detail, the input data that is not within the system requirement range but has a large amplitude, or the data that is within the system requirement range but is not the target clutter, noise, multipath interference, etc., also enters the point-to-point association judgment process due to the large amplitude, resulting in a large amount of data processed by the back-end, consuming more processing resources and time for processing useless data outside the requirement range, clutter, noise, multipath interference, etc., which reduces the processing performance of the system to a certain extent. In this embodiment, the input video signal is filtered twice through step 1 and step 2:

[0069] Step 1 filters the input video signal by distance and azimuth to remove input signals that are not within the system requirement range. This can effectively reduce additional system input, reduce back-end processing load, save processing resources and time, and does not affect the data within the system requirement range, thus improving the system processing performance to a certain extent.

[0070] Step 2 further performs a second filtering according to the requirements of the video signal reception accumulation times. For random, sudden, strong clutter, noise and multipath interference, due to their randomness and suddenness, their reception accumulation times are usually much smaller than the reception accumulation times of the real target. Even if the input data is within the system requirements, it can be effectively filtered out.

[0071] Specifically, step 3 is as follows: according to the video signal input after two filtering, the amplitude information is extracted, and the amplitude threshold frame based on the amplitude feature is set in combination with the system processing model, including the minimum amplitude threshold A min and the maximum amplitude threshold A max

[0072] In this embodiment, the system processing model in step 3 includes multiple system processing models such as taking an average value, taking a maximum value, taking a minimum value, taking an operational weighted value, etc.

[0073] Step 4 The process of establishing the distance\azimuth dimension frame is as follows:

[0074] Step 41: According to the characteristics of the video signal reflected by the actual target, combined with the system's minimum distance resolution unit, target type, actual size of each type of target, radiation angle, target motion posture and other characteristics, set the minimum number of distance resolution units and the maximum number of distance resolution units occupied by the video signal reflected by the target in terms of distance size. For example, if the system's minimum distance resolution unit is X meters, and the actual length of the target is Y meters, according to the characteristics of the video signal reflected by the actual target, it may change due to factors such as the radiation angle and the target posture, resulting in the video signal reflected by the target being Z meters in terms of distance size. With X meters as the minimum distance resolution unit, set the number of X meters occupied by Z meters, and the minimum number Z meters. min and the maximum number Z max ;

[0075] Step 42: According to the characteristics of the video signal reflected by the actual target, combined with the system's minimum distance size resolution unit, target type, actual size of each type of target, radiation angle, target motion posture and other characteristics, set the minimum number of azimuth resolution units and the maximum number of azimuth resolution units occupied by the video signal reflected by the target in azimuth angle. For example, if the system's minimum azimuth size resolution unit is α°, and the actual shape azimuth of the target is β°, according to the characteristics of the video signal reflected by the actual target, it may change due to factors such as radiation angle and target posture, resulting in the video signal reflected by the target being γ° in azimuth angle. With α° as the minimum azimuth size resolution unit, set the number of α° occupied by γ°, and the minimum number γ°. min and the maximum number γ max ;

[0076] Step 43: According to the characteristics of the video signal reflected by the actual target, combined with the system's minimum distance resolution unit, target type, actual size of each type of target, radiation angle, target motion posture and other characteristics, set the minimum number of distance resolution units and the maximum number of distance resolution units occupied by the video signal reflected by the target on the minimum azimuth resolution unit. For example, if the system's minimum azimuth resolution unit is α°, the target is actually an irregular shape, and the distance dimensions at various azimuth angles are not the same, and may change due to factors such as radiation angle and target posture, resulting in different distance resolution units occupied by the video signal reflected by the target at different azimuth angles. Taking α° as the minimum azimuth resolution unit, set the distance dimension occupied by the target on α° to D meters, and still take X meters as the minimum distance resolution unit, set the number of D meters occupying X meters, and the minimum number D min and the maximum number D max ;

[0077] Step 44: Set the target distance size range Z min ~Z max , target azimuth angle range γ min ~γ max, the distance dimension range D of the target azimuth angle min ~D max , you can create a distance\azimuth dimension frame.

[0078] In step 5, the multi-dimensional feature information includes: distance, azimuth, amplitude, total number of sampling points, number of sampling points of the signal within the amplitude threshold frame, number of sampling points of the signal within the distance\azimuth size frame, number of sampling points of the signal within the distance\azimuth size frame and within the amplitude threshold frame, etc. The specific quantization process is:

[0079] Step 51: Sample and quantize the video signal after two filtering according to the sampling rate set by the system, extract its distance feature information, and quantize it using the system's minimum distance size resolution unit X meters, quantized to R i ;

[0080] Step 52: Sample and quantize the video signal after two filtrations according to the sampling rate set by the system, extract its azimuth feature information, and quantize it using the system's minimum azimuth size resolution unit α°, quantized to A i ;

[0081] Step 53: Sample and quantize the video signal after two filtrations according to the sampling rate set by the system, extract its amplitude feature information, and perform statistics according to the amplitude threshold frame based on the amplitude feature set in step 3. The total number of sampling points is Si, and the number of sampling points located in the amplitude threshold frame A is 1. min ~A max The number of sampling points within the range is W i , the proportion of sampling points within the amplitude threshold frame to the total number of sampling points (W / S) i ;

[0082] Step 54: Sample and quantize the video signal input after two filtering according to the sampling rate set by the system, extract its distance and azimuth feature information, and perform statistics according to the distance\azimuth size frame based on the distance and azimuth features set in step 4. The number of sampling points of the signal within the distance\azimuth size frame is RA i , the ratio of the number of sampling points within the range\azimuth dimension box to the total number of sampling points (RA / S) i ;

[0083] Step 55: Sample and quantize the video signal input after two filtering according to the sampling rate set by the system, extract its distance, azimuth, and amplitude feature information, and perform statistics according to the amplitude threshold frame based on the amplitude feature set in step 3 and the distance\azimuth size frame based on the distance and azimuth features set in step 4. The sampling points of the signal located in the distance\azimuth size frame and in the amplitude threshold frame are RAW. i, the ratio of the number of sampling points within the range\azimuth size frame and within the amplitude threshold frame to the total number of sampling points (RAW / S) i .

[0084] In view of the fact that the extraction and determination of point traces in the prior art only rely on whether they fall into the association gate and the amplitude of the point traces, the establishment of the association gate is usually based on target features such as distance and azimuth, and the target characteristics based on them are very limited. There is a high probability that false point traces cannot be filtered out, resulting in the problem that false point traces are determined as valid point traces. In this embodiment, multi-dimensional target characteristics are added through steps 3-5 to reduce the probability of misjudgment of point traces. Step 3 sets an amplitude threshold frame based on amplitude characteristics, excludes marks such as slight clutter with small reflections and buildings with very strong reflections from the amplitude threshold frame, and only selects and marks the data that meets the requirements of the amplitude threshold frame to enter the subsequent point trace determination calculation; Step 4 sets a distance and azimuth size frame based on distance and azimuth characteristics, and especially adds the target feature of the distance size occupied on the minimum azimuth unit, which provides a further reference target feature for the subsequent preliminary judgment; Step 5 samples and quantizes the extraction of multi-dimensional target features, which provides more comprehensive and multi-dimensional data support for subsequent judgments, and can set filtering, identification and judgment conditions according to multiple dimensions.

[0085] The specific process in step 6 is: according to the distance feature information R obtained in steps 51 and 52 i , orientation feature information A i , combined with the distance\azimuth size frame obtained in step 4, the judgment logic is: the distance size of the input video signal is within the distance size range, and the azimuth feature of the input video signal is greater than or equal to the minimum azimuth angle range, and the distance size of the input video signal per unit angle is greater than or equal to the distance size at the minimum azimuth angle, and the azimuth size of the input video signal is less than or equal to the maximum azimuth angle range or the distance size per unit angle is less than or equal to the distance size at the maximum azimuth angle, that is, (R i ≥Z min and R i ≤Z max )and(A i ≥γ min and on α°A i ≥D min )and(A i ≤γ max or on α°A i ≤D max ), it is initially determined as a point trace with quality, otherwise it is determined as a point trace without quality. It also supports the synchronous output of point traces with quality and point traces without quality.

[0086] Step 7 specifically includes: combining the target type, target size, target radiation angle, target motion posture, etc. in the actual application scenario, statistically analyzing the target feature information output by the mass point traces and the massless point traces determined in step 6 when the real target exists, and statistically forming the multi-dimensional feature information of each type of target corresponding to the high probability (such as greater than 90%) of determining that the target point trace is mass, to form a multi-dimensional feature information set, including: the distance size range Zˊ of the target min ~Zˊ max 、Azimuth angle range γˊ min ~γˊ max , Amplitude range Aˊ min ~Aˊ max , the minimum number of sampling points W within the amplitude threshold frame min , the ratio of the minimum number of sampling points within the amplitude threshold frame to the total number of sampling points (W / S) min , the ratio of the minimum number of sampling points in the distance\azimuth size box to the total number of sampling points (RA / S) min , the ratio of the minimum number of sampling points within the range\azimuth size frame and the amplitude threshold frame to the total number of sampling points (RAW / S) min The multi-dimensional target feature information obtained based on actual use cases and test statistics is more real, reliable, effective, and highly verifiable. Its reliability and credibility are greatly improved, which greatly increases the probability of extracting valid points and reduces the probability of misjudgment of false points. It is particularly suitable for false targets caused by multipath effects that meet the distance, azimuth, and amplitude characteristics, and the existing methods cannot achieve correct judgment.

[0087] In step 8, the logic of setting point trace extraction and determination is adjusted by the obtained multi-dimensional feature information set, specifically:

[0088] The target distance size range Zˊ is determined to be mass based on a high statistical probability (e.g. greater than 90%) min ~Zˊ max 、Azimuth angle range γˊ min ~γˊ max , adjust the distance\azimuth size box created in step 4;

[0089] When the target point is judged to have mass according to a high statistical probability (such as greater than 90%), the target amplitude range Aˊ min ~Aˊ max , adjust the amplitude threshold frame established in step 3;

[0090] When the target point is judged to have mass according to a high probability (such as greater than 90%), the minimum number of sampling points W located within the amplitude threshold frame is min , adjust and set the minimum sampling point threshold value W within the amplitude threshold framef , which is used as the key parameter for point extraction and judgment;

[0091] When the target point is judged to have mass based on a high probability (such as greater than 90%), the proportion of the minimum number of sampling points within the amplitude threshold frame to the total number of sampling points (W / S) min , adjust the minimum number of sampling points within the amplitude threshold frame to the total number of sampling points (W / S) f , which is used as the key parameter for point extraction and judgment;

[0092] When the target point is judged to have mass according to a high probability (such as greater than 90%), the proportion of the minimum number of sampling points in the range\azimuth size box to the total number of sampling points (RA / S) min , adjust the minimum number of sampling points in the distance\azimuth size box to the total number of sampling points (RA / S) f , which is used as the key parameter for point extraction and judgment;

[0093] When the target point is judged to have mass based on a high probability (e.g. greater than 90%), the proportion of the minimum number of sampling points within the range\azimuth size frame and within the amplitude threshold frame to the total number of sampling points (RAW / S) min , adjust the minimum number of sampling points in the range\azimuth size box and in the amplitude threshold box to the total number of sampling points (RAW / S) f , which is used as the key parameter for point extraction and judgment;

[0094] If the points of most of the real targets (more than 90%) are larger than the designated size range, the size frame can be appropriately enlarged; if the points of most of the real targets (more than 90%) are smaller than the designated size range, the size frame can be appropriately reduced; whether the size frame is reasonable is determined by whether it can cover more than 90% of the targets and can be quantified.

[0095] The logic of point extraction and judgment is established by combining the adjusted amplitude threshold frame, distance\azimuth frame and related key parameters.

[0096] Step 9 directly compares the input traces with the logic one by one. The input traces that meet all the conditions are determined to be valid traces and their feature information is extracted. Otherwise, they are determined to be invalid traces.

[0097] Step 10 is to obtain the judgment result of step 9, and add a credibility characteristic mark based on the multi-dimensional target characteristics. For example, for the valid point traces judged in step 8, the credibility mark is set to 1; for the invalid point traces judged in step 8, the credibility mark is set to 0 and output to the back-end reference.

[0098] When false traces caused by clutter, noise, multipath interference, etc. in the prior art fall into the association gate, if there are multiple traces in the association gate, it may lead to misjudgment of the traces, especially when the false traces caused by the multipath effect are close to the real traces, there is a high probability of misjudgment of the traces; when the amplitude of the false traces is large, the probability of misjudgment of the traces is high, which is solved by steps 6 to 10.

[0099] Usually, the traditional method has no obvious distinction in distance, direction, and amplitude characteristics, and the unrecognizable false points will appear in one or more target features in other dimensions. For example, the number of sampling points W located within the amplitude threshold frame i , The proportion of sampling points within the amplitude threshold frame to the total number of sampling points (W / S) i , the ratio of the number of sampling points within the distance\azimuth dimension box to the total number of sampling points (RA / S) i , the ratio of the number of sampling points within the range\azimuth size frame and within the amplitude threshold frame to the total number of sampling points (RAW / S) i Even if both the false traces and the real traces fall into the associated gate, that is, there are multiple traces in the associated gate, the false traces are close to the real traces, or even the false traces have a large amplitude, the optimization can be directly performed according to the judgment logic of steps 7 and 8 and the credibility characteristics of the trace data output by steps 9 and 10.

[0100] Since step 7 is based on multi-dimensional target feature information obtained from actual use cases and test statistics, it is more real, reliable, effective, and has strong verifiability. Its reliability and credibility are greatly improved, which greatly increases the probability of extracting effective points and reduces the probability of misjudgment of false points. It is particularly suitable for false targets caused by multipath effects, which meet the distance, azimuth, and amplitude characteristics, but the existing technology cannot achieve correct judgment.

[0101] The present invention extracts and determines the input data before the point navigation association through operations such as filtering, threshold measurement, collection quantification, preliminary judgment, actual measurement statistics, feedback adjustment, logical judgment, and credibility identification of multi-dimensional target features, and retains both the quality point traces and the qualityless point traces and outputs them to the back end, and marks the point traces with credibility, which not only retains the output of the point traces to the maximum extent, but also provides more reference target characteristics for the point navigation association, making the point navigation association more reliable, and reducing the probability of false point traces and error correlation. In addition, step 6 marks and outputs both the quality point traces and the qualityless point traces, without deleting the filter, so as to facilitate step 7 to perform statistical comparison according to the actual application scenario and target feature information: if some point traces marked as qualityless are actually real targets; or if some point traces marked as quality are actually false targets, the amplitude threshold frame set in step 3 and the distance azimuth size frame set in step 4 can be adjusted to further improve the probability of correct point trace extraction and reduce the probability of false point trace misjudgment. In particular, some faint targets are easily marked as massless points by the amplitude threshold frame. Through feedback adjustment, they can be marked as mass points first, and then their multi-dimensional target characteristics can be extracted. In actual scenes, we focus on finding target characteristics that can be clearly distinguished from clutter and extract them. This processing mechanism of continuous feedback adjustment can further improve the probability of correct point extraction and judgment.

[0102] The present invention is not limited to the above-mentioned specific embodiments. The present invention extends to any new features or any new combination disclosed in this specification, as well as any new method or process steps or any new combination disclosed. If non-substantial changes or improvements made by those skilled in the art without departing from the spirit of the present invention should fall within the scope of protection of the claims of the present invention.

[0103] All features disclosed in this specification, or steps in all methods or processes disclosed, except mutually exclusive features and / or steps, can be combined in any manner.

[0104] Any feature disclosed in this specification, unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. A point trace extraction and judgment method based on multi-dimensional target features, characterized in that: The following steps are involved: Step 1: According to the distance range and azimuth range that the radar system needs to pay attention to, the input video signal is filtered once to filter out the signals outside the range; Step 2, performing a second filtering on the video signal after the first filtering according to the number of times the video signal is received and accumulated; Step 3: According to the input video signal amplitude and in combination with the system processing model, an amplitude threshold frame based on the amplitude feature is set; Step 4: According to the characteristics of the video signal reflected by the actual target, set the distance\azimuth size frame based on the distance and azimuth characteristics; Step 5: Sampling and quantizing the video signal after secondary filtering to extract its multi-dimensional feature information; Step 6: Based on the distance and range feature information of the video signal and the distance\azimuth size frame, the traces are divided into quality traces and non-quality traces, and marked; Step 7, combining the video signals of multiple known real targets, executing the process of step 1 to step 6, obtaining the marked traces, and counting the multi-dimensional feature information corresponding to the real target when the proportion of the traces with quality is greater than 90%, to form a multi-dimensional feature information set; Step 8: Based on the multi-dimensional feature information set obtained in step 7, the amplitude threshold frame set in step 3 and the distance / azimuth size set in step 4 are fed back and updated to obtain the logic of point extraction and determination; Step 9: Compare the sampled and quantized input trace data with the logic of trace extraction and determination, and take the trace data that meets the logic of trace extraction and determination as the extraction and determination result; Step 10: According to the results of extraction and determination, a credibility characteristic mark is added to the point trace data and output to the back-end reference.

2. The point trace extraction and judgment method based on multi-dimensional target features according to claim 1 is characterized in that: The step 2 includes the following sub-steps: Step 21, according to the beam dwell time and the system transmission and reception signal cycle, calculate the number of video signals that should be received at the corresponding position of the target when the target exists during the beam dwell time, which is M; Step 22: according to the system processing tolerance requirement, set the minimum number of video signals to be received, N, N≤M; Step 23: Perform reception accumulation statistics within the beam dwell time for the input video signal after the first filtering, and the number of times the target actually receives the video signal at the corresponding position is L; Step 24, compare the number of times L of actually received video signals with the set number of times N of video signals that should be received at least, if L≥N, retain the input video signal received this time, if L<N, filter the input video signal.

3. The point trace extraction and judgment method based on multi-dimensional target features according to claim 1 or 2, characterized in that: The step 4 includes the following sub-steps: Step 41: According to the characteristics of the video signal reflected by the actual target, combined with the system's minimum distance resolution unit and minimum azimuth resolution unit, set the minimum distance resolution unit number Z occupied by the video signal reflected by the target in the distance dimension. min and the maximum number of distance resolution units Z max , set the minimum number of azimuth resolution units and γ occupied by the video signal reflected by the target in azimuth angle min Maximum number of azimuth resolution units γ max , set the minimum number of distance resolution units occupied by the video signal reflected by the target on the minimum azimuth size resolution unit D min and the maximum number of distance resolution units D max ; Step 42: According to the set target distance size range Z min ~Z max , target azimuth angle range γ min ~γ max , the distance dimension range D of the target azimuth angle min ~D max , create a distance\azimuth dimension box.

4. The point trace extraction and judgment method based on multi-dimensional target features according to claim 3 is characterized in that: The multidimensional feature information in step 5 includes: distance feature information, azimuth feature information, and the number of sampling points within the amplitude threshold frame, the proportion of the sampling points within the amplitude threshold frame in the total number of sampling points, the number of sampling points of the signal within the distance\azimuth size frame, the proportion of the sampling points within the distance\azimuth size frame in the total number of sampling points, the number of sampling points of the signal within the distance\azimuth size frame and within the amplitude threshold frame, and the proportion of the sampling points within the distance\azimuth size frame and within the amplitude threshold frame to the total number of sampling points.

5. The point trace extraction and judgment method based on multi-dimensional target features according to claim 4 is characterized in that: In step 6, the judgment method is: when the distance dimension of the input video signal is within the distance dimension range, and the azimuth feature of the input video signal is greater than or equal to the minimum azimuth angle range, and the distance dimension of the input video signal at the azimuth angle is greater than or equal to the distance dimension at the minimum azimuth angle, and the azimuth dimension of the input video signal is less than or equal to the maximum azimuth angle range or the distance dimension per unit angle is less than or equal to the distance dimension at the maximum azimuth angle, it is judged as a point trace with quality, otherwise it is judged as a point trace without quality.

6. The point trace extraction and judgment method based on multi-dimensional target features according to claim 5 is characterized in that: In step 7, the multidimensional feature information set is composed of multidimensional feature information of multiple targets, including: the distance size range, azimuth angle range, amplitude range, the minimum number of sampling points within the amplitude threshold frame, the proportion of the minimum number of sampling points within the amplitude threshold frame to the total number of sampling points, the proportion of the minimum number of sampling points within the distance\azimuth size frame to the total number of sampling points, and the proportion of the minimum number of sampling points within the distance\azimuth size frame and within the amplitude threshold frame to the total number of sampling points.

7. The point trace extraction and judgment method based on multi-dimensional target features according to claim 6 is characterized in that: In step 8, the logic acquisition method for point trace extraction and determination is: The distance\azimuth size frame is adjusted according to the distance size range and azimuth angle range of the target; the amplitude threshold frame is adjusted according to the amplitude range of the target; and the minimum number of sampling points in the amplitude threshold frame, the proportion of the minimum number of sampling points in the amplitude threshold frame to the total number of sampling points, the proportion of the minimum number of sampling points in the distance\azimuth size frame to the total number of sampling points, and the proportion of the minimum number of sampling points in the distance\azimuth size frame and in the amplitude threshold frame to the total number of sampling points are used as judgment parameters; the adjusted distance\azimuth size frame, amplitude threshold frame and judgment parameters are combined to form the logic of point trace extraction and judgment.

8. The point trace extraction and judgment method based on multi-dimensional target features according to claim 1 is characterized in that: In step 8, the input point trace and the point trace extraction are compared with all conditions in the judgment logic. If all conditions are met, it is judged as a valid point trace and feature information is extracted; Otherwise it is judged as invalid point trace.

Images