CN115015903B - Radar sequential image moving target detection method and system - Google Patents
Radar sequential image moving target detection method and system Download PDFInfo
- Publication number
- CN115015903B CN115015903B CN202210626011.5A CN202210626011A CN115015903B CN 115015903 B CN115015903 B CN 115015903B CN 202210626011 A CN202210626011 A CN 202210626011A CN 115015903 B CN115015903 B CN 115015903B
- Authority
- CN
- China
- Prior art keywords
- kernel function
- radar
- frame
- preset
- vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 55
- 239000013598 vector Substances 0.000 claims abstract description 136
- 239000011159 matrix material Substances 0.000 claims abstract description 65
- 238000010606 normalization Methods 0.000 claims abstract description 59
- 238000000034 method Methods 0.000 claims abstract description 27
- 230000006835 compression Effects 0.000 claims description 38
- 238000007906 compression Methods 0.000 claims description 38
- 238000004364 calculation method Methods 0.000 claims description 34
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 230000011218 segmentation Effects 0.000 claims description 6
- 238000003384 imaging method Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012892 rational function Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/02—Systems using reflection of radio waves, e.g. primary radar systems; Analogous systems
- G01S13/50—Systems of measurement based on relative movement of target
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
- G01S7/415—Identification of targets based on measurements of movement associated with the target
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention relates to a radar sequential image moving target detection method and system, in particular to the technical field of signal processing. The method comprises the following steps: sequentially dividing radar real data and performing fast Fourier transform to obtain images corresponding to each frame of radar distance compressed data; respectively carrying out normalization processing on each image to obtain normalized images; obtaining a time dimension vector of each pixel point according to all the normalized images; calculating a secondary rational kernel function result vector of the pixel point time dimension vector according to the time dimension vector and some preset parameters; obtaining a two-dimensional matrix according to the result vector of the secondary rational kernel function; respectively carrying out normalization processing on each element in the two-dimensional matrix according to the secondary rational kernel function result vectors of all the pixel points to obtain a kernel function normalization result; and detecting moving targets of the radar sequential images according to a preset judgment threshold value and a kernel function normalization result. The invention can obviously improve the detection performance of the moving target under the condition of low signal to noise ratio and improve the accuracy of the detection result.
Description
Technical Field
The invention relates to the technical field of signal processing, in particular to a radar sequential image moving target detection method and system.
Background
The imaging radar has the characteristics of being capable of working all the time and all the weather, has wide application fields, and can realize high-resolution imaging detection on the ground and imaging detection on a moving target. As imaging radar technology advances, the complexity of its detection of the type of target and the detection environment increases. Aiming at similar research background, a radar sequential image signal processing method is developed, radar data are processed into multi-frame sequential images similar to video, and then moving target detection is carried out according to multi-frame imaging results.
In conventional moving-target detection, an imaging radar generally acquires a moving-target image with a sufficient signal-to-noise ratio, and then utilizes a threshold detection technique to complete the moving-target detection. However, as the scattering characteristics of the targets become weaker and the radar detection environment becomes more and more targets, the moving targets tend to be submerged in radar image clutter and noise, so that the conventional threshold detection technology is difficult to meet the actual application requirements, and the obtained detection result is inaccurate.
Disclosure of Invention
The invention aims to provide a radar sequential image moving target detection method and a system, which can remarkably improve the detection performance of a moving target under low signal to noise ratio and improve the accuracy of a detection result.
In order to achieve the above object, the present invention provides the following solutions:
a radar sequential image moving target detection method, comprising:
acquiring related parameters of radar sequential image moving target detection and radar real data subjected to distance compression processing; the relevant parameters include: pulse repetition frequency and total duration of radar data;
calculating the output length of the sequential image kernel function according to the total duration, the preset single frame time length, the preset frame interval, the preset effective frame number estimated value of the target signal and the preset kernel function sliding window step length;
dividing the radar real data subjected to the distance compression processing along the azimuth direction according to the pulse repetition frequency, the preset single-frame time length and the preset frame interval to obtain multi-frame radar distance compressed data, and respectively performing fast Fourier transform on each frame of radar distance compressed data to obtain an image corresponding to each frame of radar distance compressed data;
respectively carrying out normalization processing on images corresponding to the radar distance compressed data of each frame to obtain normalized images;
for any pixel point, determining that pixel values of the pixel points on all normalized images form a time dimension vector of the pixel point;
calculating a secondary rational kernel function result vector of the pixel point time dimension vector according to the pixel point time dimension vector, the preset target signal occurrence effective frame number estimated value, the preset secondary rational kernel function parameter and the sequential image kernel function output length;
determining the minimum element in the secondary rational kernel function result vector of each pixel point to form a two-dimensional matrix;
respectively carrying out normalization processing on each element in the two-dimensional matrix according to a three-dimensional matrix formed by the secondary rational kernel function result vectors of all the pixel points to obtain a kernel function normalization result of each pixel point;
and detecting moving targets of the radar sequential images according to a preset judgment threshold value and a kernel function normalization result of each pixel point.
Optionally, the dividing the radar real data subjected to the distance compression processing along the azimuth direction according to the pulse repetition frequency, the preset single-frame time length and the preset frame interval to obtain multi-frame radar distance compressed data specifically includes:
calculating the number of single-frame azimuth points according to the pulse repetition frequency and the preset single-frame time length;
calculating the number of frame interval points according to the pulse repetition frequency and the preset frame interval;
and dividing the radar real data subjected to the distance compression processing along the azimuth direction according to the Shan Zhen azimuth point number and the frame interval point number to obtain multi-frame radar distance compression data.
Optionally, the normalizing processing is performed on the images corresponding to the radar distance compressed data of each frame to obtain normalized images, which specifically includes:
for an image corresponding to any frame of radar distance compressed data, respectively calculating the average value and standard deviation of the image;
and carrying out normalization processing on the image according to the average value and the standard deviation of the image to obtain a normalized image.
Optionally, the calculating the sequential image kernel function output length according to the total duration, the preset single frame time length, the preset frame interval, the preset effective frame number estimated value of the target signal and the preset kernel function sliding window step length specifically includes:
calculating the number of frames for converting the real data of the radar according to the total duration, the preset single frame time length and the preset frame interval;
and calculating the sequential image kernel function output length according to the effective frame number estimated value of the preset target signal, the preset kernel function sliding window step length and the frame number converted by the radar real data.
Optionally, the normalizing processing is performed on each element in the two-dimensional matrix according to the three-dimensional matrix formed by the secondary rational kernel function result vectors of all the pixel points to obtain a kernel function normalization result of each pixel point, which specifically includes:
intercepting a three-dimensional matrix formed by the secondary rational kernel function result vectors of all the pixel points along the azimuth direction and the distance direction to obtain a background noise kernel function result three-dimensional matrix;
calculating the mean value and standard deviation of the three-dimensional matrix of the background noise kernel function result;
and respectively carrying out normalization processing on each element in the two-dimensional matrix according to the mean value and standard deviation of the background noise kernel function result three-dimensional matrix to obtain a kernel function normalization result of each pixel point.
Optionally, the calculating the secondary rational kernel function result vector of the pixel point time dimension vector according to the pixel point time dimension vector, the preset target signal occurrence valid frame number estimated value, the preset secondary rational kernel function parameter and the sequential image kernel function output length specifically includes:
under the current iteration times, selecting two vectors with an interval of which the effective frame number estimated value of the preset target signal is generated on the time dimension vector of the pixel point as a first target vector and a second target vector;
obtaining a secondary rational kernel function result of a time dimension vector of the pixel point under the current iteration number according to the preset secondary rational kernel function parameter, the valid frame number estimated value of the preset target signal, the first target vector and the second target vector;
and updating the first target vector and the second target vector to enter the next iteration until the iteration times are equal to the output length of the sequential image kernel function, and determining the secondary rational kernel function result under all the iteration times as the secondary rational kernel function result vector of the pixel point time dimension vector.
A radar sequential image moving object detection system, comprising:
the acquisition module is used for acquiring related parameters of radar sequential image moving target detection and radar real data subjected to distance compression processing; the relevant parameters include: pulse repetition frequency and total duration of radar data;
the sequential image kernel function output length calculation module is used for calculating the sequential image kernel function output length according to the total time length, the preset single frame time length, the preset frame interval, the effective frame number estimated value of the preset target signal and the preset kernel function sliding window step length;
the segmentation module is used for segmenting the radar real data subjected to the distance compression processing along the azimuth direction according to the pulse repetition frequency, the preset single-frame time length and the preset frame interval to obtain multi-frame radar distance compressed data, and respectively carrying out fast Fourier transform on each frame of radar distance compressed data to obtain an image corresponding to each frame of radar distance compressed data;
the normalized image calculation module is used for respectively carrying out normalization processing on images corresponding to the radar distance compressed data of each frame to obtain normalized images;
the time dimension vector determining module is used for determining the time dimension vector of the pixel points, which is formed by the pixel values of the pixel points on all normalized images, for any one pixel point;
the secondary rational kernel function result vector calculation module is used for calculating a secondary rational kernel function result vector of the pixel point time dimension vector according to the pixel point time dimension vector, the preset target signal valid frame number estimated value, the preset secondary rational kernel function parameter and the sequential image kernel function output length;
the two-dimensional matrix determining module is used for determining that the minimum element in the secondary rational kernel function result vector of each pixel point forms a two-dimensional matrix;
the kernel function normalization module is used for respectively carrying out normalization processing on each element in the two-dimensional matrix according to a three-dimensional matrix formed by the secondary rational kernel function result vectors of all the pixel points to obtain a kernel function normalization result of each pixel point;
and the moving target detection module is used for detecting the moving target of the radar sequential image according to a preset judgment threshold value and a kernel function normalization result of each pixel point.
Optionally, the dividing module specifically includes:
the single-frame azimuth point calculation unit is used for calculating single-frame azimuth point according to the pulse repetition frequency and the preset single-frame time length;
a frame interval point number calculation unit for calculating a frame interval point number according to the pulse repetition frequency and the preset frame interval;
the segmentation unit is used for segmenting the radar real data subjected to the distance compression processing along the azimuth direction according to the Shan Zhen azimuth point number and the frame interval point number to obtain multi-frame radar distance compressed data.
Optionally, the normalized image calculation module specifically includes:
the average value and standard deviation calculation unit is used for calculating the average value and standard deviation of the image corresponding to the radar distance compressed data of any frame;
and the normalized image calculation unit is used for carrying out normalization processing on the image according to the average value and the standard deviation of the image to obtain a normalized image.
Optionally, the sequential image kernel function output length calculating module specifically includes:
the frame number calculating unit is used for calculating the frame number of the real data conversion of the radar according to the total duration, the preset single frame time length and the preset frame interval;
and the sequential image kernel function output length calculation unit is used for calculating the sequential image kernel function output length according to the effective frame number estimated value of the preset target signal, the preset kernel function sliding window step length and the frame number converted by the radar real data.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: according to the method, the sequential image kernel function output length is calculated according to the total duration, the preset single frame time length, the preset frame interval, the valid frame number estimated value of the preset target signal and the preset kernel function sliding window step length; dividing the radar real data subjected to distance compression processing along the azimuth direction according to the pulse repetition frequency, the preset single-frame time length and the preset frame interval to obtain multi-frame radar distance compressed data, and respectively performing fast Fourier transform on each frame of radar distance compressed data to obtain an image corresponding to each frame of radar distance compressed data; respectively carrying out normalization processing on images corresponding to the radar distance compressed data of each frame to obtain normalized images; for any pixel point, determining the time dimension vector of the pixel point formed by the pixel values of the pixel points on all normalized images; calculating a secondary rational kernel function result vector of the pixel point time dimension vector according to the pixel point time dimension vector and the sequential image kernel function output length; determining the minimum element in the secondary rational kernel function result vector of each pixel point as a two-dimensional matrix; respectively carrying out normalization processing on each element in the two-dimensional matrix according to a three-dimensional matrix formed by the secondary rational kernel function result vectors of all the pixel points to obtain a kernel function normalization result of each pixel point; and detecting the moving target of the radar sequential image according to a preset judgment threshold value and a kernel function normalization result of each pixel point, so that the detection performance of the moving target under the condition of low signal to noise ratio can be remarkably improved, and the accuracy of the detection result is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a method for detecting a moving target of a radar sequential image based on a quadratic rational kernel function according to an embodiment of the present invention;
FIG. 2 is a diagram of a result of a kernel function of a radar sequential image based on a quadratic rational kernel function according to an embodiment of the present invention;
fig. 3 is a diagram of a radar sequential image moving target detection result based on a quadratic rational kernel function according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
Background similarity, moving object difference and the like exist among multi-frame sequential radar image data, and the detection of the moving object can be facilitated by fully utilizing the information and mining the high-dimensional characteristics of the objects in the image data. Therefore, a new method for researching moving target detection based on multi-frame sequential radar image data is necessary, and the moving target detection performance under low signal-to-noise ratio can be remarkably improved.
As shown in fig. 1, the method includes:
a radar sequential image moving target detection method, comprising:
acquiring related parameters of radar sequential image moving target detection and radar real data subjected to distance compression processing; the relevant parameters include: pulse repetition frequency PRF and total duration T of radar data a 。
According to the total time length T a Preset single frame time length T fl Preset frame interval T fs Effective frame number estimated value L of preset target signal valid Presetting a kernel function sliding window step length N step Calculating the output length L of the kernel function of the sequential images ker 。
According to the pulse repetition frequency PRF and the preset single frame time length T fl And the preset frame interval T fs And dividing the radar real data subjected to the distance compression processing along the azimuth direction to obtain multi-frame radar distance compression data, and respectively performing fast Fourier transform on each frame of radar distance compression data to obtain an image corresponding to each frame of radar distance compression data.
And respectively carrying out normalization processing on the images corresponding to the radar distance compressed data of each frame to obtain normalized images.
For any one pixel point, determining that the pixel values of the pixel points on all normalized images form a time dimension vector of the pixel point.
According to the time dimension vector of the pixel point and the valid frame number estimated value L of the preset target signal valid Calculating a second rational kernel function result vector K of the pixel point time dimension vector by presetting a second rational kernel function parameter and the sequential image kernel function output length m,n 。
Determining the minimum element in the quadratic rational kernel function result vector of each pixel point to form a two-dimensional matrix K min 。
The two-dimensional matrix K is paired according to a three-dimensional matrix K consisting of secondary rational kernel function result vectors of all pixel points min And respectively carrying out normalization processing on each element in the array to obtain a kernel function normalization result of each pixel point.
And detecting moving targets of the radar sequential images according to a preset judgment threshold value and a kernel function normalization result of each pixel point.
In practical application, the pulse repetition frequency PRF and the preset single frame time length T are used for fl And the preset frame interval T fs Dividing the radar real data subjected to the distance compression processing along the azimuth direction to obtain multi-frame radar distance compression data, wherein the method specifically comprises the following steps of:
according to the pulse repetition frequency PRF and the preset single frame time length T fl Calculating the number N of single-frame azimuth points a 。
According to the pulse repetition frequency PRF and the preset frame interval T fs Calculating the number of frame interval points N fs 。
According to the Shan Zhen azimuth point number N a And the number of frame interval points N fs And dividing the radar real data subjected to the distance compression processing along the azimuth direction to obtain multi-frame radar distance compression data.
In practical application, the normalizing processing is performed on the images corresponding to the radar distance compressed data of each frame to obtain normalized images, which specifically includes:
and respectively calculating the average value and standard deviation of the images corresponding to the radar distance compressed data of any frame.
And carrying out normalization processing on the image according to the average value and the standard deviation of the image to obtain a normalized image.
In practical application, the method is based on the total time length T a Preset single frame time length T fl Preset frame interval T fs Effective frame number estimated value L of preset target signal valid Presetting a kernel function sliding window step length N step Calculating the output length L of the kernel function of the sequential images ker The method specifically comprises the following steps:
according to the total time length T a The preset single frame time length T fl And the preset frame interval T fs Calculating the frame number N of the conversion of the real data of the radar fra 。
Effective frame number estimation according to the preset target signalMetering value L valid The preset kernel function sliding window step length N step And the frame number N of the radar real data conversion fra Calculating the output length L of the sequential image kernel function ker 。
In practical application, the normalizing process is performed on each element in the two-dimensional matrix according to the three-dimensional matrix formed by the quadratic rational kernel function result vectors of all the pixel points to obtain the kernel function normalization result of each pixel point, which specifically includes:
intercepting a three-dimensional matrix K consisting of secondary rational kernel function result vectors of all pixel points along azimuth and distance directions to obtain a background noise kernel function result three-dimensional matrix K B 。
Calculating the three-dimensional matrix K of the background noise kernel function result B Mean B of (2) mean And standard deviation sigma B 。
According to the average value B of the three-dimensional matrix of the background noise kernel function result mean And standard deviation sigma B For the two-dimensional matrix K min And respectively carrying out normalization processing on each element in the array to obtain a kernel function normalization result of each pixel point.
In practical application, the effective frame number estimation value L is generated according to the time dimension vector of the pixel point and the preset target signal valid The method comprises the steps of presetting a secondary rational kernel function parameter and calculating a secondary rational kernel function result vector of the pixel point time dimension vector according to the sequential image kernel function output length, and specifically comprises the following steps:
and under the current iteration times, selecting two vectors with an interval of which the effective frame number estimated value is generated for the preset target signal on the time dimension vector of the pixel point as a first target vector and a second target vector.
Obtaining a secondary rational kernel function result of the time dimension vector of the pixel point under the current iteration number i according to the preset secondary rational kernel function parameter, the first target vector and the second target vector
Updating the first target vector and the second target vector to enter the next iteration until the iteration times are equal to the output length of the sequential image kernel function, and determining the secondary rational kernel function result under all the iteration times as a secondary rational kernel function result vector K of the pixel point (m, n) time dimension vector m,n 。
In practical application, according to the total time length T a The preset single frame time length T fl And the preset frame interval T fs Calculating the frame number N of the conversion of the real data of the radar fra Specifically, according to the formula
In practical application, according to the pulse repetition frequency PRF and the preset single frame time length T fl Calculating the number N of single-frame azimuth points a The method comprises the steps of carrying out a first treatment on the surface of the Specifically, according to the formula
N a ＝[PRF×T fl ](2) And (5) calculating.
In practical application, according to the pulse repetition frequency PRF and the preset frame interval T fs Calculating the number of frame interval points N fs Specifically, according to the formula
N fs ＝[PRF×T fs ](3) And (5) calculating.
In practical application, normalizing the image according to the average value u and standard deviation sigma to obtain a normalized image x nor Specifically, according to the formulaAnd (5) calculating.
In practical application, the effective frame number estimated value L is generated according to the preset target signal valid The preset kernel function sliding window step length N step And the frame number N of the radar real data conversion fra Calculating the output length L of the sequential image kernel function ker Specifically, according to the formula
In practical application, the effective frame number estimated value L is generated according to the preset secondary rational kernel function parameter and the preset target signal valid Obtaining a second rational kernel function result of the time dimension vector of the pixel point under the current iteration number i by the first target vector and the second target vectorThe method specifically comprises the following steps:
(a) According to the formula (6), the preset quadratic rational kernel function parameter and the first target vector X i And the second target vector Y i Obtaining a secondary rational kernel function result of a certain pixel point time dimension vector under the current iteration number i
Wherein,for X in a time-dimensional vector according to pixel (m, n) i And Y i And (3) vector calculation to obtain a secondary rational kernel function result of the time dimension vector of the ith pixel point (m, n). m and n are pixel point coordinates, X i ，Y i The interval of the time dimension vector formed for the pixel point is L valid I is the kernel result sequence number, x of the pixel point output j ，y j Represents X i ，Y i Corresponding elements in the two vectors, j representing the sequence number of the element in the vector.
In practical application, updating the first target vector and the second target vector to enter the next iteration until the iteration times are equal to the output length of the sequential image kernel function, and determining allThe result of the quadratic rational kernel function under the iteration times is a result vector K of the quadratic rational kernel function of the pixel point (m, n) time dimension vector m,n The method comprises the following steps: updating the loop variable i to make i=i+1 until the output kernel function result sequence number i=l ker Obtaining a second rational kernel function result vector K of the pixel point time dimension vector m,n 。
In practical application, determining that the minimum element in the quadratic rational kernel function result vector of each pixel point forms a two-dimensional matrix to obtain a two-dimensional matrix K min Specifically according to the formula
In practical application, a three-dimensional matrix K composed of the result vectors of the quadratic rational kernel function of all pixel points is intercepted along the azimuth direction and the distance direction to obtain a three-dimensional matrix K of the result of the background noise kernel function B The method specifically comprises the following steps: combining the kernel function result three-dimensional matrix K of the time dimension vectors of all pixel points, and intercepting part of edge image kernel function results along azimuth and distance directions to serve as a background noise kernel function result three-dimensional matrix K B The specific formula is as follows:
K B ＝K M,N (8)
and M and N are azimuth point numbers and distance point numbers of the intercepted part of the edge image.
In practical application, the average value B of the three-dimensional matrix is obtained according to the background noise kernel function result mean And standard deviation sigma B For the two-dimensional matrix K min Each element of (a)Respectively carrying out normalization processing to obtain kernel function normalization results of all pixel points>In particular according to the formula
And (5) calculating.
Where m, n are pixel coordinates.
Normalizing the result according to a preset judgment threshold THR and a kernel function of each pixel pointAnd detecting a moving target of the radar sequential image, wherein the specific operation flow is as follows:
(a) Judging the kernel function normalization result of a certain pixel point (m, n)And the magnitude of the decision threshold THR, ifJudging that the pixel point has no moving object passing through, if +.>And judging that the pixel point has a moving object passing through, wherein m and n are pixel point coordinates.
(b) And (c) repeating the operation (a) of the step, updating the variables m and n to obtain moving target detection results of all pixel points, and directly drawing a radar sequential image moving target detection result graph.
A radar sequential image moving object detection system, comprising:
the acquisition module is used for acquiring related parameters of radar sequential image moving target detection and radar real data subjected to distance compression processing; the relevant parameters include: pulse repetition frequency PRF and total duration T of radar data a 。
A sequential image kernel function output length calculation module for calculating the total time length T a Preset single frame time length T fl Preset frame interval T fs Effective frame number estimated value L of preset target signal valid Presetting a kernel function sliding window step length N step Calculating the output length L of the kernel function of the sequential images ker 。
A dividing module for dividing the preset single frame time length T according to the pulse repetition frequency PRF fl And the preset frame interval T fs And dividing the radar real data subjected to the distance compression processing along the azimuth direction to obtain multi-frame radar distance compression data, and respectively performing fast Fourier transform on each frame of radar distance compression data to obtain an image corresponding to each frame of radar distance compression data.
And the normalized image calculation module is used for respectively carrying out normalization processing on the images corresponding to the radar distance compressed data of each frame to obtain normalized images.
And the time dimension vector determining module is used for determining the pixel values of the pixel points on all the normalized images to form the time dimension vector of the pixel points for any one pixel point.
The second rational kernel function result vector calculation module is used for generating an effective frame number estimated value L according to the time dimension vector of the pixel point and the preset target signal valid And presetting a secondary rational kernel function parameter and the sequential image kernel function output length to calculate a secondary rational kernel function result vector of the pixel point time dimension vector.
A two-dimensional matrix determining module for determining that the minimum element in the quadratic rational kernel function result vector of each pixel point forms a two-dimensional matrix K min 。
And the kernel function normalization module is used for respectively carrying out normalization processing on each element in the two-dimensional matrix according to a three-dimensional matrix formed by the secondary rational kernel function result vectors of all the pixel points to obtain a kernel function normalization result of each pixel point.
And the moving target detection module is used for detecting the moving target of the radar sequential image according to a preset judgment threshold value and a kernel function normalization result of each pixel point.
In practical application, the segmentation module specifically includes:
a single-frame azimuth point number calculation unit for calculating the pulse repetition frequency PRF and the preset single-frame time length T fl Calculating the number N of single-frame azimuth points a 。
A frame interval point number calculation unit for calculating a frame interval according to the pulse repetition frequency PRF and the preset frame interval T fs Calculating the number of frame interval points N fs 。
A dividing unit for counting N according to the Shan Zhen azimuth a And the number of frame interval points N fs And dividing the radar real data subjected to the distance compression processing along the azimuth direction to obtain multi-frame radar distance compression data.
In practical application, the normalized image calculation module specifically includes:
and the average value and standard deviation calculation unit is used for calculating the average value and standard deviation of the image corresponding to the radar distance compressed data of any frame.
And the normalized image calculation unit is used for carrying out normalization processing on the image according to the average value and the standard deviation of the image to obtain a normalized image.
In practical application, the sequential image kernel function output length calculation module specifically includes:
a frame number calculating unit for calculating a total time length T a The preset single frame time length T fl And the preset frame interval T fs Calculating the frame number N of the conversion of the real data of the radar fra 。
A sequential image kernel function output length calculation unit for generating an effective frame number estimation value L according to the preset target signal valid The preset kernel function sliding window step length N step And the frame number N of the radar real data conversion fra Calculating the output length L of the sequential image kernel function ker 。
The present embodiment proposes a simulation process using the radar sequential image moving target detection method based on the quadratic rational kernel function provided in the foregoing embodiment, where the required parameters are shown in table 1.
Table 1 example parameters
The embodiment specifically comprises the following steps:
step one: reading radar sequential image moving target detection related parameters and radar real data subjected to distance compression processing, wherein the related parameters specifically comprise: pulse repetition frequency PRF, total duration T a The preset simulation parameters specifically comprise: preset single frame time length T of frames of the generated radar sequential image fl Presetting a frame interval T fs Presetting a kernel function sliding window step length N step Presetting a secondary rational kernel function parameter C, and presetting an effective frame number estimated value L of a target signal valid And a preset decision threshold THR.
Step two: combining the total time length T read in step one a Presetting a single frame time length T fl Presetting a frame interval T fs Calculating the frame number N of the radar real data conversion by using the formula (1) fra 。
Step three: presetting a single frame time length T by combining the pulse repetition frequency PRF read in the step one fl Presetting a frame interval T fs And radar real data subjected to distance compression processing, and frame number N calculated in the step two fra Generating a subframe of radar real data subjected to distance compression processing, and performing fast Fourier transform calculation on each frame of data to obtain an imaging result of each frame of radar data, wherein the specific operation flow is as follows.
(a) Calculating the number N of single-frame azimuth points by using the method (2) a 。
(b) Calculating the number of frame interval points N using (3) fs 。
(c) Combining single frame azimuth point number N a And number of frame interval points N fs And dividing the radar real data subjected to the distance compression processing along the azimuth direction to generate subframes of radar distance compressed data.
(d) And performing fast Fourier transform calculation on the radar distance compressed data of each frame to obtain a radar data imaging result of each frame.
Step four: and (3) calculating a normalization result of each frame of image by combining the radar data imaging result of each frame calculated in the step (III), wherein the specific operation flow is as follows:
(a) An average value u of the image data of each frame is calculated.
(b) The standard deviation σ of the image data per frame is calculated.
(c) And calculating the normalization result of each frame of image by using the formula (4).
Step five: combining the preset target signal read in the first step to generate an effective frame number estimated value L valid Presetting a kernel function sliding window step length N step And the number of frames N calculated in the second step fra Calculating the sequential image kernel output length L using equation (5) ker 。
Step six: and (3) extracting the values of the same pixel point of each frame to form a time dimension vector according to the normalization result of each frame of image calculated in the step (IV), namely extracting the values of the same pixel point on each frame of image to form the time dimension vector of the pixel point.
Step seven: combining the preset secondary rational kernel function parameter C read in the step one, and presetting the kernel function sliding window step length N step Preset target signal occurrence effective frame number estimated value L valid And step five, calculating the output length L of the sequential image kernel function ker The extracted values of the same pixel points of each frame form a time dimension vector, a kernel function result matrix K of the time dimension vector of all the pixel points and a kernel function result minimum value K of the time dimension vector of each pixel point are calculated min The specific operation flow is as follows:
(a) Calculating the ith secondary rational function result of a pixel point time dimension vector by using the method (6)
(b) Updating the loop variable i to make i=i+1 until the output kernel function result sequence number i=l ker Obtaining a second rational kernel function result vector K of the pixel point time dimension vector m,n (at all times)Constituted by.
(c) Repeating the operations (a) and (b) of the step, updating the variables m and n, and obtaining a secondary rational kernel function result three-dimensional matrix K of all pixel point time dimension vectors, namely, the secondary rational kernel function result vector composition K of all pixel points.
(d) Calculating the minimum value of the quadratic rational function result matrix K of the time dimension vector of all pixel points in the time dimension by using the method (7) to obtain a two-dimensional matrix K min One pixel point corresponds to a minimum value, and then the minimum values of all the pixel points form a two-dimensional matrix K min The dimensions are distance and azimuth, respectively.
Step eight: calculating the three-dimensional matrix K of the background noise kernel function result by using the formula (8) in combination with the three-dimensional matrix K of the kernel function result of the time-dimensional vectors of all the pixel points calculated in the step seven B 。
Step nine: combining the background noise kernel function result three-dimensional matrix K extracted in the step eight B Calculating a three-dimensional matrix K of background noise kernel function results B Mean B of (2) mean And standard deviation sigma B 。
Step ten: combining the least value K of the kernel function result of the time dimension vector of each pixel point calculated in the step seven min And step nine, calculating the mean value B of the background noise kernel function result three-dimensional matrix mean And standard deviation sigma B Calculating the normalization result of each kernel function by using the formula (9) to obtain a plurality of normalization results K nor 。
Step eleven: combining the judgment threshold THR read in the step one and the kernel function normalization result K calculated in the step ten nor By comparing K nor And THR to obtain the detection result of the radar sequential image moving target.
Through the processing of the above steps, the kernel function calculation result of the radar sequential image obtained by combining the parameters of table 1 and the radar real data subjected to the distance compression processing can be obtained, as shown in fig. 2, and the radar sequential image moving target detection result (white position) is shown in fig. 3. As can be seen from the results of fig. 2 and fig. 3, the method provided by the present invention can distinguish the moving object from the original data of the radar including noise by adding the time dimension and calculating the kernel function value of each pixel point of the sequential image of the radar, and the result is clear and discernable.
The invention has the advantages that:
(1) The invention provides a radar sequential image moving target detection method based on a secondary rational kernel function, which can realize effective moving target detection under the condition of low signal-to-noise ratio compared with the existing radar moving target detection method.
(2) The invention provides a radar sequential image moving target detection method based on a secondary rational kernel function, which increases the correlation calculation of time dimension, and utilizes the secondary rational kernel function to map an input space into a high-dimensional feature space, thereby reducing the classification difficulty and improving the moving target detection precision.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present invention and the core ideas thereof; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (10)
1. A radar sequential image moving target detection method, characterized by comprising:
acquiring related parameters of radar sequential image moving target detection and radar real data subjected to distance compression processing; the relevant parameters include: pulse repetition frequency and total duration of radar data;
calculating the output length of the sequential image kernel function according to the total duration, the preset single frame time length, the preset frame interval, the preset effective frame number estimated value of the target signal and the preset kernel function sliding window step length;
dividing the radar real data subjected to the distance compression processing along the azimuth direction according to the pulse repetition frequency, the preset single-frame time length and the preset frame interval to obtain multi-frame radar distance compressed data, and respectively performing fast Fourier transform on each frame of radar distance compressed data to obtain an image corresponding to each frame of radar distance compressed data;
respectively carrying out normalization processing on images corresponding to the radar distance compressed data of each frame to obtain normalized images;
for any pixel point, determining that pixel values of the pixel points on all normalized images form a time dimension vector of the pixel point;
calculating a secondary rational kernel function result vector of the pixel point time dimension vector according to the pixel point time dimension vector, the preset target signal occurrence effective frame number estimated value, the preset secondary rational kernel function parameter and the sequential image kernel function output length;
determining the minimum element in the secondary rational kernel function result vector of each pixel point to form a two-dimensional matrix;
respectively carrying out normalization processing on each element in the two-dimensional matrix according to a three-dimensional matrix formed by the secondary rational kernel function result vectors of all the pixel points to obtain a kernel function normalization result of each pixel point;
and detecting moving targets of the radar sequential images according to a preset judgment threshold value and a kernel function normalization result of each pixel point.
2. The method for detecting a moving target of a radar sequential image according to claim 1, wherein the dividing the radar real data subjected to the distance compression processing along the azimuth direction according to the pulse repetition frequency, the preset single-frame time length and the preset frame interval to obtain multi-frame radar distance compressed data specifically comprises:
calculating the number of single-frame azimuth points according to the pulse repetition frequency and the preset single-frame time length;
calculating the number of frame interval points according to the pulse repetition frequency and the preset frame interval;
and dividing the radar real data subjected to the distance compression processing along the azimuth direction according to the Shan Zhen azimuth point number and the frame interval point number to obtain multi-frame radar distance compression data.
3. The method for detecting the moving target of the radar sequential image according to claim 1, wherein the normalizing processing is performed on the images corresponding to the radar distance compressed data of each frame to obtain normalized images, specifically comprising:
for an image corresponding to any frame of radar distance compressed data, respectively calculating the average value and standard deviation of the image;
and carrying out normalization processing on the image according to the average value and the standard deviation of the image to obtain a normalized image.
4. The method for detecting the moving target of the sequential image of the radar according to claim 1, wherein the calculating the output length of the sequential image kernel function according to the total duration, the preset single frame time length, the preset frame interval, the estimated value of the effective frame number of the preset target signal and the preset kernel function sliding window step length specifically comprises:
calculating the number of frames for converting the real data of the radar according to the total duration, the preset single frame time length and the preset frame interval;
and calculating the sequential image kernel function output length according to the effective frame number estimated value of the preset target signal, the preset kernel function sliding window step length and the frame number converted by the radar real data.
5. The method for detecting the moving target of the radar sequential image according to claim 1, wherein the normalizing processing is performed on each element in the two-dimensional matrix according to a three-dimensional matrix formed by quadratic rational kernel function result vectors of all pixel points to obtain a kernel function normalization result of each pixel point, specifically comprising:
intercepting a three-dimensional matrix formed by the secondary rational kernel function result vectors of all the pixel points along the azimuth direction and the distance direction to obtain a background noise kernel function result three-dimensional matrix;
calculating the mean value and standard deviation of the three-dimensional matrix of the background noise kernel function result;
and respectively carrying out normalization processing on each element in the two-dimensional matrix according to the mean value and standard deviation of the background noise kernel function result three-dimensional matrix to obtain a kernel function normalization result of each pixel point.
6. The method for detecting a moving target of a radar sequential image according to claim 1, wherein the calculating a second order rational kernel result vector of the pixel time dimension vector according to the pixel time dimension vector, the estimated value of the number of valid frames of the preset target signal, a preset second order rational kernel parameter and the sequential image kernel output length specifically comprises:
under the current iteration times, selecting two vectors with an interval of which the effective frame number estimated value of the preset target signal is generated on the time dimension vector of the pixel point as a first target vector and a second target vector;
obtaining a secondary rational kernel function result of a time dimension vector of the pixel point under the current iteration number according to the preset secondary rational kernel function parameter, the valid frame number estimated value of the preset target signal, the first target vector and the second target vector;
and updating the first target vector and the second target vector to enter the next iteration until the iteration times are equal to the output length of the sequential image kernel function, and determining the secondary rational kernel function result under all the iteration times as the secondary rational kernel function result vector of the pixel point time dimension vector.
7. A radar sequential image moving object detection system, characterized by comprising:
the acquisition module is used for acquiring related parameters of radar sequential image moving target detection and radar real data subjected to distance compression processing; the relevant parameters include: pulse repetition frequency and total duration of radar data;
the sequential image kernel function output length calculation module is used for calculating the sequential image kernel function output length according to the total time length, the preset single frame time length, the preset frame interval, the effective frame number estimated value of the preset target signal and the preset kernel function sliding window step length;
the segmentation module is used for segmenting the radar real data subjected to the distance compression processing along the azimuth direction according to the pulse repetition frequency, the preset single-frame time length and the preset frame interval to obtain multi-frame radar distance compressed data, and respectively carrying out fast Fourier transform on each frame of radar distance compressed data to obtain an image corresponding to each frame of radar distance compressed data;
the normalized image calculation module is used for respectively carrying out normalization processing on images corresponding to the radar distance compressed data of each frame to obtain normalized images;
the time dimension vector determining module is used for determining the time dimension vector of the pixel points, which is formed by the pixel values of the pixel points on all normalized images, for any one pixel point;
the secondary rational kernel function result vector calculation module is used for calculating a secondary rational kernel function result vector of the pixel point time dimension vector according to the pixel point time dimension vector, the preset target signal valid frame number estimated value, the preset secondary rational kernel function parameter and the sequential image kernel function output length;
the two-dimensional matrix determining module is used for determining that the minimum element in the secondary rational kernel function result vector of each pixel point forms a two-dimensional matrix;
the kernel function normalization module is used for respectively carrying out normalization processing on each element in the two-dimensional matrix according to a three-dimensional matrix formed by the secondary rational kernel function result vectors of all the pixel points to obtain a kernel function normalization result of each pixel point;
and the moving target detection module is used for detecting the moving target of the radar sequential image according to a preset judgment threshold value and a kernel function normalization result of each pixel point.
8. The radar sequential image moving object detection system according to claim 7, wherein said segmentation module specifically comprises:
the single-frame azimuth point calculation unit is used for calculating single-frame azimuth point according to the pulse repetition frequency and the preset single-frame time length;
a frame interval point number calculation unit for calculating a frame interval point number according to the pulse repetition frequency and the preset frame interval;
the segmentation unit is used for segmenting the radar real data subjected to the distance compression processing along the azimuth direction according to the Shan Zhen azimuth point number and the frame interval point number to obtain multi-frame radar distance compressed data.
9. The radar sequential image moving object detection system according to claim 7, wherein said normalized image calculation module specifically comprises:
the average value and standard deviation calculation unit is used for calculating the average value and standard deviation of the image corresponding to the radar distance compressed data of any frame;
and the normalized image calculation unit is used for carrying out normalization processing on the image according to the average value and the standard deviation of the image to obtain a normalized image.
10. The radar sequential image moving object detection system according to claim 7, wherein said sequential image kernel output length calculation module specifically comprises:
the frame number calculating unit is used for calculating the frame number of the real data conversion of the radar according to the total duration, the preset single frame time length and the preset frame interval;
and the sequential image kernel function output length calculation unit is used for calculating the sequential image kernel function output length according to the effective frame number estimated value of the preset target signal, the preset kernel function sliding window step length and the frame number converted by the radar real data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210626011.5A CN115015903B (en) | 2022-06-02 | 2022-06-02 | Radar sequential image moving target detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210626011.5A CN115015903B (en) | 2022-06-02 | 2022-06-02 | Radar sequential image moving target detection method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115015903A CN115015903A (en) | 2022-09-06 |
| CN115015903B true CN115015903B (en) | 2024-04-09 |
Family
ID=83072631
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210626011.5A Active CN115015903B (en) | 2022-06-02 | 2022-06-02 | Radar sequential image moving target detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115015903B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108710126A (en) * | 2018-03-14 | 2018-10-26 | 上海鹰觉科技有限公司 | Automation detection expulsion goal approach and its system |
| CN111781590A (en) * | 2020-05-26 | 2020-10-16 | 中国空间技术研究院 | Radar target parameter estimation method based on efficient FRFT |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10852419B2 (en) * | 2017-10-20 | 2020-12-01 | Texas Instruments Incorporated | System and method for camera radar fusion |
-
2022
- 2022-06-02 CN CN202210626011.5A patent/CN115015903B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108710126A (en) * | 2018-03-14 | 2018-10-26 | 上海鹰觉科技有限公司 | Automation detection expulsion goal approach and its system |
| CN111781590A (en) * | 2020-05-26 | 2020-10-16 | 中国空间技术研究院 | Radar target parameter estimation method based on efficient FRFT |
Non-Patent Citations (2)
| Title |
|---|
| Zhou, Xinka等.A Modified Radon Fourier Transform for GNSS-Based Bistatic Radar Target Detection.IEEE GEOSCIENCE AND REMOTE SENSING LETTERS.2020,全文. * |
| 朱岱寅 ; 杨鸣冬 ; 宋伟 ; 叶少华 ; .高分辨率极化合成孔径雷达成像研究进展.数据采集与处理.2016,(第04期),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115015903A (en) | 2022-09-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107229918B (en) | SAR image target detection method based on full convolution neural network | |
| CN110109060B (en) | Radar radiation source signal sorting method and system based on deep learning network | |
| CN104851097B (en) | The multichannel SAR GMTI methods aided in based on target shape and shade | |
| CN107808138B (en) | Communication signal identification method based on FasterR-CNN | |
| CN110287819A (en) | Moving target detection method under dynamic background based on low-rank and sparse decomposition | |
| CN110378924B (en) | Level set image segmentation method based on local entropy | |
| CN116012364B (en) | SAR image change detection method and device | |
| CN113536963B (en) | SAR image airplane target detection method based on lightweight YOLO network | |
| CN110827262B (en) | Weak and small target detection method based on continuous limited frame infrared image | |
| CN111311644B (en) | Moving target detection method based on video SAR | |
| CN112986950A (en) | Single-pulse laser radar echo feature extraction method based on deep learning | |
| CN113222824B (en) | Infrared image super-resolution and small target detection method | |
| CN103077499B (en) | SAR (Synthetic Aperture Radar) image pre-processing method based on similar block | |
| CN103093432B (en) | Polarized synthetic aperture radar (SAR) image speckle reduction method based on polarization decomposition and image block similarity | |
| Wang et al. | A novel and high-speed local contrast method for infrared small-target detection | |
| CN111915558A (en) | Pin state detection method for high-voltage transmission line | |
| CN114779185A (en) | Radar signal anti-interference low-loss recovery method of coding and decoding convolutional neural network | |
| CN115015903B (en) | Radar sequential image moving target detection method and system | |
| CN109285148B (en) | Infrared weak and small target detection method based on heavily weighted low rank and enhanced sparsity | |
| CN111783656A (en) | Self-adaptive infrared small target detection method based on statistical characteristic pre-segmentation | |
| CN113822361A (en) | SAR image similarity measurement method and system based on Hamming distance | |
| Wu et al. | Object-Enhanced YOLO Networks for Synthetic Aperture Radar Ship Detection | |
| CN114283296B (en) | Target identification self-evaluation method based on structural similarity | |
| CN113159157B (en) | Improved low-frequency UWB SAR leaf cluster hidden target fusion change detection method | |
| Sun et al. | Image segmentation algorithm based on top-hat transformation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |