CN111323757B - Target detection method and device for marine radar - Google Patents
Target detection method and device for marine radar Download PDFInfo
- Publication number
- CN111323757B CN111323757B CN202010208714.7A CN202010208714A CN111323757B CN 111323757 B CN111323757 B CN 111323757B CN 202010208714 A CN202010208714 A CN 202010208714A CN 111323757 B CN111323757 B CN 111323757B
- Authority
- CN
- China
- Prior art keywords
- target
- information
- processing
- radar
- digital signal
- 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
Images
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
- 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
- G01S7/28—Details of pulse systems
- G01S7/285—Receivers
- G01S7/292—Extracting wanted echo-signals
-
- 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
- G01S7/28—Details of pulse systems
- G01S7/285—Receivers
- G01S7/295—Means for transforming co-ordinates or for evaluating data, e.g. using computers
-
- 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
- G01S7/35—Details of non-pulse systems
- G01S7/352—Receivers
- G01S7/354—Extracting wanted echo-signals
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The invention provides a method and a device for detecting a marine radar target, wherein the method comprises the following steps: acquiring radar original data information acquired by a radar receiver; performing analog-to-digital conversion processing on the radar original data information to obtain a digital signal; detecting and processing the digital signal according to a first filtering algorithm to determine a target; when the first filtering algorithm does not detect the target, detecting the digital signal according to a deep learning algorithm to determine the target; adjusting a threshold in the first filtering algorithm according to the information of the target; and processing the digital signal according to the first filtering algorithm of the adjusted threshold, and outputting the detected target. The invention provides more accurate target information for users, thereby providing reliable reference basis for navigation, collision avoidance and target monitoring.
Description
Technical Field
The invention relates to the technical field of radars, in particular to a method and a device for detecting a target of a marine radar.
Background
Radars are widely used in the marine field, and for users they correspond to their "eyes". The large ship needs to be equipped with two radars, one is an S-band radar working at 3GHz, the other is an X-band radar working at 9GHz, and the early warning and navigation functions are realized by performing long-distance detection on the X-band; and short-distance measurement is carried out through the S wave band, and the navigation and collision avoidance functions are realized.
In the prior art, a marine radar detects a target according to a traditional threshold method of echo quality (size, roundness and the like), and the condition that the target is not detected but the echo exists affects the accuracy of radar target detection, so that accurate and reliable target information cannot be accurately provided for a user.
Disclosure of Invention
The invention provides a method and a device for detecting a target of a marine radar, which solve the problem of poor accuracy of radar target detection.
In order to solve the above technical problem, an embodiment of the present invention provides the following technical solutions:
a marine radar target detection method, comprising:
acquiring radar original data information acquired by a radar receiver;
performing analog-to-digital conversion processing on the radar original data information to obtain a digital signal;
detecting and processing the digital signal according to a first filtering algorithm to determine a target;
when the first filtering algorithm does not detect the target, detecting the digital signal according to a deep learning algorithm to determine the target;
adjusting a threshold in the first filtering algorithm according to the information of the target;
and processing the digital signal according to the first filtering algorithm of the adjusted threshold, and outputting the detected target.
Optionally, the detecting and processing the digital signal according to a deep learning algorithm to determine the target includes:
carrying out format conversion on the digital signal to obtain first image information;
processing the first image information to obtain original echo image information;
processing the original echo image to obtain second image information;
and inputting the second image information into a deep learning model for processing to obtain an image detection result and determine a target.
Optionally, performing format conversion on the digital signal to obtain first image information, including:
and carrying out format conversion on the digital signal according to a fixed format of radar signal processing to obtain the first image information.
Optionally, processing the first image information to obtain original echo image information includes:
and processing the pixel value information of the first image information to obtain the original echo image information.
Optionally, processing the original echo image to obtain second image information includes: and filtering the original echo image by using a preset filtering algorithm to obtain second image information.
Optionally, inputting the second image information into a deep learning model for processing, obtaining an image detection result, and determining a target, including:
dividing the second image into a plurality of k × k grids;
determining that a target in a grid is detected when a target center falls on the grid;
using a minimum rectangular frame to frame the target object;
and calculating the coordinates of the central point of the frame of the target, the width of the frame, the height and the confidence coefficient of the frame to obtain the image detection result, and extracting target information.
Optionally, the method for detecting a target of a marine radar further includes: and if the image detection result contains N frames of images, wherein M frames of images are matched with the target, determining that the target is accurate, wherein M and N are positive integers, and M is greater than N/2.
Optionally, the method for detecting a target of a marine radar further includes: identifying the target by using a circumscribed rectangle frame;
and taking the intersection point of the diagonal lines of the rectangular frame as the coordinate information of the target, and converting the rectangular coordinate and the polar coordinate to obtain the azimuth and distance information of the target.
Optionally, processing the digital signal according to the first filtering algorithm of the adjusted threshold, and outputting a detected target, includes: and detecting the data signal according to the detection threshold after at least one stage of adjustment to determine a target.
Optionally, the method for detecting a target of a marine radar further includes: and if the antenna of the radar scans the X-circle image, and the Y-circle image is matched with the target, determining that the target is a real target, and outputting the target, wherein X and Y are positive integers, and Y is more than X/2.
Optionally, the method for detecting a target of a marine radar further includes: and tracking the information of the target obtained according to the first filtering algorithm of the adjusted threshold, and displaying the target.
Optionally, the method for detecting a target of a marine radar further includes: and displaying the output target.
An embodiment of the present invention further provides a marine radar target detection device, including:
the acquisition module is used for acquiring radar original data information acquired by the radar receiver;
the processing module is used for carrying out analog-to-digital conversion processing on the radar original data information to obtain a digital signal; detecting and processing the digital signal according to a first filtering algorithm to determine a target; when the first filtering algorithm does not detect the target, detecting the digital signal according to a deep learning algorithm to determine the target; adjusting a threshold in the first filtering algorithm according to the information of the target; and processing the digital signal according to the first filtering algorithm of the adjusted threshold, and outputting the detected target.
The technical scheme of the invention has the beneficial effects that: the method has the advantages that the radar target is detected through the deep learning model, the method is suitable for the marine radar with different wave bands, target information is judged without or by assisting the size threshold value of radar echo, more accurate target information is provided for users, and reliable reference basis is provided for navigation and collision avoidance.
Drawings
FIG. 1 shows a schematic flow diagram of a method of marine radar target detection according to an embodiment of the invention;
FIG. 2 is a flow chart illustrating an implementation of a method for detecting a target in a marine radar according to an embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating an algorithm flow of a deep learning model according to an embodiment of the present invention;
FIG. 4 is a functional diagram of a display module according to an embodiment of the invention;
FIG. 5 shows a functional schematic of an alarm module in an embodiment of the invention;
FIG. 6 is a functional diagram of a data management module in an embodiment of the invention;
fig. 7 is a schematic block diagram of a deep learning-based marine radar target detection apparatus according to an embodiment of the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments.
As shown in fig. 1, an embodiment of the present invention provides a method for detecting a marine radar target, which specifically includes the following steps:
step 11: and acquiring radar original data information acquired by a radar receiver.
Step 12: and performing analog-to-digital conversion processing on the radar original data information to obtain a digital signal.
Step 13: detecting and processing the digital signal according to a first filtering algorithm to determine a target;
step 14: when the first filtering algorithm does not detect the target, detecting the digital signal according to a deep learning algorithm to determine the target;
step 15: adjusting a threshold in the first filtering algorithm according to the information of the target;
step 16: and processing the digital signal according to the first filtering algorithm of the adjusted threshold, and outputting the detected target.
In an optional embodiment of the present invention, in step 11, the radar receiver may be an X-band radar receiver, an S-band radar receiver, or the like. The navigation radar can be a ship radar or a shore-based radar. It should be noted that the application scenarios and bands of the radar receiver are not limited by the implementation of the present invention, and the above is only an exemplary illustration.
In an optional embodiment of the present invention, in step 12, the radar receiver receives radar raw data information, performs AD (analog-to-digital) conversion on the radar raw data information, converts the radar raw data information into a discrete digital signal, and performs digital signal processing.
In an optional embodiment of the present invention, in step 13, the digital signal is detected according to a first filtering algorithm to determine a target, and if the target is detected, the target may be directly output;
in an optional embodiment of the present invention, the step 14 may include:
step 141, performing format conversion on the digital signal to obtain first image information; specifically, the format conversion may be performed on the digital signal according to a fixed format of radar signal processing, so as to obtain the first image information.
Step 142, processing the first image information to obtain original echo image information; specifically, the pixel value information of the first image information may be processed to obtain the original echo image information.
Step 143, processing the original echo image to obtain second image information; specifically, a preset filtering algorithm may be used to perform filtering processing on the original echo image to obtain second image information. The preset filtering algorithm herein includes but is not limited to: median filtering algorithm, mean filtering, wavelet transform, etc. In the step, the original echo image is filtered to remove certain clutter noise, and then the filtered original echo image can be input into a deep learning model for further processing and can also be input into a display module for displaying.
And step 134, inputting the second image information into a deep learning model for processing to obtain an image detection result, and determining a target. The deep learning model refers to a model which is trained by a plurality of groups of marked original echo images through a deep learning algorithm, and the format of the model can be a caffemodel or a model using other frames; the deep learning method adopts supervised learning, and can adopt open-source Yolo v3, but is not limited to other open-source target detection algorithms, such as Fast RCNN, SSD, Retina-Net, Anchor-Free and the like. The deep learning model has a self-learning function, and is continuously trained along with the increase of target data, so that the accuracy and generalization capability of the deep learning model are improved.
In an alternative embodiment of the present invention, step 134 may include:
step 1341, divide the second image into a plurality of k × k grids;
step 1342, when the target center falls on a certain grid, determining that the target in the grid is detected;
step 1343, framing the target object with a minimum rectangular frame;
and step 1344, calculating the coordinates of the center point of the frame of the target, the width of the frame, the height of the frame and the confidence coefficient to obtain the image detection result, and extracting target information.
In this embodiment, when the deep learning model detects a radar echo image, the image is divided into a plurality of k × k grids. When the target center falls on a certain grid, namely a target in the network is detected, the target is drawn by a minimum rectangular frame. And calculating the coordinates, width, height and confidence coefficient of the central point of the target frame, and finally extracting target information. If the target center does not fall on a certain network, the target is not detected, and the network finishes the detection. Like the model is similar to a black box, as the input samples increase, the model is continuously self-learned, namely, the network works and learns at the same time, so that the generalization capability and the target detection capability of the model are increased. In addition, the confidence degree of the output target can be adjusted according to the intensity of noise such as sea clutter, rain clutter and the like, the output of the target is further optimized, and the accuracy of target detection is improved.
In an optional embodiment of the present invention, after step 134, the method for detecting a marine radar target based on deep learning may further include:
step 1340, if the image detection result includes N frames of images, wherein M frames of images are matched with the target, determining that the target is accurate, wherein M and N are positive integers, and M is greater than N/2.
In this embodiment, the result after the deep learning detection is subjected to multi-frame image matching, for example, if 3 or more than 3 of 4 frames correspond to the target, the target is considered to be the target, otherwise, the target is noise. The embodiment further filters the interference of random noise and rain and snow noise, and reduces the false detection of the target.
In an optional embodiment of the present invention, after step 135, the method for detecting a marine radar target based on deep learning may further include:
and step 136, identifying the target by using a circumscribed rectangle frame.
In the embodiment, the target information detected by the deep learning is extracted, and the target is identified in the form of a circumscribed rectangle. In order to obtain more accurate target information, coordinate information in which the intersection point (i.e., the center point) of the diagonal lines of the rectangular frame is the target can be obtained.
In an optional embodiment of the present invention, after step 136, the method for detecting a target of a marine radar may further include:
and 137, taking the intersection point of the diagonal lines of the rectangular frame as the coordinate information of the target, and converting the rectangular coordinate and the polar coordinate to obtain the azimuth and distance information of the target.
In this embodiment, the conversion of the rectangular coordinates and the polar coordinates enables the user to more intuitively observe the specific azimuth and distance information of the target.
In an alternative embodiment of the present invention, step 15 may include:
step 151, according to the detection threshold adjusted by at least one stage, performing detection processing on the data signal to determine a target; specifically, the digital signal is detected according to the adjusted multi-level detection threshold, for example, the target is detected by using a threshold adjustment method such as a roundness and a size of an echo, or a first amplitude detection threshold, a second amplitude detection threshold, and the like.
In an optional embodiment of the present invention, the method for detecting a marine radar target based on deep learning may further include:
and 152, if the X-circle image is scanned by the antenna of the radar, and the Y-circle image is matched with the target, determining that the target is a real target and outputting the target, wherein X and Y are positive integers, and Y is larger than X/2. Specifically, the result of antenna multi-turn scanning is adopted for relevant matching, and if the target can be matched within 3 or more turns in 4 turns of antenna scanning, the target is further determined to be a real target, namely, the interference of random noise and rain and snow noise is further filtered.
In an optional embodiment of the present invention, after step 15, the method for detecting a target of a marine radar may further include:
In an optional embodiment of the present invention, the method for detecting a target of a marine radar may further include:
and step 17, displaying the output target. In this embodiment, the target detected by the first filtering algorithm is output to the client for reference by the user.
The following describes a specific implementation process of the above embodiment with reference to a specific flowchart, and as shown in fig. 2, the method for detecting a marine radar target based on deep learning of the present invention may specifically include:
and step 21, acquiring radar original data information acquired by the radar receiver through a signal acquisition module, converting the radar original data information into discrete digital signals through AD (analog-to-digital) and processing the digital signals, and further performing format conversion and processing through a first filtering algorithm.
And step 22, converting the digital signals of the radar into fixed format data processed by the radar signals through a format conversion module, and further converting the fixed format data into first image information.
And step 23, processing the pixel value information of the first image information through the original echo image module to obtain an original echo image.
And 24, preprocessing the original radar echo image by using algorithms such as median filtering and the like through the image preprocessing module, filtering interference of noise waves such as rain clutter and snow clutter and the like to obtain second image information, and then inputting the second image information into the deep learning model or inputting the second image information into the display module for displaying.
Step 25, processing a deep learning algorithm on the second image information through a deep learning model, wherein the deep learning model may be in a format of a califfemod, or may be in a format of a model using other frames; the deep learning method adopts supervised learning, and can adopt open-source Yolo v3, but is not limited to other open-source target detection algorithms, such as Fast RCNN, SSD, Retina-Net, Anchor-Free and the like. The deep learning adopts a supervised learning method to train a plurality of input radar echo images to obtain a better network model so as to realize target detection. When the radar antenna scans for one circle, a complete radar echo image can be obtained, and is preprocessed through algorithms such as median filtering and the like, noise signals such as rain clutter and snow clutter are filtered out, and then the radar echo image is sent into the deep learning model. As shown in fig. 3, the process of the deep learning model includes: when detecting a radar echo image, the image is divided into k × k grids. When the target center falls on a certain grid, namely a target in the network is detected, the target is drawn by a minimum rectangular frame. And calculating the coordinates, width, height and confidence coefficient of the central point of the target frame, and finally extracting target information. If the target center does not fall on a certain network, the target is not detected, and the network finishes the detection. The model is similar to a black box, and as the input samples increase, the model is continuously self-learned, namely, the network works and learns at the same time, so that the generalization capability and the target detection capability of the model are increased. In addition, the confidence coefficient can be adjusted according to the intensity of noise such as sea clutter, rain clutter and the like, the output of the target is further optimized, and the accuracy of target detection is improved.
And 26, extracting target information detected by deep learning through a target information extraction module, identifying the target in a form of an external rectangular frame, further matching the rectangular frame with the target by using a K nearest neighbor algorithm, and extracting coordinates of the rectangular frame to obtain position information of the target.
And 27, converting the rectangular coordinates and the polar coordinates through a coordinate conversion module, so that a user can more intuitively observe specific azimuth and distance information of the target.
Step 28, the threshold in the first filtering algorithm is adjusted by the threshold adjusting module.
And step 29, matching the multi-frame images through a multi-frame correlation module according to the result of the deep learning detection, wherein if 3 or more than 3 of the 4 frames correspond to the target, the target is considered, and if not, the target is noise. The multi-frame related module further filters the interference of random noise and rain and snow foreign waves, and reduces the false detection of the target.
And step 30, the first filtering algorithm judges whether the existence of the target is detected or not according to the adjusted threshold, if the target is detected, the target output is kept, and if not, no target is output.
And step 31, outputting the target information detected by the first filtering algorithm through the target information output module, and outputting the target information to the target tracking module.
And step 32, continuously tracking the target information obtained by the first filtering algorithm through a target tracking module, wherein the module algorithm can be a tracking algorithm such as a detection before tracking TBD algorithm or an extended Kalman filtering EKF algorithm. And a data association algorithm such as Hungarian and the like can also be adopted, and methods such as acceleration introduction and Kalman filtering expansion are adopted to realize target tracking by using a Singer model. But are not limited to these data associations, processing algorithms, and tracking methods.
And step 33, outputting the target obtained by the first filtering algorithm and the target detected by the deep learning algorithm to the client through the display module for reference and use by the user. As shown in fig. 4, the display module may also display all targets detected by two radars in the X-band and the S-band simultaneously, may also display a target detected by one radar in the X-band or the S-band separately, or may set a common or different area for displaying only two radars; in addition, the display module can set the display range according to the distance of the target, and can display information such as the speed, the direction, the position, the ID and the like of the target.
In the above embodiment of the present invention, as shown in fig. 5, when the target approaches to the ship or a specific area, the sound alarm or the light alarm of the alarm module may be used to remind the user to take a corresponding collision avoidance measure.
In the above embodiment of the present invention, as shown in fig. 6, a user may store various parameters of the radar through the data management module, and look up, analyze, and print target information data, so that the user may look up and analyze historical information of the radar target, and a reliable basis is provided for analyzing a motion trajectory, an intention, and a track plan of the target.
The embodiment of the invention also supports the chart overlay function, and the target information can be displayed on the chart, so that the track and state information of the target can be displayed more intuitively.
The embodiment of the invention can also prompt the position, the direction and the speed information of the user target through the phenomena of sound, light flicker and the like, and can reset and mute the alarm module through the mute and reset button.
In the embodiment of the invention, the ID management module is arranged, so that the ID information of the targets can be managed, the uniqueness of each historical target ID number is ensured, and the flight path planning and management of the targets are facilitated.
In the above embodiments of the present invention, the approaching, departing, and standing targets are set according to the moving direction of the targets, and are classified according to the moving state.
The radar target detection method provided by the embodiment of the invention detects the radar target through the deep learning model, is suitable for the radars with different wave bands, does not need or assist the threshold value of the radar echo to judge the target information, provides more accurate target information for users, and thus provides reliable reference basis for navigation and collision avoidance.
As shown in fig. 7, an embodiment of the present invention further provides a deep learning-based marine radar target detection apparatus 70, including:
the acquisition module 71 is configured to acquire radar original data information acquired by a radar receiver;
the processing module 72 is configured to perform analog-to-digital conversion processing on the radar original data information to obtain a digital signal; detecting and processing the digital signal according to a first filtering algorithm to determine a target; when the first filtering algorithm does not detect the target, detecting the digital signal according to a deep learning algorithm to determine the target; adjusting a threshold in the first filtering algorithm according to the information of the target; and processing the digital signal according to the first filtering algorithm of the adjusted threshold, and outputting the detected target.
Optionally, adjusting the threshold in the first filtering algorithm according to a deep learning algorithm includes: carrying out format conversion on the digital signal to obtain first image information;
processing the first image information to obtain original echo image information;
processing the original echo image to obtain second image information;
and inputting the second image information into a deep learning model for processing to obtain an image detection result and determine a target.
Optionally, performing format conversion on the digital signal to obtain first image information, including: and carrying out format conversion on the digital signal according to a fixed format of radar signal processing to obtain the first image information.
Optionally, processing the first image information to obtain original echo image information includes: and processing the pixel value information of the first image information to obtain the original echo image information.
Optionally, processing the original echo image to obtain second image information includes: and filtering the original echo image by using a preset filtering algorithm to obtain second image information.
Optionally, inputting the second image information into a deep learning model for processing, obtaining an image detection result, and determining a target, including:
dividing the second image into a plurality of k × k grids;
determining that a target in a grid is detected when a target center falls on the grid;
using a minimum rectangular frame to frame the target object;
and calculating the coordinates of the central point of the frame of the target, the width of the frame, the height and the confidence coefficient of the frame to obtain the image detection result, and extracting target information.
Optionally, the processing module 72 is further configured to determine that the target is accurate if N frames of images exist in the image detection result, where M frames of images are matched with the target, M and N are positive integers, and M is greater than N/2.
Optionally, the processing module 72 is further configured to identify the target by using a circumscribed rectangle.
Optionally, the processing module 72 is further configured to use an intersection of diagonal lines of the rectangular frame as coordinate information of the target, and perform conversion between rectangular coordinates and polar coordinates to obtain azimuth and distance information of the target.
Optionally, processing the digital signal according to the first filtering algorithm of the adjusted threshold, and outputting a detected target, includes:
and detecting the data signal according to the detection threshold after at least one stage of adjustment to determine a target.
Optionally, the processing module 72 is further configured to determine that the target is a real target and output the target if an image of a circle Y matches the target in an image of a circle X scanned by an antenna of the radar, where X and Y are positive integers, and Y is greater than X/2.
Optionally, the processing module 72 is further configured to track information of the target obtained according to the first filtering algorithm of the adjusted threshold;
optionally, the apparatus further comprises: and the display module is used for displaying the target.
It should be noted that the apparatus is an apparatus corresponding to the above method, and all the implementations in the above method embodiment are applicable to the embodiment of the apparatus, and the same technical effects can be achieved.
In a specific implementation, as shown in fig. 2, the obtaining module 71 in this embodiment may be the signal acquiring module, and the processing module 72 may be at least one of the format conversion module, the original echo image module, the image preprocessing module, the deep learning model, the multi-frame correlation module, the target information extracting module, the coordinate conversion module, the first filtering algorithm module, the threshold adjusting module, the target information output module, and the target tracking module, or may include these modules. The threshold adjusting module may be a part of the first filtering algorithm module, or may not belong to the first filtering algorithm module, but may adjust the threshold of the first filtering algorithm. In the embodiment of the device, a fault alarm module as shown in fig. 5, a data management module as shown in fig. 6, and the like can be further included.
It is to be noted that, in the embodiment of the present invention, the function implemented by each step of the method corresponds to a module that implements the corresponding function in the system, and accordingly, the examples in the method embodiment and the system embodiment also correspond to each other, and to avoid repetition, the example described in the method embodiment may be applied to the example of the corresponding module in the system embodiment, and the example of the module described in the system embodiment may also be applied to the implementation example of the corresponding step in the method embodiment.
According to the embodiment of the invention, the radar target is detected through the deep learning model, and more accurate target information is provided for a user, so that a reliable reference basis is provided for navigation and collision avoidance.
Furthermore, it is to be noted that in the device and method of the invention, it is obvious that the individual components or steps can be decomposed and/or recombined. These decompositions and/or recombinations are to be regarded as equivalents of the present invention. Also, the steps of performing the series of processes described above may naturally be performed chronologically in the order described, but need not necessarily be performed chronologically, and some steps may be performed in parallel or independently of each other. It will be understood by those skilled in the art that all or any of the steps or elements of the method and apparatus of the present invention may be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in hardware, firmware, software, or any combination thereof, which can be implemented by those skilled in the art using their basic programming skills after reading the description of the present invention.
Thus, the objects of the invention may also be achieved by running a program or a set of programs on any computing device. The computing device may be a general purpose device as is well known. The object of the invention is thus also achieved solely by providing a program product comprising program code for implementing the method or the apparatus. That is, such a program product also constitutes the present invention, and a storage medium storing such a program product also constitutes the present invention. It is to be understood that the storage medium may be any known storage medium or any storage medium developed in the future. It is further noted that in the apparatus and method of the present invention, it is apparent that each component or step can be decomposed and/or recombined. These decompositions and/or recombinations are to be regarded as equivalents of the present invention. Also, the steps of executing the series of processes described above may naturally be executed chronologically in the order described, but need not necessarily be executed chronologically. Some steps may be performed in parallel or independently of each other.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (12)
1. A marine radar target detection method is characterized by comprising the following steps:
acquiring radar original data information acquired by a radar receiver;
performing analog-to-digital conversion processing on the radar original data information to obtain a digital signal;
detecting and processing the digital signal according to a first filtering algorithm to determine a target;
when the first filtering algorithm does not detect the target, detecting the digital signal according to a deep learning algorithm to determine the target;
adjusting a threshold in the first filtering algorithm according to the information of the target;
processing the digital signal according to the first filtering algorithm of the adjusted threshold, and outputting a detected target;
wherein, according to the information of the target, adjusting the threshold in the first filtering algorithm comprises:
and adjusting the threshold in the first filtering algorithm according to the amplitude detection threshold.
2. The method according to claim 1, wherein the detecting the digital signal according to a deep learning algorithm to determine the target comprises:
carrying out format conversion on the digital signal to obtain first image information;
processing the first image information to obtain original echo image information;
processing the original echo image to obtain second image information;
and inputting the second image information into a deep learning model for processing to obtain an image detection result and determine a target.
3. The method of claim 2, wherein converting the format of the digital signal to obtain first image information comprises:
and carrying out format conversion on the digital signal according to a fixed format of radar signal processing to obtain the first image information.
4. The method of claim 2, wherein processing the first image information to obtain raw echo image information comprises:
and processing the pixel value information of the first image information to obtain the original echo image information.
5. The method of claim 2, wherein processing the raw echo image to obtain second image information comprises:
and filtering the original echo image by using a preset filtering algorithm to obtain second image information.
6. The method according to claim 2, wherein inputting the second image information into a deep learning model for processing to obtain an image detection result and determining the target comprises:
dividing the second image into a plurality of k × k grids;
determining that a target in a grid is detected when a target center falls on the grid;
using a minimum rectangular frame to frame the target object;
and calculating the coordinates of the central point of the frame of the target, the width of the frame, the height and the confidence coefficient of the frame to obtain the image detection result, and extracting target information.
7. The marine radar target detection method of claim 2, further comprising:
and if the image detection result contains N frames of images, wherein M frames of images are matched with the target, determining that the target is accurate, wherein M and N are positive integers, and M is greater than N/2.
8. The marine radar target detection method of claim 7, further comprising:
identifying the target by using a circumscribed rectangle frame;
and taking the intersection point of the diagonal lines of the rectangular frame as the coordinate information of the target, and converting the rectangular coordinate and the polar coordinate to obtain the azimuth and distance information of the target.
9. The method of claim 1, wherein processing the digital signal according to the first filtering algorithm that adjusts the threshold and outputting the detected target comprises:
and detecting the data signal according to the detection threshold after at least one stage of adjustment, and outputting the detected target.
10. The marine radar target detection method of claim 9, further comprising:
and if the antenna of the radar scans the X-circle image, and the Y-circle image is matched with the target, determining that the target is a real target, and outputting the target, wherein X and Y are positive integers, and Y is more than X/2.
11. The marine radar target detection method of claim 10, further comprising:
and tracking the information of the target obtained according to the first filtering algorithm of the adjusted threshold, and displaying the target.
12. A marine radar target detection apparatus, comprising:
the acquisition module is used for acquiring radar original data information acquired by the radar receiver;
the processing module is used for carrying out analog-to-digital conversion processing on the radar original data information to obtain a digital signal; detecting and processing the digital signal according to a first filtering algorithm to determine a target; detecting and processing the digital signal according to a deep learning algorithm to determine a target; adjusting a threshold in the first filtering algorithm according to the information of the target; processing the digital signal according to the first filtering algorithm of the adjusted threshold, and outputting a detected target;
wherein, according to the information of the target, adjusting the threshold in the first filtering algorithm comprises:
and adjusting the threshold in the first filtering algorithm according to the amplitude detection threshold.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911388295 | 2019-12-30 | ||
| CN2019113882953 | 2019-12-30 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111323757A CN111323757A (en) | 2020-06-23 |
| CN111323757B true CN111323757B (en) | 2022-04-05 |
Family
ID=71164037
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010208714.7A Active CN111323757B (en) | 2019-12-30 | 2020-03-23 | Target detection method and device for marine radar |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111323757B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112649791B (en) * | 2020-12-24 | 2022-08-05 | 北京海兰信数据科技股份有限公司 | Radar echo processing method and device |
| CN112816956A (en) * | 2020-12-31 | 2021-05-18 | 北京海兰信数据科技股份有限公司 | Method and device for acquiring radar target information |
| CN116109845A (en) * | 2023-02-13 | 2023-05-12 | 广东海洋大学 | Navigation radar oil spill identification method and system based on YOLO network |
Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1959269A1 (en) * | 2007-02-14 | 2008-08-20 | Delphi Technologies, Inc. | Method of adaptively adjusting the target detection sensitivity of a motor vehicle radar |
| CN101937081A (en) * | 2010-08-16 | 2011-01-05 | 北京海兰信数据科技股份有限公司 | Ship navigation radar system and ship navigation method |
| GB201202642D0 (en) * | 2011-02-17 | 2012-04-04 | Bosch Gmbh Robert | Method and device for sensing an environment of a vehicle |
| CN103995257A (en) * | 2014-05-28 | 2014-08-20 | 北京海兰盈华科技有限公司 | Target detection method and device |
| CN105116396A (en) * | 2015-07-17 | 2015-12-02 | 西安空间无线电技术研究所 | Continuous wave radar Doppler echo detection method |
| CN105116381A (en) * | 2015-08-12 | 2015-12-02 | 西安电子科技大学 | Multi-beam radar time power resource joint distribution method |
| CN105793728A (en) * | 2013-12-03 | 2016-07-20 | 松下知识产权经营株式会社 | Signal processing device |
| CN106249232A (en) * | 2016-08-24 | 2016-12-21 | 中国电子科技集团公司第二十八研究所 | Method for tracking target based on target travel situation information data association strategy |
| CN106842194A (en) * | 2017-03-07 | 2017-06-13 | 北京海兰信数据科技股份有限公司 | A kind of adaptive targets detection method and device |
| CN108197582A (en) * | 2018-01-10 | 2018-06-22 | 武汉理工大学 | Maritime affairs radar image processing method based on deep learning |
| CN108872961A (en) * | 2018-06-28 | 2018-11-23 | 西安电子科技大学 | Radar Weak target detecting method based on low threshold |
| CN109188430A (en) * | 2018-07-31 | 2019-01-11 | 安徽四创电子股份有限公司 | A kind of target extraction method based on ground surveillance radar system |
| CN109891262A (en) * | 2016-10-27 | 2019-06-14 | 株式会社电装 | Object detection device |
| CN109961462A (en) * | 2019-03-25 | 2019-07-02 | 华瑞新智科技（北京）有限公司 | Method for tracking target, device and system |
| CN110058239A (en) * | 2019-04-29 | 2019-07-26 | 上海保隆汽车科技股份有限公司 | A kind of vehicle-mounted millimeter wave radar installations and object detection method |
| CN110082754A (en) * | 2019-04-19 | 2019-08-02 | 北京海兰信数据科技股份有限公司 | Weak target self-adapting detecting system and detection method based on tracking feedback |
| CN110398720A (en) * | 2019-08-21 | 2019-11-01 | 深圳耐杰电子技术有限公司 | A kind of anti-unmanned plane detection tracking interference system and photoelectric follow-up working method |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7236123B2 (en) * | 2004-09-03 | 2007-06-26 | American Systems Corporation | System for enhanced detection of a target |
| AU2012325362B2 (en) * | 2011-10-19 | 2014-08-07 | Balu Subramanya | Directional speed and distance sensor |
| JP2014025916A (en) * | 2012-06-21 | 2014-02-06 | Furuno Electric Co Ltd | Target detection device and echo signal processing method |
| US9933514B1 (en) * | 2015-03-12 | 2018-04-03 | The Boeing Company | LADAR systems with dynamic receiver filters |
| CN106199555A (en) * | 2016-08-31 | 2016-12-07 | 上海鹰觉科技有限公司 | A kind of unmanned boat navigation radar for collision avoidance detection method |
| CN108562879B (en) * | 2018-04-18 | 2022-03-22 | 南京理工大学 | Constant false alarm detection method for ship-borne radar based on FPGA |
-
2020
- 2020-03-23 CN CN202010208714.7A patent/CN111323757B/en active Active
Patent Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1959269A1 (en) * | 2007-02-14 | 2008-08-20 | Delphi Technologies, Inc. | Method of adaptively adjusting the target detection sensitivity of a motor vehicle radar |
| CN101937081A (en) * | 2010-08-16 | 2011-01-05 | 北京海兰信数据科技股份有限公司 | Ship navigation radar system and ship navigation method |
| GB201202642D0 (en) * | 2011-02-17 | 2012-04-04 | Bosch Gmbh Robert | Method and device for sensing an environment of a vehicle |
| CN105793728A (en) * | 2013-12-03 | 2016-07-20 | 松下知识产权经营株式会社 | Signal processing device |
| CN103995257A (en) * | 2014-05-28 | 2014-08-20 | 北京海兰盈华科技有限公司 | Target detection method and device |
| CN105116396A (en) * | 2015-07-17 | 2015-12-02 | 西安空间无线电技术研究所 | Continuous wave radar Doppler echo detection method |
| CN105116381A (en) * | 2015-08-12 | 2015-12-02 | 西安电子科技大学 | Multi-beam radar time power resource joint distribution method |
| CN106249232A (en) * | 2016-08-24 | 2016-12-21 | 中国电子科技集团公司第二十八研究所 | Method for tracking target based on target travel situation information data association strategy |
| CN109891262A (en) * | 2016-10-27 | 2019-06-14 | 株式会社电装 | Object detection device |
| CN106842194A (en) * | 2017-03-07 | 2017-06-13 | 北京海兰信数据科技股份有限公司 | A kind of adaptive targets detection method and device |
| CN108197582A (en) * | 2018-01-10 | 2018-06-22 | 武汉理工大学 | Maritime affairs radar image processing method based on deep learning |
| CN108872961A (en) * | 2018-06-28 | 2018-11-23 | 西安电子科技大学 | Radar Weak target detecting method based on low threshold |
| CN109188430A (en) * | 2018-07-31 | 2019-01-11 | 安徽四创电子股份有限公司 | A kind of target extraction method based on ground surveillance radar system |
| CN109961462A (en) * | 2019-03-25 | 2019-07-02 | 华瑞新智科技（北京）有限公司 | Method for tracking target, device and system |
| CN110082754A (en) * | 2019-04-19 | 2019-08-02 | 北京海兰信数据科技股份有限公司 | Weak target self-adapting detecting system and detection method based on tracking feedback |
| CN110058239A (en) * | 2019-04-29 | 2019-07-26 | 上海保隆汽车科技股份有限公司 | A kind of vehicle-mounted millimeter wave radar installations and object detection method |
| CN110398720A (en) * | 2019-08-21 | 2019-11-01 | 深圳耐杰电子技术有限公司 | A kind of anti-unmanned plane detection tracking interference system and photoelectric follow-up working method |
Non-Patent Citations (8)
| Title |
|---|
| A study on the verification of collision avoidance support system in real voyages;Dajung Kim等;《2015 International Association of Institutes of Navigation World Congress (IAIN)》;20151210;全文 * |
| Adaptive Ship Detection for Single-Look Complex SAR Images Based on SVWIE-Noncircularity Decomposition;Zhao Yu-Huan等;《Sensors》;20181231;第18卷(第10期);全文 * |
| Real-Time Target Detection and Tracking System Based on Stereo Camera for Quadruped Robots;Jiang, ZY等;《PROCEEDINGS OF THE 2019 31ST CHINESE CONTROL AND DECISION CONFERENCE (CCDC 2019)》;20190605;全文 * |
| Target Tracking in Clutter with Extremum Seeking Control for Adaptive Detection Thresholding;Park, SH等;《SENSORS》;20190320;第19卷(第6期);全文 * |
| 基于RBF海杂波微弱目标的检测与提取;赵福立;《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》;20130815;全文 * |
| 基于图像不变量特征的自动目标识别技术研究;朱旭锋;《中国优秀博硕士学位论文全文数据库(博士)信息科技辑》;20130515;全文 * |
| 基于深度学习的杂波抑制技术研究;吴文磊;《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》;20190715;全文 * |
| 大幅面可见光遥感图像典型目标识别关键技术研究;韩现伟;《中国优秀博硕士学位论文全文数据库(博士)信息科技辑》;20140115;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111323757A (en) | 2020-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111323756B (en) | Marine radar target detection method and device based on deep learning | |
| CN111323757B (en) | Target detection method and device for marine radar | |
| US8422738B1 (en) | Adaptive automated synthetic aperture radar vessel detection method with false alarm mitigation | |
| EP0888560B1 (en) | Improved method of moment estimation and feature extraction for devices which measure spectra as a function of range or time | |
| CN111899568B (en) | Bridge anti-collision early warning system, method and device and storage medium | |
| CN101915910B (en) | Method and system for identifying marine oil spill object by marine radar | |
| CN111247525A (en) | Lane detection method and device, lane detection equipment and mobile platform | |
| JP2017156219A (en) | Tracking device, tracking method, and program | |
| de Oliveira et al. | Deep convolutional autoencoder applied for noise reduction in range-Doppler maps of FMCW radars | |
| CN109782247B (en) | Method for correcting radar echo by using track information | |
| CN107942329A (en) | Motor platform single-channel SAR is to surface vessel object detection method | |
| CN110263646A (en) | A kind of sea weak target detection method and system based on convolutional neural networks | |
| CN105427342A (en) | Method and system for detecting and tracking underwater small-target sonar image target | |
| CN113743385A (en) | Unmanned ship water surface target detection method and device and unmanned ship | |
| CN111046877A (en) | Millimeter wave image suspicious article detection method and system | |
| CN115061113A (en) | Target detection model training method and device for radar and storage medium | |
| CN111160260A (en) | SAR image target detection method and system | |
| CN108983194B (en) | Target extraction and condensation method based on ground monitoring radar system | |
| CN112305531B (en) | Target detection method and device | |
| CN112967320B (en) | Ship target detection tracking method based on bridge anti-collision | |
| CN111856445A (en) | Target detection method, device, equipment and system | |
| CN112435249B (en) | Dynamic small target detection method based on circumferential scanning infrared search system | |
| CN117173215A (en) | Inland navigation ship whole-course track identification method and system crossing cameras | |
| Tang et al. | A novel approach for automatic recognition of LPI radar waveforms based on CNN and attention mechanisms | |
| CN112816956A (en) | Method and device for acquiring radar target information |
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 |