Medium and high orbit SAR ship shaking target detection method

Abstract

The invention discloses a method for detecting a shaking target of a medium-high orbit SAR (Synthetic Aperture Radar) ship, which belongs to the technical field of radar signal processing, and comprises the following steps of: firstly, extracting one-dimensional range feature information of each fixed azimuth of the shaking target by adopting an azimuth grouping and accumulating mode, then carrying out CFAR (Constant False Alarm Rate) detection on the extracted one-dimensional signal, and finally, carrying out CAR detection on the extracted one-dimensional signal; the method comprises the following steps: detecting a shaking target, performing over-detection point condensation on each group of over-detection points, intercepting information of a target center point and a distance direction width value, performing re-condensation on an over-detection group, and finally performing re-focusing processing on a target row and column number provided after detection to obtain an accurate two-dimensional image of the shaking target, thereby verifying a detection result. According to the method, target detection can be carried out aiming at the condition that the imaged shaking target is in a long strip shape and loses the two-dimensional characteristic information characteristic of a ship.

Description

Technical Field

[0001] The invention belongs to the technical field of radar signal processing, and in particular relates to a medium- and high-orbit SAR ship swaying target detection method. Background Art

[0002] When a surface ship is sailing on the sea, it will experience spatial translation and complex swaying driven by waves. The synthetic aperture time of traditional low-orbit SAR is usually around a few seconds, and surface ships can usually be approximately considered to have only spatial translation, ignoring the swaying. The medium- and high-orbit SAR has a high orbit altitude and a long synthetic aperture time, which can reach up to tens of seconds. The swaying of surface ships cannot be ignored. The swaying of the target causes a large difference in Doppler at different points on the target, which will cause serious envelope offset and phase deviation in the echo, and then cause the imaging of the swaying ship target to be severely defocused, and it is presented in the form of a long strip in the two-dimensional image, making it difficult to detect and locate the position of the ship target. Therefore, how to detect the swaying ship target of the medium- and high-orbit SAR in an efficient way is of great significance. Summary of the invention

[0003] In view of this, the present invention provides a medium-high orbit SAR ship swaying target detection method, which can detect the swaying target after imaging, which is in the shape of a long strip and has lost the two-dimensional characteristic information characteristics of the ship.

[0004] The technical solution for implementing the present invention is as follows: A medium-high orbit SAR ship swaying target detection method comprises the following steps: S1: Divide the shaking target into regions in the azimuth direction and perform incoherent accumulation in the azimuth direction to extract the one-dimensional distance feature information; S2: The one-dimensional signal of the shaking target is detected based on the unit average constant false alarm rate (CA-CFAR); S3: Gather the inspection points and intercept the target center point and distance width value information; S4: The inspection group is condensed to determine the row and column number of the target center; S5: According to the row and column number of the target center, the ISAR imaging processing method is selected for refocusing to obtain accurate two-dimensional imaging of the shaking target, thereby verifying the detection result.

[0005] Furthermore, the specific process of condensing the check points is as follows: subtract the coordinates of each check point from its subsequent coordinates, and if the threshold condition is met, they are considered to belong to the same target, and the same target flag is set; if the condensation threshold is not met, a new target flag is set; after all targets are traversed, the flag information of all coordinates after condensation is obtained, and then the isolated points are removed from the condensed coordinate information, the target center point and distance width value information are intercepted, and finally the information of all targets after condensation is obtained.

[0006] Furthermore, the strategy for intercepting the target center point is: in the group signals obtained after point aggregation, the maximum signal of each group is extracted as the center point of the shaking target distance. If there are several maximum values ​​at the same time, the middle position value between the first maximum value and the last maximum value in each group is taken as the center point.

[0007] Furthermore, the strategy for intercepting the distance width value information is: save the maximum value of each group after the inspection points are condensed, and record the row numbers of the first and last values ​​in each group that differ from the maximum value by a fixed pixel value, and obtain the distance width value of the shaking target by the difference.

[0008] Furthermore, the specific process of the inspection group cohesion is as follows: first, the number of distance-related association channels is set, and the targets in the inspection group with close association positions are divided into a group, and each group of pixels to be fused is traversed, and the amplitude difference between the pixel to be fused and its next pixel and the cohesion threshold are compared to determine whether it is cohesive; if it is less than the cohesion threshold, it indicates cohesion, otherwise it is not cohesive; after all targets in the group are traversed, the target information after cohesion in the group is obtained; then the row and column numbers of the target center point are determined in each group, and the row is determined first, and the number of pixels in each row whose pixel amplitude exceeds the set threshold is counted, and the row with the largest number is taken as the row where the target center point is located, and then the column where the pixel with the largest amplitude in this row is taken as the column where the target center point is located; if the number of values ​​that meet the conditions in each row is the same, the row and column where the maximum amplitude is located is directly taken, so as to more accurately determine the row and column numbers where the target center point is located.

[0009] Beneficial effects: 1. The present invention utilizes the over-detection point aggregation technology to eliminate isolated points and redundant targets detected by CFAR, further improving the target detection effect.

[0010] 2. The present invention uses the inspection group aggregation technology to simplify the related targets, so as to obtain the target position more accurately.

[0011] 3. The present invention eliminates isolated points and redundant targets detected by CFAR, simplifies related targets, and further accurately obtains the target position through the over-inspection point aggregation technology, over-inspection group aggregation technology and refocusing verification technology, thereby solving the problem of severe defocusing of the imaging of swaying targets of medium and high orbit SAR ships and difficulty in detecting and locating the target position, and greatly improves the detection effect of swaying targets. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is an overall flow chart of the shaking target detection of the present invention.

[0013] Figure 2 It is a one-dimensional image after incoherent superposition of the shaking target in azimuth.

[0014] Figure 3is a two-dimensional image of the shaking target.

[0015] Figure 4 This is the flow chart of CFAR detection technology for shaking targets.

[0016] Figure 5 To condense the overall flow chart for inspection.

[0017] Figure 6 Re-compile the overall flow chart for the inspection team.

[0018] Figure 7 This is an example of the ship swaying target detection result.

[0019] Figure 8 This is an example of a ship target after refocusing. DETAILED DESCRIPTION

[0020] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0021] The present invention provides a method for detecting swaying targets of medium- and high-orbit SAR ships. First, the swaying targets are grouped and accumulated in azimuth, and one-dimensional distance feature information is extracted for each fixed azimuth. Then, CFAR detection is performed on the extracted one-dimensional signal, and each group of over-detection points is condensed, and the target center point and distance width value information are intercepted. Next, the over-detection group is recondensed, and finally the target row and column numbers provided after the detection are refocused to obtain accurate two-dimensional imaging of the swaying target, thereby verifying the detection result. The overall process is as follows: Figure 1 shown. S1: One-dimensional feature information extraction: Due to the synthetic aperture time of medium and high orbit SAR, the shaking of ship targets will cause serious defocusing of the image, affecting the detection performance of ship targets. Therefore, the imaging processing of the two-dimensional NCS algorithm is adopted to reduce the degree of defocusing of the ship target shaking to a certain extent and improve the detection performance of the target.

[0022] After imaging, the swaying target of the ship appears in a long strip shape, and the two-dimensional feature information of the ship has been lost. In order to detect the swaying target, the present invention designs a strategy for extracting one-dimensional feature information of the swaying target. First, the azimuth accumulation method of the swaying target is adopted, that is, the image is first divided into small blocks with fixed azimuth areas in the azimuth direction, and incoherently superimposed along the azimuth direction to obtain the one-dimensional distance information of the swaying target. Figure 2 It is a one-dimensional image after incoherent superposition in azimuth. The targets from top to bottom are shaking targets, stationary targets, and translational targets. Figure 3 It is a two-dimensional image of the shaking target, and the three targets correspond one to one.

[0023] S2: One-dimensional signal CFAR detection: The one-dimensional signal of the shaking target is detected by the detection technology based on the unit average constant false alarm rate (CA-CFAR). The technical flow chart is as follows: Figure 4 shown.

[0024] The pulse compression process is performed on the one-dimensional distance characteristic information of the shaking target to obtain the signal to be detected. Assume that the signal to be detected is Contains the sample values ​​of n distance units in total , if there is no ship target at the i-th distance unit, then the assumption is satisfied , otherwise the assumption is satisfied , as shown below:

[0025] Among them, the clutter amplitude of different distance units are independent of each other and have the same Rayleigh distribution. s is the target signal, which represents the signal component generated by the target in the i-th distance unit when there is a ship target.

[0026] After square-law detection, the square value of the clutter amplitude follows the same exponential distribution. Assuming that the parameter of the Rayleigh distribution is b, the parameter of the exponential distribution is ,but The square value of the signal amplitude at each distance unit The corresponding probability density function is:

[0027] In the assumption When , the range cell contains only the clutter part. Assume When established, the range unit contains two parts: clutter and ship target. , SCR is the ratio of the target signal to the average power of the clutter. reference units and protection units, assuming that the signal in the reference unit contains only the clutter part, the square values ​​of the amplitudes obey the same exponential distribution and are independent of each other. The sum of the squared amplitudes of the reference units follows the Gamma distribution. , the corresponding probability density function is:

[0028] CA-CFAR Assay Selection As the detection threshold, is the factorial symbol, T is a number corresponding to the reference unit and false alarm probability If the distance unit to be detected does not contain the ship target, the false alarm probability is:

[0029] in, x is the signal amplitude value of the distance unit to be detected. It can be seen from the above formula that the false alarm probability It has nothing to do with the parameters and intensity of the clutter distribution, but only with the constant T and the number of reference cells. The detection threshold can be adjusted as the background clutter changes, thus ensuring that the false alarm probability remains approximately unchanged.

[0030] S3: Checkpoint aggregation: After CFAR detection, there are multiple checkpoints, but multiple checkpoints may belong to the same target. The checkpoint aggregation flow chart is as follows: Figure 5 As shown in the figure, the coordinates of each check point are subtracted from its subsequent coordinates. If the threshold condition is met, they are considered to belong to the same target and the same target flag is set. If the condensation threshold is not met, a new target flag is set. After all targets are traversed, the flag information of all coordinates after condensation is obtained, and then the isolated points are removed from the condensed coordinate information, and then the target center point and distance width value information are intercepted, and finally the information of all targets after condensation is obtained. The strategy for intercepting the target center point and distance width value information is as follows: (1) Intercepting the center point: In the group signals obtained after point aggregation, extract the maximum value of each group as the center point of the shaking target distance. If there are several maximum values ​​at the same time, take the middle position value between the first maximum value and the last maximum value in each group as the center point.

[0031] (2) Intercept the distance width value of the shaking target: Save the maximum value of each group after the inspection points are condensed, and record the row numbers of the first and last values ​​in each group that differ from the maximum value by a fixed pixel value, and obtain the distance width value of the shaking target by the difference. S4: Over-inspection group aggregation: After the above steps, multiple sliding areas with fixed values ​​in the orientation area can be obtained. In order to distinguish between shaking targets and horizontal and stationary targets, a threshold of the number of target orientation areas is set. When the target length is greater than the threshold, these orientation areas are determined to be an over-inspection group, and the information of the maximum amplitude point is taken as the target information of the over-inspection group. These over-inspection groups have one or more targets. In order to eliminate the same targets, a group aggregation strategy is adopted. The over-inspection group re-aggregation flow chart is as follows Figure 6 shown.

[0032] The inspection group cohesion first sets the number of distance-related channels, divides the targets in the inspection group with close associated positions into a group, traverses each group of pixels to be fused, and compares the amplitude difference between the pixel to be fused and the next pixel and the cohesion threshold to determine whether it is cohesive; if it is less than the cohesion threshold, it means cohesion, otherwise it is not cohesive; after all targets in the group are traversed, the target information after cohesion in the group is obtained; then determine the row and column numbers of the target center point in each group, first determine the row, count the number of pixels in each row whose pixel amplitude exceeds the set threshold, take the row with the largest number as the row where the target center point is located, and then take the column where the pixel with the largest amplitude in this row is located as the column where the target center point is located; if the number of values ​​that meet the conditions in each row is the same, directly take the row and column where the maximum amplitude is located, so as to more accurately determine the row and column numbers of the target center point.

[0033] S5: Refocusing: The medium-high orbit SAR has a high orbit height and a long synthetic aperture time, which can reach up to tens of seconds. The target shakes and has relative motion with the satellite. The traditional SAR imaging method cannot focus the ship target. Therefore, the present invention adopts the ISAR imaging method, which uses the Doppler change caused by the rotation of the ship relative to the radar line of sight to compensate for the translation component, and uses the space-varying Doppler component of the ship target to achieve the acquisition of ISAR images. According to the row and column number of the target center corresponding to the ship target detection result, the corresponding echo of the ship target is intercepted, and the ISAR imaging processing method is used to achieve the refocusing of the ship target, and the accurate two-dimensional imaging of the ship shaking target is obtained, so as to verify the detection result.

[0034] S6: During the algorithm verification process, the detection performance of the algorithm is verified by simulating the echo of the ship surface target under the medium and high orbit conditions and the measured data, and performing imaging detection on the echo and refocusing processing on the detection results. Figure 7 As shown in the figure, the example of the ship target result after refocusing the detected row and column numbers is as follows Figure 8 shown.

[0035] In summary, the above are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A medium-high orbit SAR ship swaying target detection method, characterized in that: The following steps are involved: S1: Divide the shaking target into regions in the azimuth direction and perform incoherent accumulation in the azimuth direction to extract the one-dimensional distance feature information; S2: The one-dimensional signal of the shaking target is detected based on the unit average constant false alarm rate; S3: Gather the inspection points and intercept the target center point and distance width value information; S4: The inspection group is condensed to determine the row and column number of the target center; S5: According to the row and column number of the target center, the ISAR imaging processing method is selected for refocusing to obtain accurate two-dimensional imaging of the shaking target, thereby verifying the detection result.

2. The shaking target detection method according to claim 1, characterized in that: The specific process of the checkpoint aggregation is as follows: the coordinates of each checkpoint are subtracted from its subsequent coordinates. If the threshold condition is met, they are considered to belong to the same target and the same target flag is set. If the aggregation threshold is not met, a new target flag is set. After all targets are traversed, the mark information of all coordinates after condensation is obtained, and then the isolated points are removed from the condensed coordinate information, the target center point and distance width value information are intercepted, and finally the information of all targets after condensation is obtained.

3. The shaking target detection method according to claim 2, characterized in that: The strategy for intercepting the target center point is: in the group signals obtained after point aggregation, the maximum signal of each group is extracted as the center point of the shaking target distance. If there are several maximum values ​​at the same time, the middle position value between the first maximum value and the last maximum value in each group is taken as the center point.

4. The shaking target detection method according to claim 2 or 3, characterized in that: The strategy for intercepting the distance width value information is: save the maximum value of each group after the inspection points are condensed, and record the row numbers of the first and last values ​​in each group that differ from the maximum value by a fixed pixel value, and the difference is used to obtain the distance width value of the shaking target.

5. The shaking target detection method according to claim 2, characterized in that: The specific process of the over-inspection group condensation is as follows: first, set the number of distance-related associated channels, divide the objects in the over-inspection group with close associated positions into one group, traverse each group of pixels to be fused, compare the amplitude difference between the pixel to be fused and the next pixel and the condensation threshold, and then determine whether to condense; If it is less than the cohesion threshold, it indicates cohesion, otherwise it is not cohesive; after all targets in the group are traversed, the target information after cohesion in the group is obtained; then the row and column numbers of the target center point are determined in each group, the row is determined first, and the number of pixels in each row whose pixel amplitude exceeds the set threshold is counted, and the row with the largest number is taken as the row where the target center point is located, and then the column where the pixel with the largest amplitude in this row is located is taken as the column where the target center point is located; if the number of values ​​that meet the conditions in each row is the same, the row and column where the maximum amplitude is located is directly taken, so as to more accurately determine the row and column number where the target center point is located.

Images