CN119091172B - Weak and small target association method, equipment and system for scanning imaging - Google Patents
Weak and small target association method, equipment and system for scanning imagingInfo
- Publication number
- CN119091172B CN119091172B CN202411197928.3A CN202411197928A CN119091172B CN 119091172 B CN119091172 B CN 119091172B CN 202411197928 A CN202411197928 A CN 202411197928A CN 119091172 B CN119091172 B CN 119091172B
- Authority
- CN
- China
- Prior art keywords
- target
- coordinate system
- earth
- carrier
- image
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/04—Scanning arrangements, i.e. arrangements for the displacement of active reading or reproducing elements relative to the original or reproducing medium, or vice versa
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method, equipment and a system for associating weak and small targets in scanning imaging, which belong to the technical field of target association, and comprise the steps of converting the position of each target in a current frame image into an absolute position in an earth coordinate system or an earth polar coordinate system, forming a dataset D by the absolute position of the target detected before, and executing clustering based on density and direction, and determining the targets in the same category as targets with association, wherein the clustering comprises the steps of sequentially accessing the unviewed target in the D until the unviewed target does not exist or the currently accessed target D 0 is a core target, ending the clustering if the unviewed target does not exist in the D, otherwise, enabling the E neighborhood of D 0 and the E neighborhood of D 0 to satisfy the E neighborhood of each core target D i AndObject d j of the inter-angle θ i<θth joins the current class, followed by clustering of the next class. The invention can effectively realize the association of weak and small targets in a scanning scene.
Description
Technical Field
The invention belongs to the technical field of target association, and particularly relates to a method, equipment and a system for associating weak and small targets by scanning imaging.
Background
After the detector detects the target, the target is quickly associated so as to know the number and the position of the target, which is important to the environmental perception capability. When a detector detects a target, the detected target tends to be small due to the long distance, and is therefore generally referred to as a small target. Different from the large target, the features can be directly extracted, and the target detection and target association are completed based on the extracted features, and the weak and small targets lack sufficient texture information due to small size and cannot be subjected to feature extraction, so that the association of the weak and small targets is difficult.
The correlation of the targets is that the space domain-time domain combination is carried out, the time sequence information contained in the image sequence is fused, the time sequence information is also the motion information of the targets, the method is called TBD (Track Before Detect), the method is mainly applied to a scene with low signal-to-noise ratio, the track is screened by utilizing the continuity of the motion of the targets, the targets belonging to different moments but the same track are matched, and the number can be allocated to the targets according to actual conditions.
The detector is divided into different scenes, namely a staring scene and a scanning scene in the actual detection process. In the staring scene, as shown in fig. 1, the field of view center of the detector is relatively fixed, the motion track of the target is continuous, obvious and continuous tracks can be accumulated from the image sequence, and the association is simpler. For gaze scenes, typical conventional TBD algorithms include dynamic programming, a three-dimensional matched filter method, an inter-frame correlation method, maximum likelihood estimation, particle filtering, a multi-level hypothesis testing method and the like, and can achieve good correlation performance in scenes with low signal-to-noise ratio, mainly due to the fact that the method is not only dependent on space information of a current frame, but also integrates time sequence information of the previous N frames to perform energy accumulation, and accordingly noise and other interference is well suppressed. But is limited by too much data to be processed, the algorithm is complex, and the method has poor real-time performance and is not beneficial to practical application realization. On the other hand, the premise of the effectiveness of the method is to accurately accumulate the energy of the moving target in the continuous multiframe, which needs to overcome the problems of unstable target movement track, including but not limited to the problems of blocked target, track breakage caused by shake of a searching platform, and the like, especially in an airborne platform, the detector always needs to continuously scan a target area in a searching mode, the available information of the target between adjacent frames is very little, especially in the fast scanning of the detector, the adjacent frames may not have a coincident part or only have very little coincident part, at the moment, when the associated target aims at the next time of scanning the target by the detector, the target can be known and associated with the target detected before, the movement of the whole platform needs to be decoupled, and the hardware information of the detector, such as azimuth angle, pitch angle, and the like, needs to be fused, so that the error of the whole system is introduced, and the difficulty of target association is aggravated.
Unlike gaze scenes, in which the center of the field of view periodically changes, the object may briefly appear at the edge of the field of view and disappear, waiting for the detector to turn to the corresponding angle of the object next time, which causes a "window period" for detecting the object, and the image sequence includes multiple frames of images without object information, where the positions of the objects in the images are greatly different, and the timing information is lacking. Fig. 2 shows a scanning scenario, in which the positions of the targets in the image are different in each scanning period, the targets have no continuous track, and N frame intervals (depending on the angular velocity of scanning) exist between two adjacent targets, so that the detector can hardly make a correlation with the targets detected before when detecting the current target. Currently, there is no effective weak and small object association means in a scanning scenario. The target motion characteristics are different in different scenes, so that the weak and small target association method proposed for the staring scene cannot effectively realize weak and small target association in the scanning scene.
Disclosure of Invention
Aiming at the defects and improvement demands of the prior art, the invention provides a method, equipment and a system for associating weak and small targets in scanning imaging, and aims to provide a method capable of effectively associating the weak and small targets in a scanning scene based on the motion characteristics of the targets in the scanning scene.
In order to achieve the above object, according to one aspect of the present invention, there is provided a method for associating a small object in a frame image sequence detected by a detector in a scan scene, the method comprising:
Performing target detection on the current frame image to obtain the position of each target in the image, and then converting the position of each target into an earth coordinate system or an earth polar coordinate system to obtain the absolute position of each target;
Clustering based on density and direction is carried out on a data set D formed by the absolute positions of all targets in the current frame image and the absolute positions of all targets detected before, and targets belonging to the same category are determined to be related targets in a clustering result, wherein the clustering based on the density and the direction comprises:
step S1, initializing a class label k=1;
Step S2, sequentially accessing the objects which are not accessed in the data set D until the objects which are not accessed do not exist in the data set D, or the currently accessed object D 0 is a core object;
Step S3, if the data set D does not have an object which is not accessed, the step S6 is carried out, if the currently accessed object D 0 is a core object, class labels of objects in the E neighborhood of the object D 0 are marked as k, and the step S4 is carried out;
Step S4, for each core object d i in the E neighborhood of object d 0, satisfy in its E neighborhood And (3) withThe class label of the object d j with the included angle theta i<θth is marked as k, and theta th is a preset angle threshold;
Step S5, after updating the category label k according to k=k+1, the step S2 is carried out;
and S6, forming a clustering result by each class, and ending the clustering.
In some alternative embodiments, the absolute position is the position of the object in the earth coordinate system, and for any one object, converting its position information in the image into the earth coordinate system, includes:
according to Calculating an azimuth angle h b, a pitch angle v b and a distance rho b of the target under a carrier polar coordinate system to obtain carrier polar coordinates of the target (h b,vb,ρb);h0、v0 and rho 0 respectively represent the azimuth angle, the pitch angle and the distance of the center of a view field of the detector, alpha and beta respectively represent the high and wide view angles of the image, M and N respectively represent the number of pixels of the image, and Deltar and Deltac respectively represent the offset of the position of the target in the image relative to the center of the image in the high and wide directions;
according to Converting the target from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain a coordinate x b＝(xb,yb,zb of the target in the carrier rectangular coordinate system;
according to Converting the target position from the carrier rectangular coordinate system to a geographic coordinate system to obtain a coordinate x g＝(xg,yg,zg of the target in the geographic coordinate system; the transformation matrix is from a carrier coordinate system to a geographic coordinate system;
according to Converting the target position from a geographic coordinate system to an earth coordinate system to obtain a coordinate x e＝(xe,ye,ze of the target in the earth coordinate system), and taking the coordinate x e＝(xe,ye,ze as an absolute position of the target; is a transformation matrix from a geographic coordinate system to an earth coordinate system.
In some alternative embodiments, the absolute position is the position of the object in the earth's polar coordinate system, and for any one object, converting its position information in the image into the earth's polar coordinate system includes:
according to Calculating an azimuth angle h b, a pitch angle v b and a distance rho b of the target under a carrier polar coordinate system to obtain carrier polar coordinates of the target (h b,vb,ρb);h0、v0 and rho 0 respectively represent the azimuth angle, the pitch angle and the distance of the center of a view field of the detector, alpha and beta respectively represent the high and wide view angles of the image, M and N respectively represent the number of pixels of the image, and Deltar and Deltac respectively represent the offset of the position of the target in the image relative to the center of the image in the high and wide directions;
according to Converting the target from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain a coordinate x b＝(xb,yb,zb of the target in the carrier rectangular coordinate system;
according to Converting the target position from the carrier rectangular coordinate system to a geographic coordinate system to obtain a coordinate x g＝(xg,yg,zg of the target in the geographic coordinate system; the transformation matrix is from a carrier coordinate system to a geographic coordinate system;
according to Converting the target position from the geographic coordinate system to the earth coordinate system to obtain a coordinate x e＝(xe,ye,ze of the target in the earth coordinate system; The transformation matrix is from a geographic coordinate system to an earth coordinate system;
The coordinates of the target in the earth coordinate system are converted into the polar coordinate system of the earth as the absolute position of the target.
Further, the method for associating the weak and small targets in the scanning imaging further comprises the step of caching absolute positions of targets in the current frame image into a preset memory after the absolute positions are obtained.
According to yet another aspect of the present invention, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the above-described method of associating a small object with a scanning image provided by the present invention.
According to yet another aspect of the present invention, there is provided a computer readable storage medium comprising a stored computer program which, when executed by a processor, implements the above-described method for associating a small object with a scanning image provided by the present invention.
According to still another aspect of the present invention, there is provided a small and weak target association apparatus for scanning imaging, which includes a computer readable storage medium for storing a computer program and a processor for reading the computer program stored in the computer readable storage medium, and implementing the computer readable storage medium provided by the present invention.
According to a further aspect of the invention, a scanning imaging detection system is provided, comprising a detector and the scanning imaging weak and small target association device provided by the invention.
In general, through the above technical solutions conceived by the present invention, the following beneficial effects can be obtained:
the invention provides a clustering method based on density and direction, and clusters the absolute position of the target in the current frame image and the absolute position of the target detected before, focuses on the density accessibility among data and focuses on the direction accessibility among the data, ensures that the included angle among core objects in the same category does not exceed a preset angle threshold value, and finally can accurately and effectively realize the weak and small target association under a scanning scene based on the motion characteristics of the target.
Drawings
FIG. 1 is a schematic diagram of a sequence of frame images detected by a detector in a gaze scene;
FIG. 2 is a schematic diagram of a sequence of frame images detected by a detector in a scanning scene;
FIG. 3 is a flowchart of a method for associating small and weak targets with scanning imaging according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of the earth coordinate system;
FIG. 5 is a schematic diagram of a geographic coordinate system;
FIG. 6 is a schematic diagram of a carrier coordinate system;
Fig. 7 is a schematic diagram of a position of a target in a frame image according to an embodiment of the present invention;
fig. 8 is a schematic diagram of a clustering method based on density and direction, in which (a) is a noise identification schematic diagram and (b) is a direction reachability schematic diagram;
fig. 9 is a schematic diagram of clustering results of different clustering methods provided by the embodiment of the invention, wherein (a) is a schematic diagram of clustering results of a DBSCAN clustering method, and (b) is a schematic diagram of clustering results of a clustering method based on density and direction;
FIG. 10 shows frame images detected by a detector in a scanning scene, wherein (a) - (f) are 6 frame images detected sequentially;
FIG. 11 is a schematic diagram of clustering results when noise is not added in the embodiment of the invention;
FIG. 12 is a schematic diagram of clustering results after adding noise 1 according to an embodiment of the present invention;
fig. 13 is a schematic diagram of clustering results after adding noise 2 according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, the technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
In the present invention, the terms "first," "second," and the like in the description and in the drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order.
In order to solve the technical problem that the existing weak and small target association method can not effectively realize the weak and small target association under a scanning scene, the invention provides a weak and small target association method, equipment and a system for scanning imaging, and the whole thought is as follows: the detected position information of the target in the image is converted into an earth coordinate system or an earth polar coordinate system, so that the target has unique absolute position information, the influence of a scanning scene on the target motion track is eliminated, and the absolute position clustering of the target based on a density and direction clustering method is further provided by combining the motion characteristics of the target, so that accurate and effective weak and small target association is realized.
The clustering method based on the density and the direction is a novel clustering method obtained by adding a directivity index based on the motion characteristic of a target on the basis of a traditional clustering method DBSCAN based on the density. Before explaining the technical scheme of the invention in detail, a DBSCAN clustering method and related concepts are briefly described as follows:
The DBSCAN algorithm is based on density clustering, and the core idea is to search data points meeting requirements continuously in a search area with a fixed radius, generate new classes when the data points meeting the requirements are not available, and continue searching until no new classes are generated. Two super parameters, eps and MinPts, are referred to herein as the radius of the search area, and MinPts represents the minimum amount of data in a neighborhood that is determined to be of the same class. There are some key concepts in DBSCAN:
e neighborhood, which is to set the area in the radius of the Eps;
the core object is data with the number of data in the E neighborhood being greater than MinPts;
The direct density can reach all data in the E neighborhood of the core object, and the direct density can reach the data in the E neighborhood of the core object without symmetry, because the data in the E neighborhood of the core object is not necessarily the core object;
The density is up to the density that the direct density is up to the transmissibility, and two data can be connected if passing through a plurality of core objects, the density is up to the density;
two data that are within the E neighborhood of the same core object but not directly reachable by density are called density links, which want to have symmetry.
The flow of the DBSCAN algorithm can be described by pseudo code as shown in tables 1 and 2. Specifically, for the data set D to be clustered, none of the objects therein is accessed at the initial time. Traversing the object in the data set D, if the object D i is not accessed, determining the data quantity in the E adjacent area according to the area radius Eps, if the quantity threshold MinPts is greater than or equal to the quantity threshold MinPts, marking the object D i as a core object, dividing the data in the E adjacent area of the object D i into the current kth category, and completing the expansion of the current kth category through ExpandCluster (D i, eps, minPts, k) functions. The ExpandCluster (D i, eps, minPts, k) function further traverses the object in the current kth class, if it is not accessed and is a core object, further adds the data in the E-neighborhood of the object to the current kth class, and re-traverses the object in the current kth class until a new object can not be added to the current kth class. Then, the category label is added with 1, and the next category is determined according to the same method.
TABLE 1
TABLE 2
When the data are clustered, the meaning of the used terms is the same as that of the terms, and the targets which belong to the same category and are associated in the finally obtained clustering result.
Aiming at the problems that the positions of the targets in the images are different and continuous tracks are not generated in a scanning scene, the method converts the position information of the targets in the images into an earth coordinate system or an earth polar coordinate system before clustering, so that the position information of the targets has uniqueness. It is readily understood that the coordinate transformation may be accomplished based on servo information provided by the servo system of the detector and inertial navigation information provided by the inertial navigation system. When the position of the target is converted into the earth coordinate system, the target has unique three-dimensional coordinate information, and when the position of the target is converted into the earth polar coordinate system, the target has unique direction angle and pitch angle information. In view of the relative simplicity of calculation of the direction and pitch angles when clustering is performed, the polar coordinate system of the earth is taken as an example in the following examples without loss of generality.
The following are examples.
Example 1:
a method for associating weak and small targets in a frame image sequence detected by a detector in a scanning scene is provided.
As shown in fig. 3, the method for associating small and weak targets in scan imaging provided in this embodiment includes:
Performing target detection on the current frame image to obtain the position of each target in the image, and then converting the position of each target into an earth coordinate system or an earth polar coordinate system to obtain the absolute position of each target;
And clustering based on density and direction is carried out on a data set D formed by the absolute positions of all targets in the current frame image and the absolute positions of all targets detected before, and targets belonging to the same category in a clustering result are determined to be related targets.
In this embodiment, any kind of target detection method based on local contrast, background estimation, principal component analysis and the like may be used to complete target detection. After the position of each target in the image is obtained, the position can be converted into an earth polar coordinate system according to the servo information provided by a servo system of the detector, wherein a plurality of navigation coordinate systems and conversion between the navigation coordinate systems are involved, wherein fig. 4 is an earth coordinate system, abbreviated as an e system, fig. 5 is a geographic coordinate system, abbreviated as a g system, and fig. 6 is a carrier coordinate system, abbreviated as a b system. The specific process of coordinate transformation comprises the following steps:
firstly, converting coordinates of an object in an image into a carrier polar coordinate system to obtain carrier polar coordinates (h b,vb,ρb) of the object, wherein a related calculation expression is as follows:
ρb＝ρ0 (3)
as shown in fig. 7, h b、vb and ρ b respectively represent the azimuth angle, pitch angle and distance of the target in the carrier polar coordinate system, h 0、v0 and ρ 0 respectively represent the azimuth angle, pitch angle and distance of the detector field center, α and β respectively represent the image height and width field angle, M and N respectively represent the image height and width pixel number, (r 0,c0) represent the image center coordinates, (r, c) represent the detected target position, Δr and Δc respectively represent the shift of the position of the target in the image in the height and width directions relative to the image center;
Then, converting the target from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain a coordinate x b＝(xb,yb,zb of the target in the carrier rectangular coordinate system), and the related calculation expression is as follows:
xb＝ρb×cosvb×sinhb (4)
yb＝ρb×cosvb×coshb (5)
zb＝ρb×sinvb (6)
Then, the target position is converted from the rectangular coordinate system of the carrier to the geographic coordinate system, so as to obtain the coordinate x g＝(xg,yg,zg of the target in the geographic coordinate system), and the related calculation expression is as follows:
wherein, the The expression of the transformation matrix from the carrier coordinate system to the geographic coordinate system is as follows:
The angle variable in the equation is shown in fig. 6, where ψ represents a pitch angle, θ represents a direction angle, and γ represents a roll angle.
Further, the target position is down-converted from the geographic coordinate system to the earth coordinate system, so as to obtain the coordinate x e＝(xe,ye,ze of the target in the earth coordinate system), and the correlation calculation expression is as follows:
wherein, the The expression of the transformation matrix from the geographic coordinate system to the earth coordinate system is as follows:
The angle variable in the formula is shown in fig. 5, wherein lambda represents the included angle between the longitude line of the carrier and the primary meridian, Representing the included angle between the latitude line of the carrier and the connecting line of the sphere center and the equator.
And finally, converting the position information of the target in the earth coordinate system into the earth polar coordinate system to obtain the direction angle and the pitch angle of the target in the earth polar coordinate system, wherein the information jointly forms the absolute position of the target.
In the actual detection process, the motion of the target and the carrier is similar to linear motion, so that the included angle of the motion direction is not too large between adjacent positions, and therefore, when the correlation of the weak and small targets is realized through clustering, the density accessibility and the direction accessibility are required to be considered. In the embodiment, the clustering based on density and direction adds the direction index based on the traditional DBSCAN clustering method, and specifically comprises the following steps:
step S1, initializing a class label k=1;
Step S2, sequentially accessing the objects which are not accessed in the data set D until the objects which are not accessed do not exist in the data set D, or the currently accessed object D 0 is a core object;
Step S3, if the data set D does not have an object which is not accessed, the step S6 is carried out, if the currently accessed object D 0 is a core object, class labels of objects in the E neighborhood of the object D 0 are marked as k, and the step S4 is carried out;
Step S4, for each core object d i in the E neighborhood of object d 0, satisfy in its E neighborhood And (3) withThe class label of the object d j with the included angle theta i<θth is marked as k, and theta th is a preset angle threshold, and in practical application, the setting can be performed according to specific motion parameters of the carrier and the target, optionally, in this embodiment, theta th is set to 40 degrees;
Step S5, after updating the category label k according to k=k+1, the step S2 is carried out;
and S6, forming a clustering result by each class, and ending the clustering.
TABLE 3 Table 3
The overall flow of the clustering method based on density and direction provided in this embodiment is similar to the flow shown in table 1, except that the present embodiment improves ExpandCluster () function so that the direction reachability can be fully considered when expanding the data in the category. The flow of ExpandCluster () function modified in this embodiment can be described by the pseudo code shown in table 3.
Because the present embodiment adds consideration of direction reachability on the basis of the DBSCAN algorithm, noise points can be effectively identified in the clustering process, as shown in (a) in fig. 8. Specifically, the present embodiment does not iterate clustering for data points that do not have significant directionality when data in the density reachability and direction reachability extension categories are simultaneously considered, as shown in (b) in fig. 8.
Further as shown in fig. 9, the data set originally comprising three targets is classified into one type after being clustered based on density reachability, and has obvious errors as shown in (a) of fig. 9, and is classified into three types after adding the direction reachability index as shown in (b) of fig. 9, which indicates that the embodiment can excavate the directionality in the data meeting the density requirement after adding the direction reachability index, and obviously increases the accuracy of target association.
Fig. 10 shows 6 frames of images detected by the detector in sequence when the detector detects two targets in a scanning scene, wherein the azimuth angle and the pitch angle of a carrier at the center of a field of view, the offset of the target relative to the center of the field of view, and the azimuth angle and the pitch angle of the converted target in the polar coordinate system of the earth are shown in fig. 4 in each frame of images respectively as shown in (a) - (f) in fig. 10.
TABLE 4 Table 4
Clustering is performed based on the clustering method based on the density and the direction provided by the embodiment, and finally, the two targets can be accurately detected, and the association between the targets is realized, as shown in fig. 11. In order to further verify the effectiveness, the present embodiment adds gaussian noise with different degrees to the frame image shown in fig. 10, and then performs object association, where association results are shown in fig. 12 and 13, respectively. The correlation results shown in fig. 12 and fig. 13 both show that the noise can be effectively identified in this embodiment, and for the noise closer to the target, the noise can be effectively identified, and finally different targets can be accurately clustered.
After the current frame image is processed, the next frame image can be processed based on the same flow until the frame image sequence detected by the detector is processed. In consideration of the need to acquire the previously detected target position information when clustering the current frame, in order to acquire the information, as a preferred implementation manner, the embodiment further includes, after obtaining the absolute positions of the targets in the current frame image, buffering the absolute positions in a preset memory for facilitating subsequent access.
According to the invention, the offset of the target between continuous frames caused by the scanning motion of the detector is decoupled by combining the attitude information of the inertial navigation system, the position of the target is projected to the earth coordinate system, and the projected coordinate points are clustered by adopting a DBSCAN clustering method added with the direction reachable property, so that the targets detected in different scanning periods can be accurately and effectively associated.
Example 2:
a computer program product comprising a computer program which, when executed by a processor, implements the method for associating small and weak objects of scanning imaging provided in embodiment 1 above.
Example 3:
a computer-readable storage medium comprising a stored computer program which, when executed by a processor, implements the method for associating small and weak targets for scanning imaging provided in embodiment 1 above.
Example 4:
the weak and small target association device for scanning imaging comprises a computer readable storage medium and a processor, wherein the computer readable storage medium is used for storing a computer program, and the processor is used for reading the computer program stored in the computer readable storage medium to realize the computer readable storage medium provided by the embodiment 1.
Example 5:
A scanning imaging detection system includes a detector and a scanning imaging small target association apparatus provided in embodiment 4 above.
It will be readily appreciated by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (8)
1. A method for associating weak and small targets in a frame image sequence detected by a detector in a scanning scene is characterized by comprising the following steps:
Performing target detection on the current frame image to obtain the position of each target in the image, and then converting the position of each target into an earth coordinate system or an earth polar coordinate system to obtain the absolute position of each target;
clustering based on density and direction is carried out on a data set D formed by the absolute positions of all targets in the current frame image and the absolute positions of all targets detected before, and targets belonging to the same category in a clustering result are determined to be targets with relevance, wherein the clustering based on the density and the direction comprises:
step S1, initializing a class label k=1;
Step S2, sequentially accessing the objects which are not accessed in the data set D until the objects which are not accessed do not exist in the data set D, or the currently accessed object D 0 is a core object;
Step S3, if the data set D does not contain an object which is not accessed, the step S6 is carried out, if the currently accessed object D 0 is a core object, class labels of objects in the E neighborhood of the object D 0 are marked as k, and the step S4 is carried out;
Step S4, for each core object d i in the E neighborhood of object d 0, satisfy in its E neighborhood And (3) withThe class label of the object d j with the included angle theta i<θth is marked as k, and theta th is a preset angle threshold;
Step S5, after updating the category label k according to k=k+1, the step S2 is carried out;
and S6, forming a clustering result by each class, and ending the clustering.
2. The method for correlating a small and weak object in scan imaging according to claim 1, wherein the absolute position is a position of the object in the earth coordinate system, and wherein for any one object, converting its position information in the image into the earth coordinate system comprises:
according to Calculating an azimuth angle h b, a pitch angle v b and a distance rho b of the target under a carrier polar coordinate system to obtain carrier polar coordinates of the target (h b,vb,ρb);h0、v0 and rho 0 respectively represent the azimuth angle, the pitch angle and the distance of the center of a view field of the detector, alpha and beta respectively represent the high and wide view angles of the image, M and N respectively represent the number of pixels of the image, and Deltar and Deltac respectively represent the offset of the position of the target in the image relative to the center of the image in the high and wide directions;
according to Converting the target from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain a coordinate x b＝(xb,yb,zb of the target in the carrier rectangular coordinate system;
according to Converting the target position from the carrier rectangular coordinate system to a geographic coordinate system to obtain a coordinate x g＝(xg,yg,zg of the target in the geographic coordinate system; the transformation matrix is from a carrier coordinate system to a geographic coordinate system;
according to Converting the target position from a geographic coordinate system to an earth coordinate system to obtain a coordinate x e＝(xe,ye,ze of the target in the earth coordinate system), and taking the coordinate x e＝(xe,ye,ze as an absolute position of the target; is a transformation matrix from a geographic coordinate system to an earth coordinate system.
3. The method for correlating a small and weak object in scan imaging according to claim 1, wherein the absolute position is a position of the object in the polar coordinate system of the earth, and wherein for any one object, converting its position information in the image into the polar coordinate system of the earth comprises:
according to Calculating an azimuth angle h b, a pitch angle v b and a distance rho b of the target under a carrier polar coordinate system to obtain carrier polar coordinates of the target (h b,vb,ρb);h0、v0 and rho 0 respectively represent the azimuth angle, the pitch angle and the distance of the center of a view field of the detector, alpha and beta respectively represent the high and wide view angles of the image, M and N respectively represent the number of pixels of the image, and Deltar and Deltac respectively represent the offset of the position of the target in the image relative to the center of the image in the high and wide directions;
according to Converting the target from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain a coordinate x b＝(xb,yb,zb of the target in the carrier rectangular coordinate system;
according to Converting the target position from the carrier rectangular coordinate system to a geographic coordinate system to obtain a coordinate x g＝(xg,yg,zg of the target in the geographic coordinate system; the transformation matrix is from a carrier coordinate system to a geographic coordinate system;
according to Converting the target position from the geographic coordinate system to the earth coordinate system to obtain a coordinate x e＝(xe,ye,ze of the target in the earth coordinate system; The transformation matrix is from a geographic coordinate system to an earth coordinate system;
The coordinates of the target in the earth coordinate system are converted into the polar coordinate system of the earth as the absolute position of the target.
4. The method for associating small and weak targets with scanning imaging according to any one of claims 1 to 3, further comprising buffering absolute positions of each target in the current frame image in a preset memory after obtaining the absolute positions.
5. A computer program product comprising a computer program which, when executed by a processor, implements the method for associating small objects with scanning imaging according to any one of claims 1 to 4.
6. A computer readable storage medium comprising a stored computer program which, when executed by a processor, implements the method for associating small objects with scanning imaging according to any one of claims 1 to 4.
7. The weak and small target association device for scanning imaging is characterized by comprising a computer readable storage medium and a processor, wherein the computer readable storage medium is used for storing a computer program, and the processor is used for reading the computer program stored in the computer readable storage medium and realizing the computer readable storage medium according to any one of claims 1-4.
8. A scanning imaging detection system comprising a detector and the scanningly imaged small target association device of claim 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411197928.3A CN119091172B (en) | 2024-08-29 | 2024-08-29 | Weak and small target association method, equipment and system for scanning imaging |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411197928.3A CN119091172B (en) | 2024-08-29 | 2024-08-29 | Weak and small target association method, equipment and system for scanning imaging |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN119091172A CN119091172A (en) | 2024-12-06 |
| CN119091172B true CN119091172B (en) | 2025-10-17 |
Family
ID=93696798
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202411197928.3A Active CN119091172B (en) | 2024-08-29 | 2024-08-29 | Weak and small target association method, equipment and system for scanning imaging |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119091172B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111090103A (en) * | 2019-12-25 | 2020-05-01 | 河海大学 | Three-dimensional imaging device and method for dynamic fine detection of underwater small targets |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7483551B2 (en) * | 2004-02-24 | 2009-01-27 | Lockheed Martin Corporation | Method and system for improved unresolved target detection using multiple frame association |
| CN107063259B (en) * | 2017-03-08 | 2020-06-09 | 四川九洲电器集团有限责任公司 | Track association method and electronic equipment |
| CN115423019B (en) * | 2022-09-01 | 2025-08-12 | 西安电子科技大学 | Fuzzy clustering method and device based on density |
-
2024
- 2024-08-29 CN CN202411197928.3A patent/CN119091172B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111090103A (en) * | 2019-12-25 | 2020-05-01 | 河海大学 | Three-dimensional imaging device and method for dynamic fine detection of underwater small targets |
Also Published As
| Publication number | Publication date |
|---|---|
| CN119091172A (en) | 2024-12-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109919974B (en) | Online multi-target tracking method based on multi-candidate association in R-FCN framework | |
| CN112991391A (en) | Vehicle detection and tracking method based on radar signal and vision fusion | |
| CN111402299B (en) | Method and device for remote sensing image target tracking based on staring satellite in geostationary orbit | |
| CN114092566B (en) | Radar data calibration method based on video multiframe self-adaptive optimization | |
| CN109145803A (en) | Gesture identification method and device, electronic equipment, computer readable storage medium | |
| CN117197182B (en) | Lei Shibiao method, apparatus and storage medium | |
| Guo et al. | An anchor-free network with density map and attention mechanism for multiscale object detection in aerial images | |
| CN117036404B (en) | A monocular thermal imaging simultaneous positioning and mapping method and system | |
| CN111161309A (en) | Searching and positioning method for vehicle-mounted video dynamic target | |
| CN112509002A (en) | Target detection tracking method based on connected domain marker | |
| CN119355714A (en) | A multi-sensor fusion SLAM method suitable for dynamic rain and fog environment | |
| CN119295721A (en) | A RGB-D visual SLAM method for indoor dynamic scenes | |
| Li et al. | An oriented SAR ship detector with mixed convolution channel attention module and geometric nonmaximum suppression | |
| CN116665015B (en) | A method for detecting weak and small targets in infrared sequence images based on YOLOv5 | |
| CN119091172B (en) | Weak and small target association method, equipment and system for scanning imaging | |
| Huang et al. | Single target tracking in high-resolution satellite videos: a comprehensive review | |
| CN112561956B (en) | Video target tracking method and device, electronic equipment and storage medium | |
| CN118570709B (en) | Infrared video small target detection method and system based on temporal coding and decoding structure | |
| CN110428446B (en) | Satellite video target tracking method based on mixed kernel correlation filtering | |
| CN114648730B (en) | Infrared target tracking method and system combining gradient statistics with local matching | |
| CN115717887B (en) | Star point rapid extraction method based on gray distribution histogram | |
| CN117575966A (en) | Video image stabilizing method for unmanned aerial vehicle high-altitude hovering shooting scene | |
| CN115393281A (en) | Infrared weak and small target detection tracking method based on mask and adaptive filtering | |
| CN106875417A (en) | A kind of multi-object tracking method associated across time domain based on high-order figure | |
| CN119063732B (en) | Information processing method, system and medium for fusion of on-board real-time target pre-detection and intelligent fine detection |
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 |