A Method for Sea Clutter Suppression Processing and Performance Analysis Based on Video Signals

Abstract

The present invention provides a method for sea clutter suppression processing and performance analysis based on video signals. The processing flow designed by the present invention is as follows: First, perform high false alarm and high detection probability detection on the preprocessed radar echo, then conduct multi-channel detection through inter-frame correlation detection, SIC inter-frame accumulation detection, and traditional CFAR detection, and finally perform combined logic judgment on the detection results of multiple channels according to the sea clutter area information. This processing flow can effectively suppress sea clutter through inter-frame accumulation, which is beneficial to improving the detection performance of stationary or slow targets between frames, and ensures the detection performance of moving targets through multi-channel detection.

Description

Technical Field

[0001] The present invention relates to a signal processing and performance analysis method, in particular to a sea clutter suppression processing and performance analysis method based on video signals. Background Art

[0002] For sea-warning radars, near-shore surveillance radars, and navigation radars that focus on detecting maritime targets / low-altitude targets, in complex weather and high sea states, sea clutter exhibits non-Gaussian, non-linear, and non-stationary characteristics. Sea spikes appear frequently, and the amplitude tailing phenomenon of sea clutter is serious, resulting in many problems such as false alarms of sea clutter, false alarms of sea spikes, and missed detections of targets in clutter. Sea clutter suppression has always been a difficult problem in the field of sea radar detection. For coherent radars, moving target indication (MTI) and moving target detection (MTD) are used to suppress sea clutter, but this will also cause detection losses for stationary targets and ambiguous velocity targets. For most existing non-coherent radars, the existing target detection technologies use methods such as non-coherent integration of single-frame data and CFAR. However, due to the strong correlation of sea clutter between sweeps, sea clutter cannot be effectively suppressed, and due to the mismatch of the CFAR detection model in sea clutter, the detection performance is lost.

[0003] According to the different characteristics of sea clutter and target echoes between frames, sea clutter fully meets the condition of being uncorrelated between frames. Sea clutter may be very strong in the current frame data and very weak in the next frame data. For target echoes, the echoes between frames are still strongly correlated. Using this characteristic, through region correlation accumulation between frames, sea clutter will be suppressed and the target detection performance will be improved. This method can be called the inter-frame correlation accumulation algorithm, abbreviated as SIC (scan-to-scan-integration-correlation). Scholars at home and abroad have carried out research on this algorithm, but no complete processing flow has been given for a certain type of radar. In particular, the performance of the algorithm has not been theoretically verified, and only the effectiveness of the algorithm has been verified from the effect level. Summary of the Invention

[0004] Object of the Invention: The technical problem to be solved by the present invention is to provide a sea clutter suppression processing and performance analysis method based on video signals in view of the deficiencies of the prior art.

[0005] To solve the above technical problem, the present invention discloses a sea clutter suppression processing and performance analysis method based on video signals, including the following steps:

[0006] Step 1, preprocess the original radar video data to obtain the preprocessed video data;

[0007] Step 2, perform constant false alarm rate detection or slow threshold detection on the preprocessed video data to obtain the preliminarily detected video data;

[0008] Step 3: Perform multi-channel detection processing and performance analysis on the output video data in Step 2, i.e., the video data after preliminary detection;

[0009] Step 4: Perform combinational logic judgment and actual effect analysis on the multi-channel detection video output in Step 3.

[0010] The preprocessing described in Step 1 includes anti-synchronous interference of the same frequency and non-coherent accumulation;

[0011] The original radar video data is a frame of original video data obtained by the radar antenna scanning one circle. Assume the k-th frame of original radar video data X k , which is represented by a matrix as:

[0012] X k ={x k (i,j)}, i = 1, 2,..., M, j = 1, 2,..., N; k = 1, 2, 3,...

[0013] where i represents azimuth sampling, j is range sampling, M is the maximum value of azimuth sampling, N is the maximum value of range sampling, and x k (i,j) is the original video measurement value recorded by the detection unit (i,j) at the k-th frame moment;

[0014] The video data z after preprocessing the original radar video data k is represented as:

[0015] Z k ={z k (i,j)}, i = 1, 2,..., M, j = 1, 2,..., N; k = 1, 2, 3,...

[0016] where z k (i,j) is the video data after preprocessing x k (i,j), representing the preprocessed video value of the detection unit (i,j) at the k-th frame moment.

[0017] Z k ={z k (i,j)}, i = 1, 2,..., M, j = 1, 2,..., N; k = 1, 2, 3,... When performing constant false alarm rate detection or slow threshold detection, set a high false alarm rate threshold, that is, set the false alarm rate to 10 -5 ~10 -2 ;

[0018] The video data Y after preliminary detection k is represented as:

[0019] Y k ={y k(i, j)}, i = 1, 2, ..., M, j = 1, 2, ..., n; k = 1, 2, 3, ...

[0020] Among them, y k (i, j) represents z k The video data of z(i, j) after constant false alarm detection.

[0021] The multi-channels described in step 3 include: the first channel, i.e., correlation detection; the second channel, i.e., inter-frame accumulation detection; the third channel, i.e., moving target detection compensation;

[0022] The specific method for performing multi-channel detection processing and performance analysis includes:

[0023] Step 3-1: Perform the first channel detection processing flow, i.e., correlation detection, including constant false alarm detection and within-window correlation detection, and analyze the performance of the correlation detection to obtain the detection video of the first channel;

[0024] Step 3-2: Perform the second channel detection processing flow, i.e., inter-frame accumulation detection, including inter-frame accumulation and constant false alarm detection, analyze the performance of the inter-frame accumulation detection to obtain the detection video of the second channel;

[0025] Step 3-3: Perform the third channel detection processing flow, i.e., moving target detection compensation, to obtain the detection video of the third channel.

[0026] The correlation detection described in step 3-1 is as follows:

[0027] Assume that the data after constant false alarm detection in the first channel detection processing flow of the (k - 1)-th frame is denoted as D k-1 , and the data after constant false alarm detection in the first channel detection processing flow of the k-th frame is denoted as D k , and the output after within-window correlation detection is denoted as H1 k , and the processing logic formula is:

[0028]

[0029] That is:

[0030]

[0031] In the formula, H1 k (i, j) represents the output video data of the detection unit (i, j) after correlation detection, represents that at least one detection unit within the correlation window centered on (i, j) in the (k - 1)-th frame detects a target, represents that no target is detected;

[0032] The analysis of the performance of the correlation detection, i.e., the analysis of the detection probability under a set false alarm rate, is as follows:

[0033] In a noise background or a sea clutter background, assuming that the false alarm rate of single-frame detection is Pfa0 and the false alarm rate after association detection is Pfa′, then:

[0034]

[0035] Then the false alarm rate Pfa′ after association detection is:

[0036] Pfa′ = Pfa0 * Pfa0 * L

[0037] Where L represents the size of the association window; analyze the detection probability of the target, that is, fix the false alarm probability, analyze the relationship curve between the detection probability and the signal-to-noise ratio, and the detection logic of the target is as follows:

[0038]

[0039] Assuming that the single-frame target detection probability is Pd and the detection probability after association detection is Pd′, then:

[0040] Pd′ = Pd * Pd

[0041] After the association detection of the first channel, the detection performance of the target is the detection probability Pd′.

[0042] The inter-frame integration detection described in step 3-2 is as follows:

[0043] Assume that the video data after preliminary detection of the kth frame is Y k (k = 1, 2,...), and the video data after inter-frame integration is represented as SI k (k = 1, 2,...), then the inter-frame integration formula is:

[0044] SI k = aY k +(1 - a)SI k-1

[0045] After constant false alarm detection, the detection video H2 is output k (k = 1, 2,...), in the CA-CFAR processing, set the slow threshold aV T , where V T is the threshold of CA-CFAR in the preprocessing, and a represents the inter-frame integration coefficient;

[0046] The performance of the analyzed inter-frame integration detection, that is, using the Monte Carlo statistical method to statistically analyze the performance of the inter-frame integration detection, the specific method is as follows:

[0047] First, analyze the change of the false alarm probability in the Rayleigh noise background area:

[0048] Assume the radar raw video data Xk The noise is Rayleigh distributed. Set the number of non-coherent integrations and the false alarm rate of the preprocessing, and analyze Y k (k = 1, 2,...) for the false alarm rate of single-frame CFAR detection and inter-frame integration CFAR detection;

[0049] Secondly, analyze the detection probabilities of stationary targets and moving targets, and analyze them separately for stationary targets and moving targets:

[0050] Inter-frame integration of stationary targets is equivalent to non-coherent integration between frames. The benefit of non-coherent integration depends on the background clutter distribution and target distribution characteristics; for moving targets across range or azimuth cells, after inter-frame integration, the target amplitude has a loss of α times, and there are residual tails at the historical positions; design aV for the false alarms of the said residual tails T to eliminate them with a slow threshold;

[0051] For the detection loss of moving targets, use the Monte Carlo statistical method: Assume that the original radar video data X k has a Rayleigh distributed clutter background, the target is a constant amplitude signal, the target completely spans range and azimuth cells between frames, set the number of non-coherent integrations and the inter-frame integration coefficient a, collect background samples after inter-frame integration, and obtain the detection thresholds at different false alarm rates; collect the target sample signals after inter-frame integration, and statistically analyze the detection probabilities at different false alarm rates;

[0052] Analyze the suppression of sea clutter false alarms: Based on the non-correlation characteristics of sea clutter between frames, generate k-distributed clutter. After the preprocessing and inter-frame integration detection described in step 1, collect samples, and statistically analyze the false alarm rates of single-frame CFAR detection and inter-frame integration CFAR detection by Monte Carlo.

[0053] The moving target detection compensation described in step 3-3, that is, the detection processing flow of the third channel, is used to make up for the loss of moving targets in the detection processing flow of the second channel. The specific method is as follows:

[0054] Assume that the video data after preliminary detection of the k-th frame is Y k (k = 1, 2,...), and use traditional CA-CFAR and threshold detection to output H3 k (k = 1, 2,...).

[0055] The combined logic judgment described in step 4, that is, AND the detection videos of the first channel and the second channel, and then OR with the detection video of the third channel. The specific method is as follows:

[0056] Assume that the final detection video is H k (k = 1, 2,...), then the combined logic is expressed as the following formula:

[0057] H k (i,j)=(H1k (i,j)&H2 k (i,j))||H3 k (i,j)

[0058] Among them, H1 k (i,j) represents the detection video of the first channel; H2 k (i,j) represents the detection video of the second channel; H3 k (i,j) represents the detection video of the third channel.

[0059] For the actual effect analysis described in step 4, the specific method is: use the original radar video data collected in reality to verify the sea clutter suppression effect and the target detection effect.

[0060] Beneficial effects:

[0061] (1) Solve many problems such as sea clutter false alarms, sea spike false alarms, and target missed alarms in clutter in marine radars;

[0062] (2) The present invention gives a complete processing flow for suppressing sea clutter using inter-frame features. In particular, a theoretical analysis of the performance of the algorithm is carried out, and the effectiveness of the algorithm is verified from the perspective of the effect. Description of the Drawings

[0063] The following further specifically describes the present invention in conjunction with the drawings and specific embodiments, and the above and / or other advantages of the present invention will become clearer.

[0064] Figure 1 It is a schematic diagram of the processing flow in the method of the present invention.

[0065] Figure 2 It is a schematic diagram of the first channel, i.e., the correlation detection process, in the method of the present invention.

[0066] Figure 3 It is a schematic diagram of the correlation window in the method of the present invention.

[0067] Figure 4 It is a schematic diagram of the false alarm rate after correlation detection of the first channel, i.e., in the method of the present invention.

[0068] Figure 5 It is a schematic diagram of the target detection performance of the first channel, i.e., correlation detection, in the method of the present invention.

[0069] Figure 6 It is a schematic diagram of the detection processing flow of the second channel in the method of the present invention.

[0070] Figure 7 It is a schematic diagram of the analysis of the false alarm rate of inter-frame accumulation of the second channel in the method of the present invention.

[0071] Figure 8It is the frame accumulation P display intention of the moving target in the method of the present invention.

[0072] Figure 9 It is the schematic diagram of the echo amplitude of the moving target frame accumulation in the method of the present invention.

[0073] Figure 10 It is the schematic diagram of the performance loss of fast target detection in the method of the present invention.

[0074] Figure 11 It is the schematic diagram of the sea clutter false alarm suppression effect in the method of the present invention.

[0075] Figure 12 It is the schematic diagram of the analysis result of the sea clutter false alarm rate in the method of the present invention.

[0076] Figure 13 It is the schematic diagram of the detection and processing flow of the third channel in the method of the present invention.

[0077] Figure 14 It is the schematic diagram of the sea clutter suppression effect in the method of the present invention. Specific implementation mode

[0078] Sea clutter can be effectively suppressed through frame accumulation, which is beneficial to improving the detection performance of stationary or slow targets between frames. However, the loss of moving targets is a problem brought by this algorithm. The processing flow designed by the present invention is as follows: First, perform high false alarm and high detection probability detection on the preprocessed radar echo, then perform multi-channel detection through frame correlation detection, SIC frame accumulation detection, and traditional CFAR detection, and finally perform combined logic judgment on the detection results of multiple channels according to the sea clutter area information.

[0079] The present invention discloses a method for sea clutter suppression processing and performance analysis based on video signals, including the following steps:

[0080] Step 1, perform preprocessing such as anti-synchronous interference of the same frequency and non-coherent accumulation on the original radar video data;

[0081] Step 2, perform constant false alarm detection with high false alarm and high detection probability on the preprocessed video data;

[0082] Step 3, perform multi-channel detection processing and performance analysis on the output video data of Step 2;

[0083] Step 4, perform combined logic judgment and actual effect analysis on the multi-channel detection video output in Step 3.

[0084] The sea clutter suppression processing (as Figure 1 shown) and the performance analysis method in this embodiment Figure 1 is the processing flow schematic diagram of the method of the present invention, and the said Step 1 includes:

[0085] Further, in one implementation, step 1 includes:

[0086] The radar raw video data obtained by the radar antenna scanning one circle is called one frame of raw video data. Assume that the k-th frame of radar raw video data is represented by the matrix X k ={x k (i,j)}, i = 1, 2, ..., M, j = 1, 2, ..., N; k = 1, 2, 3, .... Among them, i represents azimuth sampling, j is range sampling, M is the maximum value of azimuth sampling, N is the maximum value of range sampling, and x k (i,j) is the measurement value recorded by the detection unit (i,j) at the k-th frame moment.

[0087] The data after preprocessing the radar raw video data, such as anti-co-channel asynchronous interference and non-coherent integration, is represented as Z k ={z k (i,j)}, i = 1, 2, ..., M, j = 1, 2, ..., N; k = 1, 2, 3, .... Among them, anti-co-channel asynchronous interference and non-coherent processing are conventional processing algorithms in the field of radar detection. For specific references, please refer to "Fundamentals of Radar Signal Processing" by Xing Mengdao, Electronic Industry Press, and "Radar Handbook (Third Edition)" by Merrill I. Skolnik.

[0088] Further, in one implementation, step 2 includes:

[0089] Perform constant false alarm rate detection (CA-CFAR) with high false alarm and high detection probability or slow threshold detection on the preprocessed video data Z k ={z k (i,j)}, i = 1, 2, ..., M, j = 1, 2, ..., N; k = 1, 2, 3, .... The processed video data is represented as Y k ={y k (i,j)}, i = 1, 2, ..., M, j = 1, 2, ..., N; k = 1, 2, 3, .... CA-CFAR and slow threshold detection are conventional processing algorithms in the field of radar detection. For specific references, please refer to "Fundamentals of Radar Signal Processing" by Xing Mengdao, Electronic Industry Press, and "Radar Handbook (Third Edition)" by Merrill I. Skolnik. It should be noted that the high false alarm rate threshold setting of CA-CFAR or slow threshold detection is crucial to ensure a high detection probability and high false alarm rate of the target. For example, set the false alarm rate to 10 -3 or 10 -2 , so as to ensure a high detection probability of the target.

[0090] Further, in one implementation, step 3 includes:

[0091] Perform multi-channel detection processing on the output video data of step 2 to greatly suppress false alarms, especially sea clutter false alarms, while ensuring the minimum loss of the target.

[0092] Step 3-1:

[0093] The detection processing flow of channel 1 is as Figure 2 shown, which can also be called correlation detection. The detailed implementation process is as follows.

[0094] Assume that the detected data after the (k-1)-th frame is denoted as D k-1 , and the detected data after the k-th frame is denoted as D k . The output after in-window correlation detection is denoted as H1 k . The processing logic formula is as follows.

[0095]

[0096] That is:

[0097]

[0098] In the formula means that at least one detection unit in the correlation window centered on (i, j) in the (k-1)-th frame detects a target. The schematic diagram of the correlation window is as Figure 3 shown.

[0099] Furthermore, analyze the performance of correlation detection, that is, analyze the detection probability under a certain false alarm rate.

[0100] In a noise background or a sea clutter background, assume that the single-frame detection false alarm rate is Pfa0, and the false alarm rate after correlation detection is Pfa'. Then

[0101]

[0102] Then the false alarm rate after correlation detection is

[0103] Pfa′ = Pfa0 * Pfa0 * L

[0104] Figure 4 is the false alarm rate improvement diagram before and after correlation detection. From the simulation results, it can be seen that the false alarm rate has been greatly suppressed. For example, when the false alarm rate before correlation detection in the figure is 10 -4 , and the size of the correlation window is 100, the false alarm rate is improved to 10 -6 after correlation detection.

[0105] Furthermore, analyze the detection probability of the target, that is, fix the false alarm probability and analyze the relationship curve between the detection probability and the signal-to-noise ratio. The detection logic of the target is as follows.

[0106]

[0107] Assume that the single-frame target detection probability is Pd, and the detection probability after associated detection is Pd'. Then

[0108] Pd′ = Pd * Pd

[0109] Assume that the movement range of the target within one scanning period is within one associated window, and the final detection false alarm rate is set to 10 -6 , and analyze the detection performance of the target. Then in the associated detection process, the number of non-coherent integrations is 20, and the false alarm rate of high false alarm detection is set to 10 -4 , the associated window is set to L = 10 * 10 = 100, and the false alarm rate after associated detection is 10 -6 , analyze the target detection probability of associated detection and compare it with the performance of traditional constant false alarm detection. The analysis results are as Figure 5 shown. The associated detection performance curve is to the left of the traditional detection performance curve, indicating that the overall performance of associated detection is improved.

[0110] After the associated detection through Channel 1, the sea clutter false alarms are effectively suppressed, and the detection performance of the target is also improved.

[0111] Step 3-2:

[0112] The detection processing flow of Channel 2 is as Figure 6 shown, which can also be called inter-frame integration detection. The specific implementation process is as follows.

[0113] Assume that the video data of the k-th frame of the original video after preprocessing is represented as Y k (k = 1, 2,...), and the video data after inter-frame integration is represented as SI k (k = 1, 2,...). Then the inter-frame integration formula is

[0114] SI k = aY k +(1 - a)SI k-1

[0115] After constant false alarm detection, the detection video H2 k (k = 1, 2,...) is output. In CA-CFAR processing, a slow threshold aV T is set to eliminate the tail false alarms left by the inter-frame integration of moving targets, where V T is the threshold of CA-CFAR in preprocessing.

[0116] Further analyze the detection performance of the inter-frame integration detection of Channel 2.

[0117] The distribution characteristics of the original video data after preprocessing and inter-frame integration can no longer be fitted by a single distribution. Here, the Monte Carlo statistical method is used. For specific references, please refer to "Radar System Modeling and Simulation" by Yang Wanhai, published by Xidian University Press. The detection performance of inter-frame integration is statistically analyzed. First, the change in the false alarm probability in the Rayleigh noise background area is analyzed.

[0118] Assume that the noise of the original video follows a Rayleigh distribution, the number of non-coherent integrations is 20, and the false alarm rate of preprocessing is 10 -4 , analyze Y k (k = 1, 2,...) The false alarm rates of single-frame constant false alarm detection and inter-frame integration constant false alarm detection are statistically analyzed. The statistical results are as Figure 7 shown. It can be seen from the figure that the false alarm rate of inter-frame integration for noise is the same as that of single-frame constant false alarm detection.

[0119] Secondly, further analyze the detection probabilities of stationary targets and moving targets. Since there is a loss for moving targets in inter-frame integration, here the stationary targets and moving targets are analyzed separately.

[0120] Inter-frame integration of stationary targets is equivalent to non-coherent integration between frames. The benefit of non-coherent integration depends on the background clutter distribution and target distribution characteristics. For detailed analysis, please refer to relevant books such as "Fundamentals of Radar Signal Processing" by Xing Mengdao, published by Electronic Industry Press, and "Radar Handbook (Third Edition)" by Merrill I. Skolnik.

[0121] For moving targets across range / azimuth cells, after inter-frame integration, the target amplitude will have a loss of α times, and there will be residual tails at the historical positions, as Figure 8 Figure 9 shown, Figure 8 (a) is the P-display map of the single-frame target echo, Figure 8 (b) is the P-display map of the inter-frame integrated target echo, with a series of historical tails behind; Figure 9 (a) is the amplitude map of the single-frame target echo, Figure 9 (b) is the amplitude map of the inter-frame integrated target echo, with a series of historical tails behind. For this kind of tail false alarm, a slow threshold of aV_T is designed to eliminate this kind of tail false alarm, as Figure 9 (c) shown.

[0122] Further quantitatively analyze the detection loss of moving targets. The analysis method still uses the Monte Carlo statistical method. Assume that the clutter background follows a Rayleigh distribution, the target is a constant amplitude signal, the target completely crosses range / azimuth cells between frames, the number of non-coherent integrations is 20, the inter-frame integration coefficient a = 0.5, 108 background samples are collected after inter-frame integration to obtain the detection thresholds at different false alarm rates; 104 target sample signals after inter-frame integration are collected to statistically analyze the detection probabilities at different false alarm rates. The statistical results are as Figure 10As shown, it can be seen from the figure that the false alarm rate is 10 -3 When it is, for fast targets, the minimum detectable signal-to-noise ratio for single-frame detection is 4.1 dB, and the minimum detectable signal-to-noise ratio for inter-frame accumulation is 4.1 dB, with a detection loss of 2.98 dB; when the false alarm rate is 10 -6 When it is, for fast moving targets, the minimum detectable signal-to-noise ratio for single-frame detection is 2.68 dB, and the minimum detectable signal-to-noise ratio for inter-frame accumulation is 6.1 dB, with a detection loss of 3.42 dB. Here, only the detection loss of moving targets in the Rayleigh clutter background is statistically analyzed. In other clutter distribution backgrounds, the analysis process is similar, but the conclusion is that there will be losses to varying degrees.

[0123] Further analyze the suppression of sea clutter false alarms. The greatest benefit of inter-frame accumulation is to suppress inter-frame uncorrelated sea clutter or sea spike echoes. Because the sea clutter distribution is complex, based on the inter-frame uncorrelated characteristics of sea clutter, k-distributed clutter is generated. After preprocessing and inter-frame accumulation, the number of samples is 105. The false alarm rates of single-frame constant false alarm detection and inter-frame accumulation constant false alarm detection are statistically analyzed by Monte Carlo, and the number of statistical times is 10 times. The analysis results are as Figure 11 shown. Figure 11 From top to bottom, the three figures are the sea spike false alarms after preprocessing, the amplitude of the sea clutter after inter-frame accumulation, and the sea spike false alarms after inter-frame accumulation constant false alarm detection. It can be seen that inter-frame accumulation detection can greatly suppress false alarms. Figure 12 For the sea clutter false alarms of single-frame constant false alarm detection and inter-frame accumulation sea clutter false alarms statistically analyzed by Monte Carlo, it can be seen that the sea clutter false alarms can be effectively suppressed through inter-frame accumulation.

[0124] Step 3-3:

[0125] The detection processing flow of Channel 3 is as Figure 13 shown, and it can also be called moving target detection. The detailed implementation process is as follows.

[0126] To make up for the loss of detecting moving targets in Channel 2, the 3rd channel is designed to detect moving targets with stronger amplitudes. Assume that the video data after preprocessing the k-th frame of the original video is represented as Y k (k = 1, 2,...). After CA-CFAR processing and high threshold V 3T detection, H3 k (k = 1, 2,...) is output. The CA-CFAR in this channel detection is traditional constant false alarm detection and traditional threshold detection. The specific processing process can refer to the constant false alarm detection chapter in "Fundamentals of Radar Signal Processing" by Xing Mengdao, published by Electronics Industry Press.

[0127] Furthermore, in one implementation, the step 4 includes:

[0128] Perform combinational logic judgment on the multi-channel detection video output in step 3. The detection videos of channel 1 and channel 2 are ANDed to further suppress sea clutter false alarms, and ORed with the detection video of channel 3 to ensure the output of moving targets. Assume the final detection video is H k (k = 1, 2,...), then the combinational logic can be expressed as the following formula.

[0129] H k (i, j) = (H1 k (i, j) & H2 k (i, j)) || H3 k (i, j)

[0130] Furthermore, use the actual collected radar raw video data to verify the sea clutter suppression effect and target detection effect, as Figure 14 shown. Figure 14 (a) is the radar raw video data actually collected when a certain coastal base warning radar is working, Figure 14 (b) is the detection result of traditional constant false alarm. The amplitude of sea clutter is strong at close range, and the correlation between sweeps is strong, resulting in high detection false alarms; Figure 14 (c) adopts the processing flow designed in this patent, and the sea clutter is greatly suppressed, and the targets are all detected.

[0131] In specific implementation, this application provides a computer storage medium and a corresponding data processing unit. Among them, the computer storage medium can store a computer program, and when the computer program is executed by the data processing unit, it can run the content of the invention and some or all of the steps in each embodiment of a method for suppressing sea clutter and analyzing performance based on video signals provided by the present invention. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0132] Those skilled in the art can clearly understand that the technical solutions in the embodiments of the present invention can be implemented by means of a computer program and its corresponding general hardware platform. Based on such an understanding, the essence of the technical solutions in the embodiments of the present invention, or the part that contributes to the prior art, can be embodied in the form of a computer program, that is, a software product. The computer program software product can be stored in the storage medium, including several instructions for causing a device (which can be a personal computer, a server, a single-chip microcomputer, a MUU, or a network device, etc.) including a data processing unit to execute the methods described in each embodiment or some parts of the embodiments of the present invention.

[0133] The present invention provides an idea and method for sea clutter suppression processing and performance analysis based on video signals. There are many methods and ways to specifically implement this technical solution. The above description is only the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention. Each component not clearly defined in this embodiment can be implemented by existing technologies.

Claims

1. A method for sea clutter suppression processing and performance analysis based on video signals, characterized in that It includes the following steps: Step 1: Preprocess the original radar video data to obtain the preprocessed video data; Step 2: Perform constant false alarm detection or slow threshold detection on the preprocessed video data to obtain the preliminarily detected video data; Step 3: Perform multi-channel detection processing and performance analysis on the output video data of Step 2, i.e., the preliminarily detected video data; Step 4: Perform combinational logic judgment and actual effect analysis on the multi-channel detection video output from Step 3; Among them, the multi-channels described in Step 3 include: the first channel, i.e., correlation detection; the second channel, i.e., inter-frame accumulation detection; the third channel, i.e., moving target detection compensation; The specific method for performing multi-channel detection processing and performance analysis includes: Step 3-1: Perform the first channel detection processing flow, i.e., correlation detection, including constant false alarm detection and within-window correlation detection, and analyze the performance of the correlation detection to obtain the detection video of the first channel; Step 3-2: Perform the second channel detection processing flow, i.e., inter-frame accumulation detection, including inter-frame accumulation and constant false alarm detection, analyze the performance of the inter-frame accumulation detection to obtain the detection video of the second channel; Step 3-3: Perform the third channel detection processing flow, i.e., moving target detection compensation, to obtain the detection video of the third channel.

2. The method for sea clutter suppression processing and performance analysis based on video signals according to claim 1, wherein The preprocessing described in Step 1 includes: anti-synchronous interference of the same frequency and non-coherent accumulation; The original radar video data is a frame of original video data obtained by a full scan of the radar antenna. Assume the k-th frame of original radar video data is X k , which is represented by a matrix as follows: X k = {x k (i, j)}, i = 1, 2, ..., M, j = 1, 2, ..., N; k = 1, 2, 3, ... where \(i\) represents azimuth sampling, \(j\) represents range sampling, \(M\) is the maximum value of azimuth sampling, \(N\) is the maximum value of range sampling, and \(x\) k (\(i,j\)) is the original video measurement value recorded by the detection unit (\(i,j\)) at the \(k\)-th frame time; Video data Z after preprocessing the original radar video data k It is expressed as: Z k = {z k (i, j)}, i = 1, 2,..., M, j = 1, 2,..., N; k = 1, 2, 3,... Among them, z k (i, j) is x k (i, j) is the preprocessed video data, representing the preprocessed video value of the detection unit (i, j) at the k-th frame moment.

3. A method for sea clutter suppression processing and performance analysis based on video signals according to claim 2, characterized in that Z described in Step 2 k ={z k (i,j)}, i = 1, 2, ..., M, j = 1, 2, ..., N; k = 1, 2, 3, ... When performing constant false alarm rate detection or slow threshold detection, a high false alarm rate threshold is set, that is, the false alarm rate is set to 10 -5 ~10 -2 ; Video data Y after preliminary detection k It is expressed as: Y k = {y k (i,j)}, i = 1, 2,..., M, j = 1, 2,..., N; k = 1, 2, 3,... where y k (i, j) represents z k (i, j) is the video data after constant false alarm rate detection.

4. A method for sea clutter suppression processing and performance analysis based on video signals according to claim 3, characterized in that, The correlation detection described in Step 3-1 is as follows: Assume that the data after constant false alarm detection in the first-channel detection and processing flow of the (k - 1)-th frame is denoted as D k-1 , and the data after constant false alarm detection in the first-channel detection and processing flow of the k-th frame is denoted as D k , and the output after in-window correlation detection is denoted as H1 k , and the processing logic formula is: That is: where H1 k (i, j) represents the output video data after the detection unit (i, j) performs associated detection, represents that at least one detection unit within the association window centered on (i, j) in the (k - 1)-th frame detects a target, represents that no target is detected; The analysis of the performance of the correlation detection, i.e., the analysis of the detection probability under the set false alarm rate, is as follows: In a noise background or sea clutter background, assuming that the single-frame detection false alarm rate is Pfa0 and the false alarm rate after correlation detection is Pfa′, then: Then the false alarm rate Pfa′ after correlation detection is: Pfa′ = Pfa0 * Pfa0 * L Among them, L represents the correlation window size; analyze the detection probability of the target, i.e., fix the false alarm probability, analyze the relationship curve between the detection probability and the signal-to-noise ratio, and the detection logic of the target is as follows: Assuming that the single-frame target detection probability is Pd and the detection probability after correlation detection is Pd′, then: Pd′ = Pd * Pd After the correlation detection of the first channel, the detection performance of the target is the detection probability Pd′.

5. A method for sea clutter suppression processing and performance analysis based on video signals according to claim 4, characterized in that, The inter-frame accumulation detection described in Step 3-2 is as follows: Suppose the video data after the preliminary detection of the k-th frame is Y k (k = 1, 2,...), and the video data after frame accumulation is denoted as SI k (k = 1, 2,...), then the frame accumulation formula is: SI k = aY k + (1 - a)SI k-1 Output the detection video H2 after constant false alarm rate detection k (k = 1, 2,...), in the CA-CFAR processing, set the slow threshold aV T , where V T is the threshold of CA-CFAR in preprocessing, and a represents the frame accumulation coefficient; The analysis of the performance of the inter-frame accumulation detection, i.e., adopt the Monte Carlo statistical method to statistically analyze the performance of the inter-frame accumulation detection, is as follows: First, analyze the change of the false alarm probability in the Rayleigh noise background area: Assume that the original radar video data is X k has Rayleigh distribution noise, set the number of non-coherent integrations and the false alarm rate of preprocessing, and analyze Y k (k = 1, 2,...) the false alarm rate of single-frame CFAR detection and inter-frame accumulation CFAR detection; Secondly, analyze the detection probabilities of stationary targets and moving targets, and analyze them separately for stationary targets and moving targets: Inter-frame accumulation of stationary targets is equivalent to non-coherent accumulation between frames. The benefit of non-coherent accumulation depends on the background clutter distribution and the characteristics of target distribution. For moving targets across range or azimuth cells, after inter-frame accumulation, the target amplitude has a loss of α times and there are residual trails at the historical positions. For the false alarms caused by the said residual trails, a slow threshold of aV T is used for rejection; The detection loss of the moving target adopts the Monte Carlo statistical method: Assume that the clutter background of the original radar video data X k obeys the Rayleigh distribution, the target is a constant amplitude signal, the target completely spans the azimuth range cells between frames, set the number of non-coherent integrations and the inter-frame accumulation coefficient a, collect the background samples after inter-frame accumulation, and obtain the detection thresholds at different false alarm rates; collect the target sample signals after inter-frame accumulation, and statistically calculate the detection probabilities at different false alarm rates; Analyze the suppression of sea clutter false alarms: Based on the non-correlation characteristics of sea clutter between frames, generate k-distributed clutter. After the preprocessing described in Step 1 and the inter-frame accumulation detection, collect samples, and statistically analyze the false alarm rate of single-frame constant false alarm detection and the false alarm rate of inter-frame accumulation constant false alarm detection by Monte Carlo.

6. A method for sea clutter suppression processing and performance analysis based on video signals according to claim 5, characterized in that, The moving target detection compensation described in Step 3-3, i.e., the third channel detection processing flow, is used to make up for the loss of moving targets in the second channel detection processing flow, and the specific method is as follows: Suppose the video data after the preliminary detection of the k-th frame is Y k (k = 1, 2,...), and the output is H3 k (k = 1, 2,...).

7. A method for sea clutter suppression processing and performance analysis based on video signals according to claim 6, characterized in that Perform the combined logic judgment described in step 4, that is, perform an AND operation on the detection videos of the first channel and the second channel, and then perform an OR operation with the detection video of the third channel.

8. A method for sea clutter suppression processing and performance analysis based on video signals according to claim 7, characterized in that, The specific method for performing the combined logic judgment described in step 4 is as follows: Assume that the final detected video is H k (k = 1, 2,...), then the combinational logic is expressed as the following formula: H k (i,j) = (H1 k (i,j) & H2 k (i,j)) || H3 k (i,j) Among them, H1 k (i, j) represents the detection video of the first channel; H2 k (i, j) represents the detection video of the second channel; H3 k (i, j) represents the detection video of the third channel.

9. A method for sea clutter suppression processing and performance analysis based on video signals according to claim 8, characterized in that, The specific method for the actual effect analysis described in step 4 is: Use the original video data of the actual collected radar to verify the sea clutter suppression effect and the target detection effect.

Images