A real-time processing method capable of self-adapting to point and planar targets

Abstract

This invention discloses a real-time processing method for adaptive point and surface targets, comprising the following steps: Step S1, compressing the original detector image p(x,y) to obtain a compressed image q(m,n); Step S2, performing image preprocessing on both the original detector image p(x,y) and the compressed image q(m,n) to obtain target clustering information P and Q; Step S3, calculating the centroids of the target clustering information P and Q using gray-level weighted centroid calculation to obtain the target centroid set C of the original image p(x,y). p The set of target centroids C of the compressed image q(m,n) q Step S4: Transform the target centroid coordinates of the compressed image to the original image, and merge them with the target centroid set of the original image to obtain the final result.

Description

Technical Field

[0001] This invention specifically relates to a real-time processing method capable of adaptively processing point and surface targets. Background Technology

[0002] Currently, satellite applications are becoming increasingly widespread and in-depth, and the development of space situational awareness capabilities is the foundation and guarantee for space offense and defense. Among these capabilities, the detection and identification of space targets is a crucial part of space situational awareness.

[0003] Space targets of different sizes appear significantly different in detector imaging under varying distances, and can be categorized into point targets and area targets. Point targets primarily appear as spots of about 3×3 pixels on the detector, such as stars and distant satellites. Area targets, on the other hand, can reach sizes of 200×200 pixels or larger, such as large targets like space stations and spacecraft at close range. When both point and area targets of significantly different sizes exist in an image, without relevant algorithms, it is impossible to simultaneously extract and locate the centroids of both point and area targets in real time, making it difficult to ensure stable target tracking and maintain the required measurement accuracy. Summary of the Invention

[0004] The purpose of this invention is to provide a real-time processing method that can adapt to point and surface targets, so as to achieve stable extraction and centroid localization for each target when there are point and surface targets of different sizes in the field of view at the same time, thereby reducing the false alarm rate and data storage space requirements, and with good real-time performance.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0006] A real-time processing method capable of adaptively processing point and surface targets, characterized by the following steps:

[0007] Step S1: Compress the original detector image p(x,y) to obtain the compressed image q(m,n);

[0008] Step S2: Perform image preprocessing on the original detector image p(x,y) and compressed image q(m,n) respectively to obtain target clustering information P and Q;

[0009] Step S3: Calculate the centroid set C of the original image p(x,y) by performing gray-level weighted centroid calculation on the target clustering information P and Q. p The set of target centroids C of the compressed image q(m,n) q ;

[0010] Step S4: Transform the target centroid coordinates of the compressed image to the original image, and merge them with the target centroid set of the original image to obtain the final result.

[0011] Further, step S1 specifically includes: assuming the original detector image p(x,y) has a size of K1×K2, dividing it into a predetermined number of sub-regions, calculating the average gray level of each sub-region, and using the average gray level of each sub-region as the pixel gray level of image q to obtain the compressed image q(m,n):

[0012] x∈(0,K1),y∈(0,K2),m∈(0,K1′),n∈(0,K2′)

[0013] Where K′1=K1 / M, K′2=K2 / N, M is the length of the subregion, and N is the width of the subregion.

[0014] Furthermore, the image preprocessing in step S2 includes: segmenting the foreground and background of the image using a threshold segmentation method, including segmenting the foreground and background of image p(x,y) and segmenting the foreground and background of image q(m,n);

[0015] The foreground and background of the segmented image p(x,y) include:

[0016]

[0017] In the formula, T is the segmentation threshold, which is the average gray value of the image background plus the threshold offset, i.e.

[0018] T = Mean BLG +ΔT

[0019] Mean BKG Let p(x,y) be the mean gray level of the background of the image p(x,y), and let the threshold offset ΔT be a constant.

[0020] The segmented image:

[0021] The foreground and background of q(m,n) include:

[0022]

[0023] In the formula, T is the segmentation threshold, which is the average gray value of the image background plus the threshold offset, i.e.

[0024] T = Mean BKG +ΔT

[0025] Mean BKG Let q(x,y) be the mean gray level of the background of the image q(x,y), and let the threshold offset ΔT be a constant.

[0026] Using the foreground of the image as the target pixels, the target pixels are clustered and merged using the 8-connected domain method to obtain a cluster set P. 1,2,…,i Cluster set Q 1,2,…,j , where i is the number of clustering targets for image p, and j is the number of clustering targets for image q;

[0027] Pi ={(x i1 ,y i1 ),(x i2 ,y i2 ),…,(x ia ,y ia )}, x∈(0,K1), y∈(0,K2)

[0028] In the formula, a is the number of pixels in cluster i;

[0029] Q j ={(m j1 ,n j1 ),(m j2 ,n j2 ),…,(m jb ,n jb )},m∈(0,K1′),n∈(0,K2′)

[0030] In the formula, b is the number of pixels in cluster j.

[0031] Further, step S4 includes: C according to formulas (1) and (2) q Coordinate C q Transforming (m,n) to the original image p coordinate system, we obtain set C. pq ;

[0032] C pq (x)=C q (m)×M (1)

[0033] C pq (y)=C q (n)×N (2)

[0034] Where M and N are the length and width of the sub-region, respectively;

[0035] For set C p After filtering, set C is obtained. p If C p There exist points in the interval that satisfy the following conditions:

[0036] x p ∈(x pq -M,x pq +M),y p ∈(y pq -M,y pq +M)

[0037] In the formula, (x pq ,y pq )∈C pq ;

[0038] Then delete (x) from the set.p ,y p );

[0039] C pq With C p Merge, with the final target set C.

[0040] Compared with the prior art, the present invention has the following advantages:

[0041] 1. When facing point and surface targets of varying sizes (ranging from a few pixels to tens of thousands of pixels) within the field of view, this invention extracts the centroid set by simultaneously processing the original image and the compressed image, and then uses a specific method to merge and process the two sets, thereby achieving stable extraction and centroid localization of various targets of different sizes.

[0042] 2. This invention uses image compression, which can compress large target (thousands of pixels or more) image information by several times (related to the size of the sub-region), reducing the data storage space requirements in the image processing process;

[0043] 3. The real-time processing method for targets proposed in this invention is in pipeline form, which is easy to implement in software and has good real-time performance. Attached Figure Description

[0044] Figure 1 This is a flowchart of a point target extraction method under high dynamic conditions according to the present invention. Detailed Implementation

[0045] The present invention will be further described below with reference to the accompanying drawings and by providing a detailed description of a preferred embodiment.

[0046] like Figure 1 As shown, a real-time processing method capable of adaptively processing point and surface targets includes the following steps:

[0047] Step S1: Compress the original detector image p(x,y) to obtain the compressed image q(m,n);

[0048] Step S2: Perform image preprocessing on the original detector image p(x,y) and compressed image q(m,n) respectively to obtain target clustering information P and Q;

[0049] Step S3: Calculate the centroid set C of the original image p(x,y) by performing gray-level weighted centroid calculation on the target clustering information P and Q. p The set of target centroids C of the compressed image q(m,n) q ;

[0050] Step S4: Transform the target centroid coordinates of the compressed image to the original image, and merge them with the target centroid set of the original image to obtain the final result.

[0051] Further, step S1 specifically includes: assuming the original detector image p(x,y) has a size of K1×K2, dividing it into a predetermined number of sub-regions, calculating the average gray value of each sub-region, and using the average gray value of each sub-region as the pixel gray value of image q to obtain the compressed image q(m,n):

[0052] x∈(0,K1),y∈(0,K2),m∈(0,K1′),n∈(0,K2′)

[0053] Where K′1=K1 / M, K′2=K2 / N, M is the length of the subregion, and N is the width of the subregion.

[0054] Furthermore, the image preprocessing in step S2 includes: segmenting the foreground and background of the image using a threshold segmentation method, including segmenting the foreground and background of image p(x,y) and segmenting the foreground and background of image q(m,n);

[0055] The foreground and background of the segmented image p(x,y) include:

[0056]

[0057] In the formula, T is the segmentation threshold, which is the average gray value of the image background plus the threshold offset, i.e.

[0058] T = Mean BLG +ΔT

[0059] Mean BKG Let p(x,y) be the mean gray level of the background of the image p(x,y), and let the threshold offset ΔT be a constant.

[0060] The segmented image:

[0061] The segmented image:

[0062] The foreground and background of q(m,n) include:

[0063]

[0064] In the formula, T is the segmentation threshold, which is the average gray value of the image background plus the threshold offset, i.e.

[0065] T = Mean BKG +ΔT

[0066] Mean BKG Let q(x,y) be the mean gray level of the background of the image q(x,y), and let the threshold offset ΔT be a constant.

[0067] It should be noted that in subsequent steps, only the foreground of the image needs to be used, that is, the background of the image is set to 0.

[0068] Using the foreground of the image as the target pixels, the target pixels are clustered and merged using the 8-connected domain method to obtain a cluster set P. 1,2,…,i Cluster set Q 1,2,…,j , where i is the number of clustering targets for image p, and j is the number of clustering targets for image q;

[0069] P i ={(x i1 ,y i1 ),(x i2 ,y i2 ),…,(x ia ,y ia )}, x∈(0,K1), y∈(0,K2)

[0070] In the formula, a is the number of pixels in cluster i;

[0071] Q j ={(m j1 ,n j1 ),(m j2 ,n j2 ),…,(m jb ,n jb )},m∈(0,K1′),n∈(0,K2′)

[0072] In the formula, b is the number of pixels in cluster j.

[0073] Further, step S4 includes: C according to formulas (1) and (2) q Coordinate C q Transforming (m,n) to the original image p coordinate system, we obtain set C. pq ;

[0074] C pq (x)=C q (m)×M (1)

[0075] C pq (y)=C q (n)×N (2)

[0076] Where M and N are the length and width of the sub-region, respectively;

[0077] For set C p After filtering, set C is obtained. p If C p There exist points in the interval that satisfy the following conditions:

[0078] x p ∈(x pq -M,x pq +M),y p ∈(ypq -M,y pq +M)

[0079] In the formula, (x pq ,y pq )∈C pq ;

[0080] Then delete (x) from the set. p ,y p );

[0081] C pq With C p Merge, with the final target set C.

[0082] Assuming the detector array size is 2048×2048 pixels, and the image output order is row-by-row from top to bottom, with each row output from left to right, the specific processing algorithm for the detector output image is as follows:

[0083] a) Merge the detector output image p to obtain a compressed image q. Divide the image p into 16384 sub-regions with 16×16 pixels as sub-regions. Calculate the average gray value of each region and use this average gray value as the pixel gray value of the compressed image q to obtain the image q.

[0084] b) Perform pixel filtering and clustering on images p and q respectively, and output the target clustering information.

[0085] c) Based on the image preprocessing information, the centroid sets Cp and Cq of the target images p and q are calculated using the gray-scale weighted centroid calculation method.

[0086] d) Convert the Cq coordinates to the p coordinate system to obtain the final target set C.

[0087] In summary, the present invention provides a point target extraction method under high dynamic conditions, which enables stable extraction and centroid localization of each target when point and surface targets of different sizes exist simultaneously in the field of view, reducing false alarm rate and data storage space requirements, and has good real-time performance.

[0088] It should be noted that, in the embodiments of the present invention, the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential," etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the accompanying drawings and are only for the convenience of describing the embodiments. They do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0089] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0090] Although the present invention has been described in detail through the preferred embodiments above, it should be understood that the above description should not be considered as a limitation of the present invention. Various modifications and substitutions to the present invention will be apparent to those skilled in the art after reading the above description. Therefore, the scope of protection of the present invention should be defined by the appended claims.

Claims

1. A real-time processing method capable of adaptively processing point and surface targets, characterized in that, Includes the following steps: Step S1, process the original image of the detector. Compress the image to obtain a compressed image. ; Step S2, process the original image of the detector respectively. and compressed images Image preprocessing is performed to obtain target clustering information. ; Step S3, through target clustering information The original image is obtained by performing centroid calculation with grayscale weighting. Target centroid set C p and compressed images Target centroid set C q ; Step S4: Transform the target centroid coordinates of the compressed image to the original image, and merge them with the target centroid set of the original image to obtain the final result; Step S4 includes: According to formulas (1) and (2), C q Coordinate C q ( m , n Convert to the original image p In the coordinate system, we obtain set C. pq ; C pq ( x ) = C q ( m )×M (1) C pq ( y ) = C q ( n )×N (2) Where M and N are the length and width of the sub-region, respectively; For sets By filtering, a set is obtained. like There exist points in the interval that satisfy the following conditions: In the formula, ; Then delete from the set ; Will and Merge, final target set .

2. The real-time processing method for adaptive point and surface targets as described in claim 1, characterized in that, Step S1 specifically includes: assuming the original image of the detector Size is The image is divided into a preset number of sub-regions, and the average gray value of each sub-region is calculated. The average gray value of each sub-region is then used as the image's mean. q The pixel grayscale values ​​are used to obtain the compressed image. : in, , M is the length of the subregion, and N is the width of the subregion.

3. The real-time processing method for adaptive point and surface targets as described in claim 2, characterized in that, The graphic preprocessing in step S2 includes: The threshold segmentation method is used to segment the foreground and background of an image, including image segmentation. Foreground and background and segmented images Foreground and background; The segmented image The foreground and background include: In the formula, T is the segmentation threshold, which is the average gray value of the image background plus the threshold offset, i.e. For image Background grayscale mean, threshold offset It is a constant value; The segmented image The foreground and background include: In the formula, T is the segmentation threshold, which is the average gray value of the image background plus the threshold offset, i.e. For image Background grayscale mean, threshold offset It is a constant value; Using the foreground of the image as the target pixels, the target pixels are clustered and merged using the 8-connected component method to obtain a cluster set. ,in i For image p The number of clustering targets, j For image q The number of clustering targets; In the formula, a For clustering i The number of pixels; In the formula, b For clustering j The number of pixels.

Images