Method and apparatus for target tracking, imaging seeker

Abstract

The present application relates to the technical fields of imaging guidance and intelligent weapon equipment, and discloses a method for target tracking. The method includes: acquiring an image frame of a target object; obtaining a corresponding two-dimensional state space for the scale scaling factor and the planar rotation angle between adjacent image frames; performing discrete sampling in the two-dimensional state space and extracting features from each discrete sampling point to obtain eigenvalue; performing frequency domain filtering on the eigenvalue and obtaining the scale scaling factor and / or the rotation angle of the target object according to the filtering result. This method only needs to sparsely sample the two-dimensional state space of scale and rotation, without repeatedly running the target tracking process multiple times, significantly improving the calculation efficiency, so as to be able to obtain the attitude of the target object more efficiently. The present application also discloses a device for target tracking and an imaging seeker.

Description

Technical Field

[0001] The present application relates to the technical field of imaging guidance and intelligent weapon equipment, for example, to a method and device for target tracking, and an imaging seeker. Background Art

[0002] The imaging seeker obtains the error information of the missile deviating from the target by taking images of the target and the background, and uses this information to control the missile to fly accurately to the target. How to stably track the target during the flight of the missile is a key issue affecting the success rate of guidance.

[0003] In actual application environments, since the missile approaches the target quickly and the missile body rotates and shakes during flight, target tracking is required to accurately estimate the scale and / or rotation angle of the target in order to achieve accurate deviation calculation.

[0004] In the process of implementing the embodiments of the present disclosure, it was found that there are at least the following problems in the related art: due to the limitation of the computing power of the missile-borne embedded platform, the efficiency of estimating the posture of the target object is not high. Summary of the invention

[0005] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical components or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0006] The embodiments of the present disclosure provide a method, an apparatus, and an imaging seeker for target tracking, so as to more efficiently obtain the posture of a target object.

[0007] In some embodiments, the method for target tracking includes:

[0008] Acquire an image frame of the target object;

[0009] For the scale scaling factor and the plane rotation angle between adjacent image frames, obtaining the corresponding two-dimensional state space;

[0010] Performing discrete sampling on the two-dimensional state space and extracting features from each discrete sampling point to obtain a feature value;

[0011] The characteristic value is subjected to frequency domain filtering, and a scale factor and / or a rotation angle of the target object is obtained according to the filtering result.

[0012] In some embodiments, obtaining the corresponding two-dimensional state space for the scale scaling factor and the plane rotation angle between adjacent image frames includes:

[0013] The scale factor and the planar rotation angle between adjacent image frames are described using a two-dimensional Cartesian coordinate system to obtain a two-dimensional state space.

[0014] In some embodiments, feature extraction is performed on each discrete sampling point to obtain feature values, including:

[0015] Extract the image regions corresponding to each of the discrete sampling points from the image frame of the target object;

[0016] Obtain the apparent features of the image regions;

[0017] Obtain the feature values corresponding to the image regions according to the apparent features.

[0018] In some embodiments, obtaining the feature values corresponding to the image regions according to the apparent features includes:

[0019] Represent the apparent features of the image regions as feature vectors;

[0020] Obtain the feature values according to the feature vectors.

[0021] In some embodiments, obtaining the scale factor and / or the rotation angle of the target object according to the filtering result includes:

[0022] Determine the scale factor and / or the rotation angle of the target object according to the coordinates of the maximum value of the filtering result in the two-dimensional state space.

[0023] In some embodiments, performing frequency domain filtering on the feature values includes:

[0024] Perform a discrete Fourier transform on the feature values to obtain the frequency domain feature values corresponding to the feature values;

[0025] Perform a spectral multiplication operation on the preset filter and the frequency domain feature values and sum by channel to obtain a first spectrum;

[0026] Perform frequency domain interpolation on the first spectrum to obtain a second spectrum;

[0027] Perform an inverse discrete Fourier transform on the second spectrum to obtain the filtering result.

[0028] In some embodiments, performing an inverse discrete Fourier transform on the second spectrum includes:

[0029] Scale the second spectrum by a set constant in equal proportion;

[0030] Perform an inverse discrete Fourier transform on the amplified second spectrum.

[0031] In some embodiments, after obtaining the scale scaling factor and / or rotation angle of the target object according to the filtering result, the method further includes:

[0032] Obtaining the scale state of the target object and the rotation state of the target object;

[0033] Updating the filter according to the position of the target object in the image frame, the scale state of the target object, and the rotation state of the target object.

[0034] In some embodiments, the apparatus for target tracking includes a processor and a memory storing program instructions, and the processor is configured to execute the above-mentioned method for target tracking when executing the program instructions.

[0035] In some embodiments, the imaging seeker includes the above-mentioned apparatus for target tracking.

[0036] The method, apparatus, and imaging seeker for target tracking provided by the embodiments of the present disclosure can achieve the following technical effects: for the scale scaling factor and planar rotation angle between adjacent image frames, obtain the corresponding two-dimensional state space; perform discrete sampling on the two-dimensional state space and extract features from each discrete sampling point, perform frequency domain filtering on the obtained feature values, and obtain the scale scaling factor and / or rotation angle of the target object according to the filtering result. In this way, only sparse sampling needs to be performed on the two-dimensional state space of scale and rotation, and it is not necessary to repeatedly run the target tracking process multiple times, significantly improving the calculation efficiency, and thus being able to obtain the pose of the target object more efficiently.

[0037] The above general description and the following description are only exemplary and explanatory, and are not used to limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] One or more embodiments are exemplarily illustrated by corresponding drawings. These exemplary illustrations and the drawings do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings are shown as similar elements. The drawings do not constitute a proportional limitation, and among them:

[0039] Figure 1 is a schematic diagram of a method for target tracking provided by an embodiment of the present disclosure;

[0040] Figure 2 is a schematic diagram of obtaining the eigenvalue corresponding to an image region provided by an embodiment of the present disclosure;

[0041] Figure 3 is a schematic diagram of performing frequency domain interpolation on a first spectrum provided by an embodiment of the present disclosure;

[0042] Figure 4It is a schematic diagram of a device for target tracking provided by an embodiment of the present disclosure. Detailed implementation manners

[0043] In order to be able to understand the features and technical content of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for reference and illustration only and are not intended to limit the embodiments of the present disclosure. In the following technical descriptions, for the sake of explanation, numerous details are provided to give a thorough understanding of the disclosed embodiments. However, one or more embodiments may still be implemented without these details. In other cases, well-known structures and devices may be shown in a simplified manner to simplify the drawings.

[0044] In the description of the embodiments of the present disclosure, the terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data may be interchanged under appropriate circumstances so as to implement the embodiments of the present disclosure described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion.

[0045] Unless otherwise specified, the term "plurality" means two or more.

[0046] In the embodiments of the present disclosure, the character " / " indicates that the objects before and after are in an "or" relationship. For example, A / B means: A or B.

[0047] The term "and / or" is a description of the associated relationship of objects, indicating that there can be three relationships. For example, A and / or B means: A or B, or, A and B these three relationships.

[0048] Combined with Figure 1 As shown, an embodiment of the present disclosure provides a method for target tracking, including:

[0049] Step S101, obtaining an image frame of the target object;

[0050] Step S102, obtaining a corresponding two-dimensional state space for the scale scaling factor and the planar rotation angle between adjacent image frames;

[0051] Step S103, performing discrete sampling on the two-dimensional state space and extracting features from each discrete sampling point to obtain feature values;

[0052] Step S104, performing frequency domain filtering on the feature values and obtaining the scale scaling factor and / or rotation angle of the target object according to the filtering result.

[0053] By using the method for target tracking provided by the embodiments of the present disclosure, for the scale scaling factor and the planar rotation angle between adjacent image frames, the corresponding two-dimensional state space is obtained; discrete sampling is performed in the two-dimensional state space and feature extraction is performed on each discrete sampling point, the obtained eigenvalues are frequency-domain filtered, and the scale scaling factor and / or the rotation angle of the target object are obtained according to the filtering result. In this way, only sparse sampling of the two-dimensional state space of scale and rotation is required, and it is not necessary to repeatedly run the target tracking process multiple times, significantly improving the calculation efficiency, so that the attitude of the target object can be obtained more efficiently.

[0054] Optionally, an image acquisition module is used to collect images of the target object in real time at a preset acquisition rate to obtain image frames of the target object.

[0055] Optionally, obtaining the corresponding two-dimensional state space for the scale scaling factor and the planar rotation angle between adjacent image frames includes: using a two-dimensional Cartesian coordinate system to describe the scale scaling factor and the planar rotation angle between adjacent image frames to obtain a two-dimensional state space.

[0056] In some embodiments, the scale and rotation of the target object between adjacent image frames have two degrees of freedom, namely the scale scaling factor and the planar rotation angle. A two-dimensional Cartesian coordinate system is used to describe the scale scaling factor and the planar rotation angle between adjacent image frames, and a two-dimensional state space corresponding to the scale scaling factor and the planar rotation angle is established in this way; the horizontal axis corresponds to the scale scaling factor, and the corresponding relationship is where s 0 is the scale scaling step, and i is the abscissa value; the vertical axis corresponds to the planar rotation angle, and the corresponding relationship is j→j·θ 0 where θ 0 is the planar rotation step, and j is the ordinate value; the polarity is determined by the right-hand screw rule; the origin of the two-dimensional coordinates represents no scale scaling and no in-plane rotation, that is, it corresponds to the original state of the target.

[0057] Optionally, discrete sampling of the two-dimensional state space includes: a point (i, j) in the two-dimensional coordinates, that is, it corresponds to a discrete sampling point (s, θ) in the two-dimensional state space, and the corresponding relationship is where s is the scale scaling factor and θ is the planar rotation angle; optionally, the scale scaling step is set to s = 1.02, and the planar rotation step is set to θ = π / 15.

[0058] Optionally, feature extraction is performed on each discrete sampling point to obtain eigenvalues, including: extracting the image region corresponding to each discrete sampling point from the image frame of the target object; obtaining the apparent features of the image region; and obtaining the eigenvalue corresponding to the image region according to the apparent features.

[0059] Optionally, obtaining the eigenvalue corresponding to the image region according to the apparent feature includes: representing the apparent feature of the image region as a feature vector; obtaining the eigenvalue according to the feature vector.

[0060] In some embodiments, the target object is scaled and rotated in the plane through similarity transformation, and the rectangular image region corresponding to the discrete sampling points is extracted from the current frame image and normalized to a fixed size; the apparent feature of the image region is calculated and represented as a one-dimensional feature vector; the feature vectors of all discrete sampling points are combined into a two-dimensional multi-channel feature according to their coordinates in the two-dimensional state space.

[0061] Optionally, for the preset initial state in the current image frame of the target object, the target object is scaled by a scale factor of s and rotated in the plane by an angle of θ through similarity transformation to obtain a hypothesis about the target scale and rotation state. The rectangular image regions corresponding to the sampling points (s, θ) of each hypothesis are extracted from the current image frame, and the extracted image regions are normalized to image regions of a fixed size; optionally, the apparent features of each image region, such as color, gradient direction, gradient magnitude, etc., are obtained; the apparent features of each image region are represented as a d-dimensional feature vector g(s, θ) ∈ R d , where d is the number of features, R is the time domain, and one of the features l, l ∈ {1,..., d}, both l and d are positive integers; optionally, the features of each discrete sampling point are denoted as a channel feature g l , then the feature vectors of all discrete sampling points constitute a two-dimensional multi-channel feature That is, the eigenvalue g corresponding to the image region.

[0062] In some embodiments, as Figure 2 shown, it is a schematic diagram for obtaining the eigenvalue corresponding to the image region. The rectangular image regions corresponding to the assumed sampling points (s 1 , θ 1 ), (s 2 , θ 2 ) and (s 3 , θ 3 ) are respectively extracted from the current image frame; the extracted image regions are normalized to image regions of a fixed size; the apparent features of each image region are obtained as the feature vectors g(s 1 , θ 1 ) ∈ R d , g(s 2 , θ 2 ) ∈ R d , g(s 3 , θ 3 ) ∈ R d ; the channel features g of each discrete sampling point are obtained 1 , g2 , g 3 ,...... g d , then the eigenvectors of all discrete sampling points constitute the two - dimensional multi - channel feature g.

[0063] Optionally, perform frequency - domain filtering on the eigenvalues, including:

[0064] Perform a discrete Fourier transform on the eigenvalues to obtain the frequency - domain eigenvalues corresponding to the eigenvalues;

[0065] Perform a spectral product operation between a preset filter and the frequency - domain eigenvalues and sum by channel to obtain the first spectrum;

[0066] Perform frequency - domain interpolation on the first spectrum to obtain the second spectrum;

[0067] Perform an inverse discrete Fourier transform on the second spectrum to obtain the filtering result.

[0068] Optionally, perform a two - dimensional discrete Fourier transform independent for each channel on the two - dimensional multi - channel feature g composed of the eigenvectors of all discrete sampling points to obtain the two - dimensional multi - channel feature in the frequency - domain representation That is, the frequency - domain eigenvalue G, where G is the two - dimensional multi - channel feature in the frequency - domain representation, and G l is the l - th channel feature in the frequency - domain representation;

[0069] Optionally, the correlation filter in the two - dimensional multi - channel frequency - domain representation for estimating the scale state and rotation state of the target object at the current moment H l is the correlation filter in the frequency - domain representation of the l - th channel feature, that is, the preset correlation filter H. Perform a spectral product operation independent for each channel between H and the frequency - domain eigenvalue G, and sum by channel to obtain the first spectrum Q of the correlation filter output;

[0070] By calculating obtain the first spectrum Q of the correlation filter output, where Q is the first spectrum of the correlation filter output, is the correlation filter in the frequency - domain representation of the l - th channel feature at time t - 1, is the frequency - domain eigenvalue of the l - th channel feature at time t, d is the number of features, l ∈ {1,..., d}, and both l and d are positive integers.

[0071] Optionally, perform a frequency - domain interpolation operation on the first spectrum output by the correlation filter H by padding zeros in partitions to obtain the second spectrum. In this way, a second spectrum with high resolution can be obtained. For example: the size of the first spectrum is w × h, and the size of the interpolated second spectrum is w′×h′.

[0072] In some embodiments, such as Figure 3As shown, it is a schematic diagram of frequency-domain interpolation for the first spectrum. The first spectrum is divided into two parts according to width, namely and , and into two parts according to height, namely and . In total, it is divided into four parts, and zero values are inserted therein to expand the first spectrum with size w×h to w′×h′, that is, the interpolated second spectrum is obtained. In this way, a second spectrum with high resolution can be obtained. and According to height, it is divided into two parts, namely and . and These are two parts, and in total it is divided into four parts. Zero values are inserted therein to expand the first spectrum with size w×h to w′×h′, that is, the interpolated second spectrum is obtained. In this way, a second spectrum with high resolution can be obtained.

[0073] In this way, by using a two-dimensional multi-channel correlation filter to simultaneously estimate the scaling scale and rotation angle of the target object and performing frequency-domain interpolation on the filter output, high-precision estimation of the scale state and rotation state of the target object can be obtained.

[0074] Optionally, performing an inverse discrete Fourier transform on the second spectrum includes: setting a constant for equal ratio amplification of the second spectrum; performing an inverse discrete Fourier transform on the amplified second spectrum.

[0075] Optionally, setting a constant for equal ratio amplification of the interpolated second spectrum, that is In this way, the spectral energy before and after interpolation can be kept unchanged.

[0076] Optionally, performing a two-dimensional inverse discrete Fourier transform calculation on the amplified second spectrum to obtain a correlation filtering result. In this way, a filtering result with high resolution can be obtained.

[0077] Optionally, obtaining the scale scaling factor and / or rotation angle of the target object according to the filtering result includes: determining the scale scaling factor and / or rotation angle of the target object according to the coordinates of the maximum value of the filtering result in the two-dimensional state space.

[0078] Optionally, determining the scale scaling factor and in-plane rotation angle of the target object at the current moment relative to the previous moment through the coordinates of the maximum value of the filtering result in the two-dimensional state space, then obtaining the estimation results of the scale state and rotation state of the target object. In this way, high-precision estimation results of the scale state and rotation state of the target object can be obtained, so as to achieve fast and accurate estimation of the scale and rotation state of the target object under conditions such as fast missile flight speed, intense movement, and background interference, thereby improving the accuracy of target tracking.

[0079] By using a correlation filter to simultaneously estimate the scale state and rotation state of the target object, the estimation results are more accurate and stable, can quickly adapt to changes in the target posture, have strong resistance to noise interference, and significantly improve the calculation efficiency; only sparse sampling needs to be performed on the two-dimensional state space of scale and rotation, and there is no need to repeatedly run the target tracking process multiple times; the estimation of the target scaling scale and rotation angle is independent of the estimation of the target position, applicable to any target tracking framework, and further improves the accuracy of target tracking.

[0080] In some embodiments, the target tracking result includes an estimation of the target object's position and an estimation of the target object's scale and rotation state. At each moment of target tracking, by sequentially performing the target position estimation and the target scale and rotation state estimation, the tracking result at the current moment can be obtained; optionally, by iteratively performing the estimation of the target object's position and the estimation of the target object's scale and rotation state, a more accurate target object tracking result can be obtained.

[0081] Optionally, after obtaining the scale scaling factor and / or rotation angle of the target object according to the filtering result, it further includes:

[0082] Obtaining the scale state of the target object and the rotation state of the target object;

[0083] Updating the filter according to the position of the target object in the image frame, the scale state of the target object, and the rotation state of the target object.

[0084] In some embodiments, learning the optimal two-dimensional multi-channel correlation filter for a single frame includes: using the two-dimensional multi-channel feature as the input, and establishing a two-dimensional state space with the position and scale rotation state of the target object in the current frame as the coordinate origin, and performing discrete sampling and feature extraction to obtain.

[0085] Optionally, obtaining the position, scale state, and rotation state of the target object at any moment, and establishing a two-dimensional state space corresponding to the scale state and rotation state with it as the coordinate origin; performing discrete sampling and feature extraction on the two-dimensional state space to obtain the two-dimensional multi-channel feature g.

[0086] Optionally, constructing the desired output o through a two-dimensional Gaussian distribution. The size of the desired output o is the same as that of the two-dimensional multi-channel feature g, and the two dimensions are independent of each other. The mean of the desired output o and the two-dimensional multi-channel feature g is the central position, and the variance of the desired output o and the two-dimensional multi-channel feature g is obtained according to the number of samples in the two-dimensional state space of the scale state and rotation state;

[0087] Optionally, the optimal two-dimensional multi-channel correlation filter h is composed of multiple channels, where, h l is the correlation filter of the l-th channel. The size of the two-dimensional multi-channel correlation filter h is the same as that of the two-dimensional multi-channel feature g and the desired output o; optionally, the optimal two-dimensional multi-channel correlation filter h is obtained by minimizing the error between the correlation filter output and the desired output;

[0088] By calculating: Obtaining the optimal two-dimensional multi-channel correlation filter h;

[0089] where, h is the optimal two-dimensional multi-channel correlation filter, h lis the correlation filter for the l-th channel, the symbol * represents the correlation operation, o is the desired output, g is the two-dimensional multi-channel feature, d is the number of features, l ∈ {1,..., d}, both l and d are positive integers, is the regularization term weighted by λ, which is used to limit the complexity of the learning model and prevent overfitting.

[0090] Optionally, an efficient analytical solution is obtained in the frequency domain through two-dimensional discrete Fourier transform. Optionally, the correlation filters h of each channel are independent l are represented as H in the frequency domain l ;

[0091] By calculating the correlation filter H of the l-th channel in the frequency domain representation is obtained l ;

[0092] wherein, the capital variables represent the frequency domain representations of the corresponding variables obtained through two-dimensional discrete Fourier transform, and the variables with a bar represent the complex conjugates of the corresponding variables; wherein, H l is the correlation filter of the l-th channel in the frequency domain representation, O is the desired output in the frequency domain representation, is the complex conjugate of the desired output in the frequency domain representation, G is the two-dimensional multi-channel feature in the frequency domain representation, l = 1,..., d, k = 1,......, d, both k, l and d are positive integers, and λ is the weight; wherein, the multiplication and division in the calculation process are matrix dot multiplication and dot division respectively, with high calculation efficiency.

[0093] In this way, the two-dimensional multi-channel correlation filter is updated online through the exponential time weighted average strategy, and the single-frame optimal two-dimensional multi-channel correlation filter learned at the current moment is weighted and summed with the historical model, so that the model can adaptively process the target appearance change. Optionally, the exponential time weighted average strategy is used to update the numerator and denominator in the formula for calculating the single-frame optimal two-dimensional multi-channel correlation filter respectively, so that the online update of the model is more accurate on the premise of slightly increasing the calculation amount.

[0094] In some embodiments, the current image frame tracking state of the target object is determined by the reliability of the target object position estimation result, that is, it depends on the tracking framework used by the seeker image tracker. If the target object position estimation result is reliable, the two-dimensional multi-channel correlation filter used to estimate the scale and rotation state of the target object is updated to adapt to the change of the target object appearance over time, such as the influence of factors such as posture, scale, and illumination, and the resistance to noise interference is enhanced; if the target object position estimation result is unreliable, at this time the target object is likely to be interfered by the complex background or occluded, and the update of the correlation filter should be avoided to reduce the decrease in the estimation accuracy of the subsequent scale rotation state caused by the noise interference of the correlation filter.

[0095] Combined Figure 4 As shown in Figure 4 , an embodiment of the present disclosure provides a device for target tracking, including a processor 100 and a memory 101 storing program instructions. Optionally, the device may further include a communication interface 102 and a bus 103. Among them, the processor 100, the communication interface 102, and the memory 101 can complete mutual communication through the bus 103. The communication interface 102 can be used for information transmission. The processor 100 can call the program instructions in the memory 101 to execute the method for target tracking in the above embodiment.

[0096] In addition, when the program instructions in the above-mentioned memory 101 are implemented in the form of software functional units and sold or used as an independent product, they can be stored in a computer-readable storage medium.

[0097] The memory 101, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as the program instructions / modules corresponding to the method in the embodiment of the present disclosure. The processor 100 executes functional applications and data processing by running the program instructions / modules stored in the memory 101, that is, implements the method for target tracking in the above embodiment.

[0098] The memory 101 may include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the terminal device, etc. In addition, the memory 101 may include a high-speed random access memory and may also include a non-volatile memory.

[0099] By using the device for target tracking provided by the embodiment of the present disclosure, for the scale scaling factor and the planar rotation angle between adjacent image frames, the corresponding two-dimensional state space is obtained; discrete sampling is performed in the two-dimensional state space and feature extraction is performed on each discrete sampling point, the obtained eigenvalues are frequency-domain filtered, and the scale scaling factor and / or rotation angle of the target object are obtained according to the filtering result. In this way, only sparse sampling of the two-dimensional state space of scale and rotation is required, and there is no need to repeatedly run the target tracking process multiple times, significantly improving the calculation efficiency, so that the attitude of the target object can be obtained more efficiently. And an estimation result of the scale state and rotation state of the target object with high precision can be obtained to realize rapid and accurate estimation of the scale and rotation state of the target object under conditions such as high missile flight speed, intense movement, and background interference, thereby improving the accuracy of target tracking.

[0100] An embodiment of the present disclosure provides an imaging seeker including the above-mentioned device for target tracking.

[0101] Optionally, the imaging seeker is provided with an image acquisition module. The imaging seeker acquires images of the target object in real time at a preset acquisition rate through the image acquisition module to obtain image frames of the target object.

[0102] For the imaging seeker, for the scale scaling factor and the planar rotation angle between adjacent image frames, it obtains the corresponding two-dimensional state space; performs discrete sampling on the two-dimensional state space and extracts features from each discrete sampling point, performs frequency domain filtering on the obtained eigenvalues, and obtains the scale scaling factor and / or rotation angle of the target object according to the filtering result. In this way, only sparse sampling of the two-dimensional state space of scale and rotation is required, and it is not necessary to repeatedly run the target tracking process multiple times, which significantly improves the calculation efficiency, and thus can obtain the attitude of the target object more efficiently. And it can obtain high-precision estimation results of the scale state and rotation state of the target object to achieve fast and accurate estimation of the scale and rotation state of the target object under conditions such as high missile flight speed, intense movement, and background interference, thereby improving the accuracy of target tracking.

[0103] The embodiments of the present disclosure provide a computer-readable storage medium storing computer-executable instructions, and the computer-executable instructions are set to execute the above method for target tracking.

[0104] The embodiments of the present disclosure provide a computer program product. The computer program product includes a computer program stored on a computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by a computer, the computer is made to execute the above method for target tracking.

[0105] The above computer-readable storage medium may be a transient computer-readable storage medium or a non-transient computer-readable storage medium.

[0106] The technical solution of the embodiments of the present disclosure can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes one or more instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method of the embodiments of the present disclosure. The foregoing storage medium may be a non-transient storage medium, including: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes, or may also be a transient storage medium.

[0107] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure, enabling those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, process, and other changes. The embodiments merely represent possible variations. Unless explicitly required, the individual components and functions are optional, and the order of operations may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. Moreover, the terms used in this application are only for describing the embodiments and do not limit the claims. As used in the description of the embodiments and the claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to also include the plural forms. Similarly, as used in this application, the term "and / or" refers to any and all possible combinations including one or more of the associated listed items. Additionally, when used in this application, the term "comprise" and its variants "comprises" and / or "comprising" etc. mean the presence of the stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or groups of these. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, or apparatus including the element. Herein, each embodiment may focus on the differences from other embodiments, and the same or similar parts among the embodiments may be referred to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method parts disclosed in the embodiments, the relevant parts may refer to the description of the method parts.

[0108] Those skilled in the art will realize that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software may depend on the specific application and design constraints of the technical solution. The skilled person may use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the embodiments of the present disclosure. The skilled person can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0109] In the embodiments disclosed herein, the disclosed methods, products (including but not limited to devices, equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units can be merely a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Additionally, the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms. The units described as separate components can be or can not be physically separated. The components displayed as units can be or can not be physical units, that is, they can be located in one place or can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to implement this embodiment. Additionally, in the embodiments of the present disclosure, the various functional units can be integrated in one processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0110] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions marked in the blocks can occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, which can depend on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks can also occur in a different order than that disclosed in the description. Sometimes, there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, which can depend on the functions involved. Each block in the block diagram and / or flowchart, as well as the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A method for target tracking, characterized in that, it includes: Obtain an image frame of the target object; For the scale scaling factor and planar rotation angle between adjacent said image frames, obtain their corresponding two-dimensional state space; Perform discrete sampling on the two-dimensional state space and extract features from each discrete sampling point to obtain eigenvalue; Perform frequency-domain filtering on the eigenvalue and obtain the scale scaling factor and / or rotation angle of the target object according to the filtering result; Extract features from each discrete sampling point to obtain eigenvalue, including: extract the image region corresponding to each discrete sampling point from the image frame of the target object; obtain the apparent feature of the image region; represent the apparent feature of the image region as a feature vector; obtain the eigenvalue according to the feature vector; For the scale scaling factor and planar rotation angle between adjacent said image frames, obtain their corresponding two-dimensional state space, including: the scale and rotation of the target object between adjacent image frames have two degrees of freedom of scale scaling factor and planar rotation angle, use a two-dimensional Cartesian coordinate system to describe the scale scaling factor and planar rotation angle between adjacent said image frames, and thus establish a two-dimensional state space corresponding to the scale scaling factor and planar rotation angle.

2. The method according to claim 1, characterized in that, Obtain the scale scaling factor and / or rotation angle of the target object according to the filtering result, including: Determine the scale scaling factor and / or rotation angle of the target object according to the coordinates of the maximum value of the filtering result in the two-dimensional state space.

3. The method according to claim 1, characterized in that, Perform frequency-domain filtering on the eigenvalue, including: Perform discrete Fourier transform on the eigenvalue to obtain the frequency-domain eigenvalue corresponding to the eigenvalue; Perform spectrum multiplication operation on the preset filter and the frequency-domain eigenvalue and sum by channel to obtain the first spectrum; Perform frequency-domain interpolation on the first spectrum to obtain the second spectrum; Perform inverse discrete Fourier transform on the second spectrum to obtain the filtering result.

4. The method according to claim 3, characterized in that, Perform inverse discrete Fourier transform on the second spectrum, including: Enlarge the second spectrum by a set constant in equal ratio; Perform inverse discrete Fourier transform on the enlarged second spectrum.

5. The method according to claim 3 or 4, characterized in that, After obtaining the scale scaling factor and / or rotation angle of the target object according to the filtering result, it further includes: Obtain the scale state of the target object and the rotation state of the target object; Update the filter according to the position of the target object in the image frame, the scale state of the target object and the rotation state of the target object.

6. A device for target tracking, including a processor and a memory storing program instructions, characterized in that, The processor is configured to execute the method for target tracking according to any one of claims 1 to 5 when executing the program instructions.

7. An imaging seeker, characterized in that, It includes the device for target tracking according to claim 6.

Images