Ship length measuring method based on adaptive bidirectional difference algorithm

Abstract

The invention provides a ship length measuring method based on a self-adaptive bidirectional difference algorithm, aiming at the problem that clutters caused by sea waves are easily misjudged as targets when the sea waves are relatively large. The method comprises the following steps: adaptively calculating the length of a selected data segment according to a signal-to-noise ratio, respectively carrying out differential operation from the left side and the right side according to the length of the selected data segment, adaptively selecting a threshold value for the result of the differential operation, filtering clutters or noises below the threshold value, and removing isolated points to obtain the radial ship length of a target. The real ship length of the target is calculated through the attitude angle, and reference is provided for next target recognition. The method effectively solves the problem that clutters caused by sea waves are likely to be misjudged as targets when the sea waves are large, and provides powerful support for ship length measurement in actual engineering.

Description

Technical Field

[0001] The invention belongs to the field of radar target recognition. Background Art

[0002] At present, as countries pay more and more attention to their own maritime rights and interests, using broadband radar to obtain high-resolution range images of sea surface ships and identify them has gradually become a research hotspot. The captain's characteristics are important information for ships, which can provide support for identifying ships and are of great significance for determining the type of ships.

[0003] At present, the use of threshold method to screen the target area is a common method for ship length estimation. However, in the case of large waves, the simple threshold method is easy to misjudge the outliers caused by the waves as targets, resulting in large radial size estimation errors. Therefore, it is difficult for the simple threshold method to accurately estimate the true length of the target. In order to solve this problem, the paper "HRRP Ship Length Extraction Algorithm for Airborne Radar under Strong Sea Clutter" proposes a ship length extraction algorithm based on mean filtering and adaptive threshold processing, which improves the distinction between target boundary information and clutter. However, due to the different perspectives of shipborne and airborne, the length measurement algorithm suitable for airborne radar cannot be directly used for shipborne radar; in addition, the patent "A Ship Length Estimation Method Fusion of Energy and Rate of Change" uses the characteristics that the amplitude and energy at the bow and stern are significantly larger than those outside the hull, and compares the ratio of the energy on both sides of the test point with the size of the threshold one by one, finds the relative position of the bow and stern, and thus calculates the length of the ship; in addition, there are also methods that use the normalized waveform entropy feature and the mean square error feature of the Doppler center frequency to determine whether there is a target in the distance unit. These two methods are still easy to misjudge in the case of large waves. The differential algorithm can effectively suppress noise signals. When the differential operator is located in the noise area, the output of the differential processing result is basically zero; when the differential operator enters the unit where the target is located, the output amplitude begins to increase; when the differential operator leaves the unit where the target is located, the output amplitude begins to decrease and gradually approaches zero. While reducing the noise signal, the differential algorithm also increases the amplitude of the target signal to a certain extent, making it easier to distinguish the target signal from the clutter signal. Summary of the invention

[0004] In order to solve the problem that the clutter caused by the sea waves is easily misjudged as a target when the sea waves are large, the present invention proposes a ship length measurement method based on an adaptive two-way differential algorithm. The length of the data segment is selected according to the signal-to-noise ratio adaptive calculation, and the differential operation is performed from the left and right sides respectively according to the selected data segment length, and the threshold value is adaptively selected for the result of the differential operation, and the clutter or noise below the threshold value is filtered out and the isolated points are removed to obtain the radial length of the target, and the real length of the target is calculated by the attitude angle, which provides a basis for the next step of target identification.

[0005] To achieve the above technical objectives, the present invention proposes a method for measuring ship length based on an adaptive two-way differential algorithm, the implementation process of which includes:

[0006] The first step is to adaptively calculate and select the length of the data segment based on the signal-to-noise ratio.

[0007] After intermediate frequency processing, the n-dimensional target feature data is obtained, the target feature data is normalized, and the normalized data is recorded as x. The signal-to-noise ratio snr of the target data is calculated, and the data segment length M is adaptively selected according to the signal-to-noise ratio. The calculation formula for selecting the data segment length is as follows:

[0008]

[0009] Among them, m1 and m2 are the minimum and maximum boundaries of the selected data segment length, and snr1 and snr2 are the minimum and maximum signal-to-noise ratio boundaries.

[0010] The second step is to perform differential operations from the left and right sides to the maximum value point.

[0011] The third step is to adaptively select the target threshold value.

[0012] During the sliding process, if the mean values ​​of two adjacent segments differ by more than three times, the reference point is recorded as the first critical point, and the data on the left and right sides of the first critical point are recorded as noise data. The first critical point of the left sliding is recorded as Pb, and the first critical point of the right sliding is recorded as Pe. The target noise area can be expressed as y(0:Pb) and y(Pe:n). By comparing the mean value of the differential data and the maximum value of the noise area, the threshold values ​​TL and TR on the left and right sides of the target are obtained by the following formula.

[0013]

[0014] According to the obtained threshold value, the data on the left and right sides are screened respectively to obtain the data y1 that exceeds the threshold. The screening formula is as follows:

[0015]

[0016] The fourth step is to remove isolated points that are far apart.

[0017] Mark the non-zero coordinates in y1 as P, and look for the first point whose adjacent difference value is not zero and whose distance exceeds d from the maximum value point to the left and right. Remove the points whose distance exceeds d to obtain the target data z after removing isolated points. The calculation formula is as follows:

[0018]

[0019] Step 5: Calculate the target's real length.

[0020] Find the starting position start and ending position end of the target data z after removing isolated points. The resolution of the radar is Δr, and the attitude angle of the target is α. Then the real length L of the target can be expressed as:

[0021]

[0022] Furthermore, in the second step, the position k of the maximum value of the normalized data is found, and a differential operation is performed from the left and right sides. The target echo range starts from the first point and slides from the left and right sides respectively. When sliding to a certain position i, 2M points on the left and right sides enter the differential window. For the left window, the mean of the first M units is LE1i, and the variance is LEv1i. The mean of the last M units is LE2i, and the variance is LEv2i. At this time, the value of i is i=1,2,…,k, then LE1i, LE2i, LEv1i, and LEv2i can be expressed as:

[0023]

[0024] As the window slides, the amplitude value of the distance unit in the difference window is processed, and the operation is stopped when the maximum value position k is reached to obtain the left differential data Ly. The processing process is as follows:

[0025]

[0026] Similarly, for the right window, at the i-th moment (i=1,2,…,k), the means and variances RE1i, RE2i, REv1i, and REv2i of the first M units and the last M units are:

[0027]

[0028]

[0029] The result Ry after the difference on the right side is:

[0030]

[0031] Combine the difference results on the left and right sides according to the following formula to obtain the target difference data y:

[0032]

[0033] Preferably, n=1500; m1=10, m2=20; snr1=40, snr2=50; d=120; Δr=0.75.

[0034] The proposal and engineering implementation of the present invention have high promotion and application value in the field of ship length measurement. The present invention uses the signal-to-noise ratio to adaptively calculate the length of the selected data segment, effectively reducing the mean error caused by improper length selection; performing differentiation from the left and right sides, effectively solving the problem of overall data translation caused by the differential algorithm; adaptively selecting the threshold according to the mean and the maximum value of the clutter, effectively separating the target from the clutter. The present invention can effectively calculate the length of the ship based on the broadband one-dimensional range image of the ship, providing a reference for subsequent target identification, and has important engineering application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Attached Figure 1 Flowchart of the ship length measurement method based on the adaptive two-way difference algorithm. DETAILED DESCRIPTION

[0036] The present invention will be further explained below in conjunction with the accompanying drawings.

[0037] The present invention proposes a ship length measurement method based on an adaptive two-way differential algorithm, which effectively solves the problem that clutter caused by waves is easily misjudged as a target when the waves are large, and provides strong support for the engineering application and promotion of ship length calculation when the waves are large.

[0038] The preferred implementation example flow chart of the present invention is as shown in the attached Figure 1 As shown, the embodiments are as follows:

[0039] Assuming that 1500-dimensional data is obtained after intermediate frequency data processing, the target is normalized, and the signal-to-noise ratio of the target data after normalization is 55, then the length of the data segment to be selected is 20.

[0040] Assuming that the maximum value of the normalized data is at the 750th position, perform differential operations on the left and right sides. For the left window, when it moves to the position of the 50th data, the means and variances of the first 20 units and the last 20 units are:

[0041]

[0042] The result after difference is:

[0043]

[0044] For the right window, at the 50th data position, the means and variances of the first 20 units and the last 20 units are:

[0045]

[0046] The result after difference is:

[0047]

[0048] The 50th bit of the differential data y obtained at this time 50 =Ly 50 and the 1450th bit data y 1450 =Ry 50 , the data y after 1500-bit difference can be obtained in the same way.

[0049] During the sliding process, the means on the left and right sides are compared. When the difference between the means on the left and right sides is more than 3 times, the first critical point of the current reference point is recorded, and the data on the left and right sides of the first critical point are recorded as noise data. The mean of the differential data and the maximum value of the noise area are compared, and the larger one is selected as the threshold to filter the differential data, and the isolated points far away from the target are removed to obtain data containing only the target.

[0050] Assuming that the radar's range resolution is 0.75 meters, the target's attitude angle is 60 degrees, and the target's starting and ending positions in the echo data are 530 and 640 respectively, the target's length is:

[0051]

Claims

1. A method for measuring ship length based on an adaptive two-way differential algorithm, characterized in that: The first step is to adaptively calculate and select the length of the data segment based on the signal-to-noise ratio; After intermediate frequency processing, the n-dimensional target feature data is obtained, the target feature data is normalized, and the normalized data is recorded as x. The signal-to-noise ratio snr of the target data is calculated, and the data segment length M is adaptively selected according to the signal-to-noise ratio. The calculation formula for selecting the data segment length is as follows: Among them, m1 and m2 are the minimum and maximum boundaries of the selected data segment length, and snr1 and snr2 are the minimum and maximum signal-to-noise ratio boundaries; The second step is to perform differential operations from the left and right sides to the maximum point position; The third step is to adaptively select the target threshold value. During the sliding process, if the mean values ​​of two adjacent segments differ by more than three times, the reference point is recorded as the first critical point, and the data on both sides of the first critical point are recorded as noise data. The first critical point of the left sliding is recorded as Pb, and the first critical point of the right sliding is recorded as Pe. The target noise area is represented by y(0:Pb) and y(Pe:n). The mean value of the differential data and the maximum value of the noise area are compared, and the threshold values ​​TL and TR on the left and right sides of the target are obtained by the following formula: According to the obtained threshold value, the data on the left and right sides are screened respectively to obtain the data y1 that exceeds the threshold. The screening formula is as follows: The fourth step is to remove isolated points that are far apart; mark the coordinates of non-zero points in y1 as P, and look for the first point whose adjacent difference value is not zero and whose distance exceeds d from the maximum value point to the left and right. Remove the points whose distance exceeds d to obtain the target data z for removing isolated points. The calculation formula is as follows: Step 5: Calculate the target's real length; Find the starting position start and ending position end of the target data z after removing isolated points. The resolution of the radar is Δr, and the attitude angle of the target is α. Then the true length L of the target can be expressed as:

2. The method for measuring ship length based on an adaptive bidirectional differential algorithm according to claim 1, characterized in that: Find the position k of the maximum value of the normalized data, perform differential operation from the left and right sides, start sliding from the first point from the left and right sides within the target echo range, and when sliding to a certain position i, there are 2M points on the left and right sides entering the differential window. For the left window, the mean of the first M units is recorded as LE1 i , the variance is LEv1 i , the mean of the last M units is LE2 i , the variance is LEv2 i , at this time, the value of i is i=1,2,…,k, then LE1 i LE2 i LEv1 i LEv2 i They can be expressed as: As the window slides, the amplitude value of the distance unit in the difference window is processed, and the operation is stopped when the maximum value position k is reached to obtain the left differential data Ly. The processing process is as follows: Similarly, for the right window, at the i-th moment (i=1,2,…,k), the mean and variance RE1 of the first M units and the last M units i 、RE2 i 、REv1 i 、REv2 i They are: The result Ry after the difference on the right side is: Combine the difference results on the left and right sides according to the following formula to obtain the target difference data y:

Images