Search

CN-122017834-A - Method for detecting wake direction of water surface ship from synthetic aperture radar image

CN122017834ACN 122017834 ACN122017834 ACN 122017834ACN-122017834-A

Abstract

The invention discloses a method for detecting the wake direction of a water surface ship from a synthetic aperture radar image, which comprises a water surface ship identification step, a rectangular frame step for surrounding the ship and the wake thereof, a Curvelet coefficient step for obtaining a rectangular frame corresponding to the radar image, a scale sequence number step for determining Curvelet coefficient used for detection, an average value step for calculating Curvelet coefficient matrixes in all directions under a selected scale, a direction step for determining the maximum coefficient matrix average value under the selected scale, a normal direction step for determining the wake of the ship, a 180-degree direction blurring step and a wake direction step for determining the wake direction of the ship, and is used for automatically detecting the wake direction of the water surface ship from the synthetic aperture radar image.

Inventors

  • Song Xingai
  • Zha Guozhen

Assignees

  • 国家卫星海洋应用中心
  • 江苏海洋大学

Dates

Publication Date
20260512
Application Date
20260130

Claims (10)

  1. 1. A method for detecting the wake direction of a water surface ship from a synthetic aperture radar image is characterized by comprising the following steps: s1, constructing and training a convolutional neural network model for detecting a target of a water surface ship; s2, inputting the to-be-detected synthetic aperture radar image into a convolutional neural network model in S1, if a water surface ship target is detected, acquiring coordinates of four vertexes of a rectangular frame surrounding the ship and a wake thereof, and determining coordinates of a center point of the rectangular frame; s3, intercepting a synthetic radar sub-image corresponding to the rectangular frame area from the original synthetic aperture radar image to be detected according to coordinates of four vertexes of the rectangular frame; s4, curvelet transformation is carried out on the radar sub-image intercepted in the S3, and Curvelet coefficient matrixes with different scales and different directions are obtained; s5, determining a scale sequence number of a Curvelet coefficient used for detecting the wake direction of the water surface ship based on the spatial resolution of the synthetic aperture radar image, the width of the ship wake and the widths of Curvelet functions of different scales, wherein the selected scale sequence number corresponds to the width of the Curvelet function and the number of pixels of the ship wake width; S6, calculating the average value of Curvelet coefficient matrixes in different directions under the selected scale for detecting the direction of the water surface ship, and sequencing the average value; S7, determining the serial number of the direction of the maximum average value of the Curvelet coefficient matrix for the selected scale for detecting the ship trail direction; s8, calculating the angle corresponding to the maximum direction of the average value of the Curvelet coefficient matrix according to the total direction number, the angle corresponding to the 1 st direction Curvelet and the serial number of the maximum direction of the average value of the coefficient matrix for the selected scale for detecting the ship trail direction; S9, determining the normal direction of the ship tail trace according to the angle corresponding to the maximum direction of the Curvelet coefficient matrix average value; S10, determining the ship trail direction according to the perpendicular relation between the ship trail and the normal line of the ship trail, and obtaining 2 angles, wherein the ship trail direction is blurred by 180 degrees; S11, the sea surface ship target has a strong backward scattering coefficient, presents a bright spot in a synthetic aperture radar image, and determines the coordinates of the center point of the bright spot; S12, determining a vector by taking the central point of a rectangular frame surrounding the ship and the wake thereof as a starting point and the central point of a bright spot corresponding to the ship as an end point, and calculating the included angle between the vector and the geographic north by taking the anticlockwise direction as the forward direction ; S13, selecting the angle closest to the included angle obtained in the S12 from 2 possible angles of the ship wake direction obtained in the S10 as the ship wake direction, and eliminating 180-degree direction ambiguity to determine the direction of the ship wake on the water surface.
  2. 2. The method of claim 1, wherein in step S1, the convolutional neural network model includes, but is not limited to, a YOLO series object detection model.
  3. 3. The method of claim 1, wherein the Curvelet transform is a post wavelet transform, and the images are decomposed in different scales and directions to obtain Curvelet coefficients of different scales, different directions and different positions, and the Curvelet coefficients of different positions form a Curvelet coefficient matrix for a certain determined scale and direction.
  4. 4. The method of claim 1, wherein the number of scales and the number of directions of the Curvelet transform are set according to the resolution of the synthetic aperture radar image and the number of rows and columns of the selected rectangular frame, the larger the number of rows and columns of the radar sub-image corresponding to the rectangular frame is, the larger the number of scales and the number of directions are, the number of scales can be set to 5 to 7, and the number of directions can be set to 64, 256, 512 or 1024.
  5. 5. The method of claim 1, wherein the number of pixels corresponding to the ship wake width is determined based on the spatial resolution of the radar image and the width of the ship wake, the scale number of Curvelet coefficients used for detecting the direction of the ship wake on the water surface is determined based on the widths of Curvelet functions of different scales, the wake width of the target ship is obtained empirically, and the target ship wake width is assumed to be recorded as The spatial resolution of the radar image is noted as The number of pixels corresponding to the width of the ship wake is Function of Representation of real numbers The method comprises the steps of rounding, determining the number of scales of Curvelet transformation, determining the space width of Curvelet functions of all scales, selecting the scale with the closest space width of Curvelet functions and the pixel number of ship trail width as the Curvelet coefficient scale used for detecting the trail direction of the water surface ship, and recording the scale sequence number.
  6. 6. The method according to claim 1, wherein in step S6, for the selected scale for detecting the direction of the ship' S wake, the sequence number of the direction of maximum average value of the Curvelet coefficient matrix is determined, and the average value of the coefficient matrix is calculated by calculating, for each direction of the coefficient matrix, the absolute value of the Curvelet coefficient at each position in the coefficient matrix, and calculating the average value of the absolute values of all elements in the matrix.
  7. 7. The method of claim 1, wherein the specific method in S8 is as follows, assuming that the first is selected A scale at which the total direction number is recorded as The angle corresponding to the 1 st direction Curvelet is recorded as The serial number of the maximum direction of the average value of the coefficient matrix is recorded as Angle step of Curvelet rotation The angle of the maximum direction of the coefficient matrix average value is The unit is angle.
  8. 8. The method of claim 1, wherein the ship ' S wake normal direction and the ship ' S wake are perpendicular in S10, and the angle between the normal direction and the ship ' S wake is 90 degrees, and the unit is the angle.
  9. 9. The method of claim 1, wherein the specific method in S12 is as follows, the center point of the rectangular frame surrounding the ship and its wake is , The central point of the corresponding bright spots of the ship is recorded as , A vector taking the center point of the rectangular frame as a starting point and the center point of the bright spot as an end point is recorded as , Taking one point above the center point of the rectangular frame randomly along the geographic north direction, and marking as , , Is an arbitrary positive integer, takes the center point of a rectangular frame as a starting point, The vector being the end point is noted as , The included angle between the two vectors is Wherein Is that Is an inverse cosine function of' "Means the dot product of two vectors.
  10. 10. The method of claim 1, wherein the specific calculation in S13 is to calculate 2 angles obtained in S10 and an included angle obtained in S12 Of the 2 differences, the angle with the smallest absolute value determines the direction of the tail ship trail.

Description

Method for detecting wake direction of water surface ship from synthetic aperture radar image Technical Field The invention belongs to the technical fields of marine remote sensing, marine target detection and harmonic analysis, and particularly relates to a method for detecting the wake direction of a water surface ship from a synthetic aperture radar image. Background The satellite remote sensing technology is utilized to obtain an observation image of a large-area sea area, the visible light image is easily influenced by cloud layers and illumination conditions, the synthetic aperture radar is not easily influenced by the cloud layers and the illumination conditions, and the sea surface can be observed under the condition that the cloud layers are blocked and at night. The prior art mostly detects ship targets from synthetic aperture radar images, for example, patent publication No. CN106170819B discloses a rapid detection method for the ship targets of the synthetic aperture radar images, which is used for primarily screening SAR image targets, dividing the images into a plurality of sub-image blocks, and finally, performing constant false alarm detection on three distributed patterns by using CUDA technology to detect effective ship targets, but cannot automatically acquire the directions of ship trails. If the ship trail direction is to be obtained, further manual interpretation is needed, time and labor are wasted, and a certain requirement is made on the technical level of interpretation personnel. The advancing direction of the ship can be judged according to the wake direction of the water surface ship. The method for obtaining the advancing direction of the surface ship has important significance for controlling the offshore traffic, striking smuggling and maintaining offshore safety, and has important significance for developing work of sea polices, naval forces and the like. Disclosure of Invention The invention aims to provide a method for detecting the wake direction of a water surface ship from a synthetic aperture radar image. In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: A method of detecting a wake direction of a surface vessel from a synthetic aperture radar image, the method comprising: s1, constructing and training a convolutional neural network model for detecting a target of a water surface ship; s2, inputting the to-be-detected synthetic aperture radar image into a convolutional neural network model in S1, if a water surface ship target is detected, acquiring coordinates of four vertexes of a rectangular frame surrounding the ship and a wake thereof, and determining coordinates of a center point of the rectangular frame; s3, intercepting a synthetic radar sub-image corresponding to the rectangular frame area from the original synthetic aperture radar image to be detected according to coordinates of four vertexes of the rectangular frame; s4, curvelet transformation is carried out on the radar sub-image intercepted in the S3, and Curvelet coefficient matrixes with different scales and different directions are obtained; s5, determining a scale sequence number of a Curvelet coefficient used for detecting the wake direction of the water surface ship based on the spatial resolution of the synthetic aperture radar image, the width of the ship wake and the widths of Curvelet functions of different scales, wherein the selected scale sequence number corresponds to the width of the Curvelet function and the number of pixels of the ship wake width; S6, calculating the average value of Curvelet coefficient matrixes in different directions under the selected scale for detecting the direction of the water surface ship, and sequencing the average value; S7, determining the serial number of the direction of the maximum average value of the Curvelet coefficient matrix for the selected scale for detecting the ship trail direction; s8, calculating the angle corresponding to the maximum direction of the average value of the Curvelet coefficient matrix according to the total direction number, the angle corresponding to the 1 st direction Curvelet and the serial number of the maximum direction of the average value of the coefficient matrix for the selected scale for detecting the ship trail direction; S9, determining the normal direction of the ship tail trace according to the angle corresponding to the maximum direction of the Curvelet coefficient matrix average value; S10, determining the ship trail direction according to the perpendicular relation between the ship trail and the normal line of the ship trail, and obtaining 2 angles, wherein the ship trail direction is blurred by 180 degrees; S11, the sea surface ship target has a strong backward scattering coefficient, presents a bright spot in a synthetic aperture radar image, and determines the coordinates of the center point of the bright spot; S12, determining a vector by taking the central poin