CN107767396A - Motion target tracking method, system, equipment and storage medium - Google Patents

Motion target tracking method, system, equipment and storage medium Download PDF

Info

Publication number
CN107767396A
CN107767396A CN201711106488.6A CN201711106488A CN107767396A CN 107767396 A CN107767396 A CN 107767396A CN 201711106488 A CN201711106488 A CN 201711106488A CN 107767396 A CN107767396 A CN 107767396A
Authority
CN
China
Prior art keywords
mrow
particle
msub
pond
msubsup
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.)
Pending
Application number
CN201711106488.6A
Other languages
Chinese (zh)
Inventor
崔苗
张秋镇
林凡
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangdong University of Technology
Original Assignee
Guangdong University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangdong University of Technology filed Critical Guangdong University of Technology
Priority to CN201711106488.6A priority Critical patent/CN107767396A/en
Publication of CN107767396A publication Critical patent/CN107767396A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/215—Motion-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)

Abstract

This application discloses a kind of motion target tracking method, system, equipment and computer-readable recording medium, this method includes：According to default particle ratio, corresponding particle is filtered out in population corresponding to predetermined movement destination image frame sequence and enters hybridization pond, obtain in pond particle assembly outside particle assembly and pond；The individual optimal solution of each particle in particle assembly outside particle assembly in the pond and the pond is updated respectively, obtains corresponding individual optimal solution, to determine the total optimization solution of the population using individual optimal solution；The population total optimization solution is optimized, determines intended particle group's total optimization solution, to realize the tracking to moving target in the movement destination image frame sequence.Motion target tracking method disclosed in the present application improves computational accuracy while accelerating convergence of algorithm speed and avoiding result of calculation from being absorbed in locally optimal solution, makes the accuracy of motion target tracking higher.

Description

Motion target tracking method, system, equipment and storage medium
Technical field
The present invention relates to tracking technique field, more particularly to a kind of motion target tracking method, system, equipment and computer Readable storage medium storing program for executing.
Background technology
Target following technology is always one of focus in computer vision research field, is guided in military affairs, vision guided navigation, The fields such as robot, intelligent transportation, public safety have a wide range of applications.Target following be exactly in continuous video sequence, Moving target interested is found in each frame monitored picture.Due to being had in the processing of visual information and movable information Some puzzlements, such as because the change of light irradiation, the change of difference, angle of background and the effect of shelter and target Change is produced, target following is always the difficult point in computer vision field.
In the prior art, mainly searched for using using the method that particle group optimizing is searched in dbjective state space, To Template matching is carried out, to obtain the tracking result of present frame.Its process is to predict mesh according to the conventional motion conditions of target Possible position in the current frame is marked, each prediction is indicated with a particle in particle swarm optimization algorithm, by target Search space as particle swarm optimization algorithm of position and scope；Secondly, in state search space, with particle group optimizing Searching algorithm finds result of the target-like state value with maximal correlation matching value as target following, and according to tracking result certainly Adapt to more new template and continue sane tracking to realize.This method for tracking target uses the stencil matching optimized algorithm of population Improve the robustness of track algorithm, and greatly reduce the complexity of algorithm, but exist convergence rate is slow, computational accuracy is not high, It is easily trapped into locally optimal solution.
As can be seen here, how a kind of motion target tracking method is provided, to improve convergence of algorithm speed and computational accuracy, Avoid Optimizing operator from being absorbed in local extremum simultaneously, be those skilled in the art's technical problem urgently to be resolved hurrily.
The content of the invention
In view of this, can it is an object of the invention to provide a kind of motion target tracking method, system, equipment and computer Storage medium is read, to improve convergence of algorithm speed and computational accuracy, while avoids Optimizing operator from being absorbed in local extremum.Its is specific Scheme is as follows：
A kind of motion target tracking method, including：
According to default particle ratio, filtered out in population corresponding to predetermined movement destination image frame sequence Corresponding particle enters hybridization pond, obtains in pond particle assembly outside particle assembly and pond；
The individual optimal solution of each particle in particle assembly outside particle assembly in the pond and the pond is carried out more respectively Newly, corresponding individual optimal solution is obtained, to determine the total optimization solution of the population using individual optimal solution；
The population total optimization solution is optimized, determines intended particle group's total optimization solution, to realize to described The tracking of moving target in movement destination image frame sequence.
Preferably, it is described according to default particle ratio, in grain corresponding to predetermined movement destination image frame sequence Before the step of corresponding particle enters hybridization pond is filtered out in subgroup, further comprise：
Background image separating treatment is carried out to predetermined sequence of video images using Three image difference, obtains moving mesh Logo image frame sequence；
It is initial to the particle in the parameter and population in particle cluster algorithm corresponding to the movement destination image frame sequence Position and speed are initialized.
Preferably, it is described that background image separating treatment is carried out to predetermined sequence of video images using Three image difference The step of, including：
Three frame video images of arbitrary continuation in the sequence of video images are obtained, obtain pending picture frame sequence；
Difference processing is carried out to the first picture frame in the pending picture frame sequence and the second picture frame, it is poor to obtain first Partial image；
Difference processing is carried out to the second picture frame in the pending picture frame sequence and the 3rd picture frame, it is poor to obtain second Partial image；
Binary conversion treatment is carried out to first difference image and second difference image respectively, obtains the first binaryzation Image and the second binary image；
To first binary image and second binary image carries out and computing, to realize background image point From.
Preferably, it is described first binary image and second binary image are carried out and the step of computing it Afterwards, in addition to：
Pair with processing after binary image carry out Morphological scale-space.
Preferably, it is described respectively in particle assembly outside particle assembly in the pond and the pond each particle individual most Before the step of excellent solution is updated, including：
Calculate weight of the population during optimal solution is searched for.
Preferably, it is described according to default particle ratio, in grain corresponding to predetermined movement destination image frame sequence The step of corresponding particle enters hybridization pond is filtered out in subgroup, including：
According to the interaction probability of particle in population, in population corresponding to predetermined movement destination image frame sequence In filter out respective numbers particle enter hybridization pond.
Preferably, the predecessor in the pond in particle assembly interacts two-by-two at random, produces the new particle of identical quantity, And the new particle substitutes predecessor, to ensure that the number of particles in the pond in population set is constant.
Preferably, it is described respectively in particle assembly outside particle assembly in the pond and the pond each particle individual most The step of excellent solution is updated, including：
According to the position of each new particle and speed in particle assembly in the following calculation formula calculating pond：
new1(xi)=piold1(xi)+(1.0-pi)old2(xi)；
new2(xi)=piold2(xi)+(1.0-pi)old1(xi)；
In formula, new1(xi) represent new particle new1Position, new2(xi) represent new particle new2Position, old1(vi) Represent predecessor old1Speed, old2(vi) represent predecessor old2Speed, piExpression be evenly distributed on [0,1] with Machine number；
The more new position and speed of each particle in particle assembly outside the pond are calculated according to following calculation formula：
In formula, i=1,2 ..., n,The more new position of particle is represented,Represent particle i d dimensions solutions in kth time iteration The current location in space, d=1,2,Represent the particle i speed that d is tieed up in kth time iteration, r1And r2Represent respectively uniform It is distributed in the random number of [0,1], c1And c2Represent accelerated factor, wkRepresent weight, pidRepresent particle i from initially to current iteration Individual optimal solution caused by number of searches, pgdRepresent the total optimization solution that the population current search arrives.
Preferably, it is described that the population total optimization solution is optimized, determine intended particle group's total optimization solution Step, including：
Step A1：The total optimization solution p for being searched the population at present according to mapping equationgdIt is mapped to logical equation Domain [0,1] on, by total optimization demapping to solution space；Wherein, the mapping equation and the logical equation point It is not：
In formula,Expression is mapped to corresponding initial value during the logical equation, and x represents vector,Represent pgdCorresponding x Value,WithMaximum and minimum value corresponding to variable x are represented respectively；
yn+1=μ yn(1-yn) (n=0,1,2；0≤μ≤4)；
In formula, μ represents the control parameter of the logical equation；
Step A2：WillPass through yn+1=μ yn(1-yn) (n=0,1,2；μ=4) M iteration is carried out, obtain sequence
Step A3：Using inverse mapping equation, by the sequences ykInverse mapping is to former solution space, to generate variable feasible solution sequence RowWherein, the inverse mapping equation is：
In formula, m=1,2,3, M；
Step A4：The adaptive value of each feasible solution vector in the variable feasible solution is calculated, and it is optimal to filter out adaptive value When corresponding feasible solution vector, obtain optimal feasible solution vector；
Step A5：Swear the position for replacing any particle in the population using the position vector of optimal feasible solution vector Amount；
Step A6：If current total optimization solution is less than default allowable error, or iterations exceed it is default most Big iterative steps, then stop iteration, and determine that current total optimization solution is intended particle group's total optimization solution；Otherwise again Into step A3, until reaching default maximum iteration or determining intended particle group's total optimization solution.
Accordingly, the present invention also provides a kind of Motion Object Tracking System, including：
Population grouping module, for according to default particle ratio, in predetermined movement destination image frame sequence Corresponding particle is filtered out in corresponding population and enters hybridization pond, obtains in pond particle assembly outside particle assembly and pond；
Individual optimal solution update module, for respectively to each in particle assembly outside particle assembly in the pond and the pond The individual optimal solution of particle is updated, and obtains corresponding individual optimal solution, to determine the population using individual optimal solution Total optimization solution；
Total optimization solution optimization module, for being optimized to the total optimization solution of the population, determine intended particle Group's total optimization solution, to realize the tracking to moving target in the movement destination image frame sequence.
Preferably, the system further comprises：
Movement destination image frame sequence determining module, for utilizing Three image difference to predetermined sequence of video images Background image separating treatment is carried out, obtains movement destination image frame sequence；
Particle cluster algorithm parameter initialization module, for particle cluster algorithm corresponding to the movement destination image frame sequence In parameter and population in particle initial position and speed initialized.
Accordingly, the present invention also provides a kind of motion target tracking equipment, including：
Processor, realize during for performing the computer program stored in memory and move as described above method for tracking target Step.
Accordingly, the present invention also provides a kind of computer-readable recording medium, is deposited on the computer-readable recording medium Computer program is contained, the step of moving as described above method for tracking target is realized when the computer program is executed by processor.
Motion target tracking method disclosed by the invention, according to default particle ratio, in predetermined moving target Corresponding particle is filtered out in population corresponding to picture frame sequence and enters hybridization pond, obtains in pond particle outside particle assembly and pond Set；Then, the individual optimal solution of each particle in particle assembly outside particle assembly in the pond and the pond is carried out respectively Renewal, obtains corresponding individual optimal solution, to determine the total optimization solution of the population using individual optimal solution；Finally, The population total optimization solution is optimized, determines intended particle group's total optimization solution, to realize to the moving target The tracking of moving target in picture frame sequence.Particle in population is divided into pond particle outside particle and pond by the present invention first, Then the individual optimal solution of particle outside particle in each pond and pond is updated respectively, obtains corresponding individual optimal solution, needed It is noted that particle can be accelerated to calculate convergence rate in pond, particle causes result of calculation to avoid being absorbed in locally optimal solution outside pond. In addition, being optimized to the population total optimization solution, intended particle group's total optimization solution is determined.By overall to population Optimal solution is optimized to obtain intended particle group's total optimization solution of enough satisfaction, i.e., final solution, to realize moving target Tracking.As can be seen here, the bright disclosed motion target tracking method of this law is accelerating convergence of algorithm speed and is avoiding calculating As a result computational accuracy is improved while being absorbed in locally optimal solution, makes the accuracy of motion target tracking higher.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing There is the required accompanying drawing used in technology description to be briefly described, it should be apparent that, drawings in the following description are only this The embodiment of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can also basis The accompanying drawing of offer obtains other accompanying drawings.
Fig. 1 is a kind of motion target tracking method flow chart disclosed by the invention；
Fig. 2 is a kind of specific motion target tracking method flow chart disclosed by the invention；
Fig. 3 separates for sequence of video images background image in a kind of specific motion target tracking method disclosed by the invention Principle flow chart；
Fig. 4 is total optimization solution schematic diagram in a kind of specific motion target tracking method disclosed by the invention；
Fig. 5 is a kind of Motion Object Tracking System structural representation disclosed by the invention；
Fig. 6 is a kind of Motion Object Tracking System structural representation disclosed by the invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made Embodiment, belong to the scope of protection of the invention.
Shown in Figure 1 the embodiment of the invention discloses a kind of motion target tracking method, this method includes：
Step S11：According to default particle ratio, in population corresponding to predetermined movement destination image frame sequence In filter out corresponding particle and enter hybridization pond, obtain in pond particle assembly outside particle assembly and pond.
In the embodiment of the present application, the movement destination image frame sequence is included beforehand through extracting motion mesh after processing Target picture frame, naturally it is also possible to be according to actual conditions and the moving target determined, using capture apparatus, for example, camera or Person's mobile phone is shot to the moving target, obtains pending frame of video, and then carry out respective handling to the frame of video To obtain the movement destination image frame.The default particle ratio can be according to experiment, can also empirically choose Corresponding proportion particle enters hybridization pond, it is necessary to illustrate, hybridizes the half that the number of particles in pond can be population quantity, The ratio of number of particles outside number of particles and pond namely in pond in particle assembly in particle assembly is 1:1.
Step S12：Respectively in particle assembly outside particle assembly in the pond and the pond each particle it is individual optimal Solution is updated, and obtains corresponding individual optimal solution, to determine the total optimization solution of the population using individual optimal solution.
It is well known that in particle cluster algorithm, the individual optimal solution finds optimal in history for each particle Positional information, and the total optimization solution of a population is found from these individual history optimal solutions, and it is overall with history Optimal solution compares, and selects total optimization solution of the optimal total optimization solution as the population.
In the embodiment of the present application, it is that different calculation formula is respectively adopted to come to particle assembly in the pond and the pond The position of each particle in outer particle assembly and speed are updated.That is, the individual optimal solution in the present embodiment includes grain The position of son and speed, accordingly, the total optimization solution of the population also include position and the speed of particle.Need what is illustrated It is that each particle can have individual optimal solution in particle assembly outside particle assembly and the pond in the pond, and particle individual is most Excellent solution will constantly update total optimization solution, and total optimization solution is to belong to whole population, particle assembly and the pond in the pond Outer particle assembly corresponds to same total optimization solution, i.e., the total optimization solution of described population.
It should be noted that in order to lift the ability that population seeks individual optimal solution and total optimization solution, can calculate The firmly weight of particle cluster algorithm, in order to ensure the accuracy during renewal particle position and speed, the embodiment of the present application Further optimization also has been made to the weight.Specifically, the power can be calculated by iteration according to below equation Weight：
In formula, wkRepresent the weight, wmaxAnd wminThe maximum and minimum value of the weight are represented respectively, and k represents to work as Preceding iterations, itermaxRepresent greatest iteration step number.
Step S13：The total optimization solution of the population is optimized, determines intended particle group's total optimization solution, with Realize the tracking to moving target in the movement destination image frame sequence.
It should be noted that the embodiment of the present application optimizes to the total optimization solution of the population, to improve particle The accuracy of group's algorithm, the solution of enough satisfaction is obtained, and then can be determined according to the solution satisfied enough in the motion mesh The current image frame of logo image frame sequence is realized to the moving target to the position of moving target described in next image frame Accurate tracking.
Motion target tracking method disclosed in the embodiment of the present application, first by the particle in population be divided into pond particle and Particle outside pond, then the individual optimal solution of particle outside particle in each pond and pond is updated respectively, obtains corresponding individual Optimal solution, wherein, the renewal of position and speed to particle in pond can be accelerated to calculate convergence rate, position to particle outside pond and The renewal of speed causes result of calculation to avoid being absorbed in locally optimal solution.In addition, the population total optimization solution is optimized, Determine intended particle group's total optimization solution.The target grain of enough satisfaction is obtained by being optimized to population total optimization solution Subgroup total optimization solution, i.e., final solution, to realize the tracking of moving target.As can be seen here, the bright disclosed moving target of this law Tracking improves calculating essence while accelerating convergence of algorithm speed and avoiding result of calculation from being absorbed in locally optimal solution Degree, makes the accuracy of motion target tracking higher.
The embodiment of the invention discloses a kind of specific motion target tracking method, relative to a upper embodiment, this implementation Example has made further instruction and optimization to technical scheme.It is shown in Figure 2, specifically include following steps：
Step S21：Background image separating treatment is carried out to predetermined sequence of video images using Three image difference, obtained To movement destination image frame sequence.
Shown in reference picture 3, the present embodiment can realize the background separation to the video sequence by following principle：
Three frame video images of arbitrary continuation in the sequence of video images are obtained, obtain pending picture frame sequence；It is right The first picture frame and the second picture frame carry out difference processing in the pending picture frame sequence, obtain the first difference image；It is right The second picture frame and the 3rd picture frame carry out difference processing in the pending picture frame sequence, obtain the second difference image；Point It is other that binary conversion treatment is carried out to first difference image and second difference image, obtain the first binary image and second Binary image；To first binary image and second binary image carries out and computing, to realize background image Separation.
It is understood that can also pair with processing after binary image carry out Morphological scale-space, such as with after processing Binary image expanded, holes filling morphological images processing.
Step S22：To in the parameter and population in particle cluster algorithm corresponding to the movement destination image frame sequence Particle initial position and speed are initialized.
It should be noted that the parameter being set by the user is needed in initialization particle cluster algorithm.Particle population size n chooses 20, accelerated factor takes 2, itermaxGreatest iteration step number is 1000, wmax=0.9, wmin=0.4, optimizing number is initialized as 500, x direction offset component Δ x, y directions offset component Δ y, anglec of rotation Δ θ maximal rate component are 30, greatest iteration Number k is 2000, and maximum non-update times are 20.And original position and speed of the n particle in solution space are generated at random.
Step S23：It is corresponding in predetermined movement destination image frame sequence according to the interaction probability of particle in population Population in filter out respective numbers particle enter hybridization pond.
It should be noted that the present embodiment assigns an interactive probability, itself and particle adaptive value to the particle in population It is unrelated, in general, 0.5 can be taken as interaction probability, a number of particle is then randomly selected according to interaction probability and entered Enter to hybridize in pond, such as the half particle entrance hybridization pond of the population can be chosen.
Particle in pond interacts two-by-two at random, produces equal number of new particle, and substitutes primary particle with new particle, To ensure that population number of particles is constant, the predecessor in the pond in particle assembly interacts two-by-two at random, produces identical quantity New particle, and the new particle substitution predecessor, to ensure that the number of particles in the pond in population set is constant.
Step S24：Respectively in particle assembly outside particle assembly in the pond and the pond each particle it is individual optimal Solution is updated, and obtains corresponding individual optimal solution, to determine the total optimization solution of the population using individual optimal solution.
It should be noted that each particle can have individual most in particle assembly outside particle assembly and the pond in the pond Excellent solution, and particle individual optimal solution will constantly update total optimization solution, total optimization solution is to belong to whole population, in the pond Particle assembly corresponds to same total optimization solution, i.e., the total optimization solution of described population outside particle assembly and the pond.
Based on above-mentioned, the step of being updated to the individual optimal solution of particle assembly particle in the pond, is specific as follows：
According to the position of each new particle and speed in particle assembly in the following calculation formula calculating pond：
new1(xi)=piold1(xi)+(1.0-pi)old2(xi)；
new2(xi)=piold2(xi)+(1.0-pi)old1(xi)；
In formula, new1(xi) represent new particle new1Position, new2(xi) represent new particle new2Position, old1(vi) Represent predecessor old1Speed, old2(vi) represent predecessor old2Speed, piExpression be evenly distributed on [0,1] with Machine number；
The step of being updated to the individual optimal solution of particle assembly particle outside the pond is specific as follows：
The more new position and speed of each particle in particle assembly outside the pond are calculated according to following calculation formula：
In formula, i=1,2 ..., n,The more new position of particle is represented,Represent particle i d dimensions solutions in kth time iteration The current location in space, d=1,2,Represent the particle i speed that d is tieed up in kth time iteration, r1And r2Represent respectively uniform It is distributed in the random number of [0,1], c1And c2Represent accelerated factor, wkRepresent weight, pidRepresent particle i from initially to current iteration Individual optimal solution caused by number of searches, pgdRepresent the total optimization solution that the population current search arrives.
Step S25：The total optimization solution of the population is optimized, determines intended particle group's total optimization solution, with Realize the tracking to moving target in the movement destination image frame sequence.
It should be noted that the present embodiment is carried out accordingly by step in detail below to the total optimization solution of the population Optimization：
Step A1：The total optimization demapping that the population searches at present is arrived by logical equation according to mapping equation In domain [0,1], by total optimization demapping to solution space；Wherein, the mapping equation and logical equation difference For：
In formula,Expression is mapped to corresponding initial value during the logical equation, and x represents vector,Represent pgdCorresponding x Value,WithMaximum and minimum value corresponding to variable x are represented respectively；
yn+1=μ yn(1-yn) (n=0,1,2；0≤μ≤4)；
In formula, μ represents the control parameter of the logical equation, ynRepresent the original state amount of the logical equation, yn+1Table Show end-state amount.
Step A2：WillPass through yn+1=μ yn(1-yn) (n=0,1,2；μ=4) M iteration is carried out, obtain sequence
Wherein, n is iterations, for arbitrary n, yn∈ [0,1], μ are an adjustable control parameter, in order to ensure to reflect Penetrate obtained ynIt is always positioned in [0,1], in the present embodiment, the control parameter μ of logical equation value takes 4, certainly Can be other reasonably values, such as 3.9.
Step A3：Using inverse mapping equation, by the sequences ykInverse mapping is to former solution space, to generate variable feasible solution sequence RowWherein, the inverse mapping equation is：
In formula, m=1,2,3, M.
Step A4：The adaptive value of each feasible solution vector in the variable feasible solution is calculated, and it is optimal to filter out adaptive value When corresponding feasible solution vector, obtain optimal feasible solution vector.
Step A5：Swear the position for replacing any particle in the population using the position vector of optimal feasible solution vector Amount.
Step A6：If current total optimization solution is less than default allowable error, or iterations exceed it is default most Big iterative steps, then stop iteration, and determine that current total optimization solution is intended particle group's total optimization solution；Otherwise again Into step A3, until reaching default maximum iteration or determining intended particle group's total optimization solution.
Wherein, the intended particle group total optimization solution includes the x directions offset component Δ being previously mentioned in above-described embodiment X, y directions offset component Δ y, anglec of rotation Δ θ.
It should be noted that different particle individual optimal solutions are not necessarily same dimension, it is therefore desirable to iteration different dimensional Optimal solution on degree, finally determine a unique total optimization solution, i.e. intended particle group total optimization solution.
Shown in reference picture 4, D is made to represent the dimension of search space, i.e., different direction dimensions, xid=(xi1, xi2..., xiD) Represent particle i in the position of D dimension solution spaces, vid=(vi1, vi2..., viD) represent particle i in search space unit iterations Displacement, pid=(pi1, pi2..., piD) show particle i from initially to individual optimal solution, p caused by current iteration number of searchesgd =(pg1, pg2..., pgD) represent the total optimization solution that whole population searches at present.
Motion target tracking method disclosed in the embodiment of the present application, in addition to the beneficial effect for possessing above-described embodiment, Background image separating treatment also is carried out to predetermined sequence of video images by using Three image difference, obtains moving target Picture frame sequence, eliminate the background image in video image and then extract more accurate moving target information, it is ensured that follow-up Calculating process and result precision are higher.
Accordingly, the embodiment of the present application discloses a kind of Motion Object Tracking System, shown in Figure 5, including：
Population grouping module 41, for according to default particle ratio, in predetermined movement destination image frame sequence Corresponding particle is filtered out in population corresponding to row and enters hybridization pond, obtains in pond particle assembly outside particle assembly and pond.
Individual optimal solution update module 42, for respectively to every in particle assembly outside particle assembly in the pond and the pond The individual optimal solution of one particle is updated, and obtains corresponding individual optimal solution, to determine the particle using individual optimal solution The total optimization solution of group.
Total optimization solution optimization module 43, for being optimized to the total optimization solution of the population, determine target grain Subgroup total optimization solution, to realize the tracking to moving target in the movement destination image frame sequence.
It should be noted that in order to ensure the accuracy of the movement destination image, the embodiment of the present application of reference picture 6 may be used also To specifically include：
Movement destination image frame sequence determining module 44, for utilizing Three image difference to predetermined video image sequence Row carry out background image separating treatment, obtain movement destination image frame sequence；
Particle cluster algorithm parameter initialization module 45, for calculating population corresponding to the movement destination image frame sequence Particle initial position in parameter and population and speed in method are initialized.
On the specific work process between modules in the present embodiment refer to moving target disclosed by the invention with Track method, will not be repeated here.
Accordingly, the embodiment of the present application also discloses a kind of motion target tracking equipment, including：
Processor, realized when the processor is used to perform the computer program stored in memory and move as described above target The step of tracking.
It should be noted that the particular content of the present embodiment technology segment can be found in hereinbefore embodiment, herein no longer Repeat.
Accordingly, the embodiment of the present application also discloses a kind of computer-readable recording medium, the computer-readable storage Computer program is stored with medium, is realized when the computer program is executed by processor and moves as described above method for tracking target The step of.Particular content can be found in hereinbefore embodiment, will not be repeated here.
Finally, it is to be noted that, herein, such as first and second or the like relational terms be used merely to by One entity or operation make a distinction with another entity or operation, and not necessarily require or imply these entities or operation Between any this actual relation or order be present.Moreover, term " comprising ", "comprising" or its any other variant meaning Covering including for nonexcludability, so that process, method, article or equipment including a series of elements not only include that A little key elements, but also the other element including being not expressly set out, or also include for this process, method, article or The intrinsic key element of equipment.In the absence of more restrictions, the key element limited by sentence "including a ...", is not arranged Except other identical element in the process including the key element, method, article or equipment being also present.
Above to a kind of motion target tracking method, system, equipment and computer-readable storage medium provided by the present invention Matter is described in detail, and specific case used herein is set forth to the principle and embodiment of the present invention, the above The explanation of embodiment is only intended to help the method and its core concept for understanding the present invention；Meanwhile for the general skill of this area Art personnel, according to the thought of the present invention, there will be changes in specific embodiments and applications, in summary, this Description should not be construed as limiting the invention.

Claims (13)

  1. A kind of 1. motion target tracking method, it is characterised in that including：
    According to default particle ratio, filtered out in population corresponding to predetermined movement destination image frame sequence corresponding Particle enter hybridization pond, obtain in pond particle assembly outside particle assembly and pond；
    The individual optimal solution of each particle in particle assembly outside particle assembly in the pond and the pond is updated respectively, obtained To corresponding individual optimal solution, to determine the total optimization solution of the population using individual optimal solution；
    The population total optimization solution is optimized, determines intended particle group's total optimization solution, to realize to the motion The tracking of moving target in target image frame sequence.
  2. 2. motion target tracking method according to claim 1, it is characterised in that it is described according to default particle ratio, The step of corresponding particle enters hybridization pond is filtered out in population corresponding to predetermined movement destination image frame sequence Before, further comprise：
    Background image separating treatment is carried out to predetermined sequence of video images using Three image difference, obtains movement destination As frame sequence；
    To the particle initial position in the parameter and population in particle cluster algorithm corresponding to the movement destination image frame sequence Initialized with speed.
  3. 3. motion target tracking method according to claim 2, it is characterised in that described to utilize Three image difference to advance The sequence of video images of determination is carried out the step of background image separating treatment, including：
    Three frame video images of arbitrary continuation in the sequence of video images are obtained, obtain pending picture frame sequence；
    Difference processing is carried out to the first picture frame in the pending picture frame sequence and the second picture frame, obtains the first difference diagram Picture；
    Difference processing is carried out to the second picture frame in the pending picture frame sequence and the 3rd picture frame, obtains the second difference diagram Picture；
    Binary conversion treatment is carried out to first difference image and second difference image respectively, obtains the first binary image With the second binary image；
    To first binary image and second binary image carries out and computing, to realize that background image separates.
  4. 4. motion target tracking method according to claim 3, it is characterised in that described to first binary image After the step of second binary image progress and computing, in addition to：
    Pair with processing after binary image carry out Morphological scale-space.
  5. 5. motion target tracking method according to claim 1, it is characterised in that described respectively to particle collection in the pond Before closing the step of being updated with the individual optimal solution of each particle in particle assembly outside the pond, including：
    Calculate weight of the population during optimal solution is searched for.
  6. 6. according to the motion target tracking method described in claim 1 to 5 any one, it is characterised in that the basis is preset Particle ratio, filtered out in population corresponding to predetermined movement destination image frame sequence corresponding particle enter it is miscellaneous The step of handing over pond, including：
    According to the interaction probability of particle in population, sieved in population corresponding to predetermined movement destination image frame sequence The particle for selecting respective numbers enters hybridization pond.
  7. 7. motion target tracking method according to claim 6, it is characterised in that original in particle assembly in the pond Particle interacts two-by-two at random, the new particle of identical quantity, and new particle substitution predecessor is produced, to ensure the pond Number of particles in middle population set is constant.
  8. 8. motion target tracking method according to claim 7, it is characterised in that described respectively to particle collection in the pond The step of being updated with the individual optimal solution of each particle in particle assembly outside the pond is closed, including：
    According to the position of each new particle and speed in particle assembly in the following calculation formula calculating pond：
    new1(xi)=piold1(xi)+(1.0-pi)old2(xi)；
    new2(xi)=piold2(xi)+(1.0-pi)old1(xi)；
    <mrow> <msub> <mi>new</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <msub> <mi>old</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>old</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <mo>|</mo> <msub> <mi>old</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>old</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>|</mo> </mrow> </mfrac> <mo>&amp;CenterDot;</mo> <mo>|</mo> <msub> <mi>old</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>|</mo> <mo>;</mo> </mrow>
    <mrow> <msub> <mi>new</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <msub> <mi>old</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>old</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <mo>|</mo> <msub> <mi>old</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>old</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>|</mo> </mrow> </mfrac> <mo>&amp;CenterDot;</mo> <mo>|</mo> <msub> <mi>old</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>v</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>|</mo> <mo>;</mo> </mrow>
    In formula, new1(xi) represent new particle new1Position, new2(xi) represent new particle new2Position, old1(vi) represent Predecessor old1Speed, old2(vi) represent predecessor old2Speed, piExpression is evenly distributed on the random of [0,1] Number；
    The more new position and speed of each particle in particle assembly outside the pond are calculated according to following calculation formula：
    <mrow> <msubsup> <mi>x</mi> <mrow> <mi>i</mi> <mi>d</mi> </mrow> <mrow> <mi>k</mi> <mo>+</mo> <mn>1</mn> </mrow> </msubsup> <mo>=</mo> <msubsup> <mi>x</mi> <mrow> <mi>i</mi> <mi>d</mi> </mrow> <mi>k</mi> </msubsup> <mo>+</mo> <msubsup> <mi>v</mi> <mrow> <mi>i</mi> <mi>d</mi> </mrow> <mi>k</mi> </msubsup> <mo>;</mo> </mrow>
    <mrow> <msubsup> <mi>v</mi> <mrow> <mi>i</mi> <mi>d</mi> </mrow> <mrow> <mi>k</mi> <mo>+</mo> <mn>1</mn> </mrow> </msubsup> <mo>=</mo> <msup> <mi>w</mi> <mi>k</mi> </msup> <msubsup> <mi>v</mi> <mrow> <mi>i</mi> <mi>d</mi> </mrow> <mi>k</mi> </msubsup> <mo>=</mo> <msub> <mi>c</mi> <mn>1</mn> </msub> <msub> <mi>r</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>p</mi> <mrow> <mi>i</mi> <mi>d</mi> </mrow> </msub> <mo>-</mo> <msubsup> <mi>x</mi> <mrow> <mi>i</mi> <mi>d</mi> </mrow> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>c</mi> <mn>2</mn> </msub> <msub> <mi>r</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <msub> <mi>p</mi> <mrow> <mi>g</mi> <mi>d</mi> </mrow> </msub> <mo>-</mo> <msubsup> <mi>x</mi> <mrow> <mi>i</mi> <mi>d</mi> </mrow> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> <mo>;</mo> </mrow>
    In formula, i=1,2 ..., n,The more new position of particle is represented,Represent particle i d dimensions solution spaces in kth time iteration Current location, d=1,2,Represent the particle i speed that d is tieed up in kth time iteration, r1And r2Represent to be uniformly distributed respectively Random number in [0,1], c1And c2Represent accelerated factor, wkRepresent weight, pidRepresent particle i from initially to current iteration number Individual optimal solution caused by search, pgdRepresent the total optimization solution that the population current search arrives.
  9. 9. motion target tracking method according to claim 8, it is characterised in that described to the population total optimization Solution optimizes, the step of determining intended particle group's total optimization solution, including：
    Step A1：Definition according to mapping equation by the total optimization demapping that the population searches at present to logical equation On domain [0,1], by total optimization demapping to solution space；Wherein, the mapping equation and the logical equation are respectively：
    <mrow> <msubsup> <mi>y</mi> <mn>1</mn> <mi>k</mi> </msubsup> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>x</mi> <mrow> <mi>g</mi> <mi>b</mi> <mi>e</mi> <mi>s</mi> <mi>t</mi> </mrow> <mi>k</mi> </msubsup> <mo>-</mo> <msubsup> <mi>x</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> <mi>k</mi> </msubsup> </mrow> <mrow> <msubsup> <mi>x</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> <mi>k</mi> </msubsup> <mo>-</mo> <msubsup> <mi>x</mi> <mi>min</mi> <mi>k</mi> </msubsup> </mrow> </mfrac> <mo>;</mo> </mrow>
    In formula,Expression is mapped to corresponding initial value during the logical equation, and x represents vector,Represent pgdCorresponding x values,WithMaximum and minimum value corresponding to variable x are represented respectively；
    yn+1=μ yn(1-yn) (n=0,1,2；0≤μ≤4)；
    In formula, μ represents the control parameter of the logical equation；
    Step A2：WillPass through yn+1=μ yn(1-yn) (n=0,1,2；μ=4) M iteration is carried out, obtain sequence
    Step A3：Using inverse mapping equation, by the sequences ykInverse mapping is to former solution space, to generate the feasible solution sequence of variableWherein, the inverse mapping equation is：
    <mrow> <msubsup> <mi>p</mi> <mrow> <mi>g</mi> <mi>d</mi> <mo>,</mo> <mi>m</mi> </mrow> <mrow> <mo>*</mo> <mi>k</mi> </mrow> </msubsup> <mo>=</mo> <msubsup> <mi>x</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> <mi>k</mi> </msubsup> <mo>+</mo> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mrow> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> <mi>k</mi> </msubsup> <mo>-</mo> <msubsup> <mi>x</mi> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> <msubsup> <mi>y</mi> <mi>m</mi> <mi>k</mi> </msubsup> <mo>;</mo> </mrow>
    In formula, m=1,2,3, M；
    Step A4：Calculate the adaptive value of each feasible solution vector in the variable feasible solution, and filter out adaptive value it is optimal when pair The feasible solution vector answered, obtains optimal feasible solution vector；
    Step A5：The position vector of any particle in the population is replaced using the position vector of optimal feasible solution vector；
    Step A6：If current total optimization solution is less than default allowable error, or iterations changes more than default maximum Ride instead of walk number, then stop iteration, and determine that current total optimization solution is intended particle group's total optimization solution；Otherwise reenter Step A3, until reaching default maximum iteration or determining intended particle group's total optimization solution.
  10. A kind of 10. Motion Object Tracking System, it is characterised in that including：
    Population grouping module, for according to default particle ratio, being corresponded in predetermined movement destination image frame sequence Population in filter out corresponding particle and enter hybridization pond, obtain in pond particle assembly outside particle assembly and pond；
    Individual optimal solution update module, for respectively to each particle in particle assembly outside particle assembly in the pond and the pond Individual optimal solution be updated, corresponding individual optimal solution is obtained, to determine the whole of the population using individual optimal solution Body optimal solution；
    Total optimization solution optimization module, for being optimized to the total optimization solution of the population, determine that intended particle group is whole Body optimal solution, to realize the tracking to moving target in the movement destination image frame sequence.
  11. 11. Motion Object Tracking System according to claim 10, it is characterised in that further comprise：
    Movement destination image frame sequence determining module, for being carried out using Three image difference to predetermined sequence of video images Background image separating treatment, obtain movement destination image frame sequence；
    Particle cluster algorithm parameter initialization module, in particle cluster algorithm corresponding to the movement destination image frame sequence Particle initial position and speed in parameter and population are initialized.
  12. A kind of 12. motion target tracking equipment, it is characterised in that including：
    Processor, realize during for performing the computer program stored in memory and moved as described in any one of claim 1 to 9 The step of method for tracking target.
  13. 13. a kind of computer-readable recording medium, it is characterised in that be stored with computer on the computer-readable recording medium Program, the motion target tracking method as described in any one of claim 1 to 9 is realized when the computer program is executed by processor The step of.
CN201711106488.6A 2017-11-10 2017-11-10 Motion target tracking method, system, equipment and storage medium Pending CN107767396A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711106488.6A CN107767396A (en) 2017-11-10 2017-11-10 Motion target tracking method, system, equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711106488.6A CN107767396A (en) 2017-11-10 2017-11-10 Motion target tracking method, system, equipment and storage medium

Publications (1)

Publication Number Publication Date
CN107767396A true CN107767396A (en) 2018-03-06

Family

ID=61273775

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711106488.6A Pending CN107767396A (en) 2017-11-10 2017-11-10 Motion target tracking method, system, equipment and storage medium

Country Status (1)

Country Link
CN (1) CN107767396A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109146922A (en) * 2018-07-11 2019-01-04 哈尔滨工程大学 A kind of Forward-looking Sonar Method for Underwater Target Tracking based on adaptive particle swarm optimization
CN109326006A (en) * 2018-09-30 2019-02-12 百度在线网络技术（北京）有限公司 Map amalgamation method and device

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102831617A (en) * 2012-07-17 2012-12-19 聊城大学 Method and system for detecting and tracking moving object
CN103381603A (en) * 2013-06-29 2013-11-06 湖南大学 Autonomous obstacle crossing programming method of deicing and line inspecting robot for high-voltage transmission line
CN103399257A (en) * 2013-07-31 2013-11-20 武汉大学 Ferromagnetic resonance failure detection method of neutral point ungrounded system
CN104915969A (en) * 2015-05-21 2015-09-16 云南大学 Template matching tracking method based on particle swarm optimization
CN106786662A (en) * 2016-11-15 2017-05-31 国家电网公司 A kind of PSS4B parameter tuning methods based on Hybrid Particle Swarm Optimization

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102831617A (en) * 2012-07-17 2012-12-19 聊城大学 Method and system for detecting and tracking moving object
CN103381603A (en) * 2013-06-29 2013-11-06 湖南大学 Autonomous obstacle crossing programming method of deicing and line inspecting robot for high-voltage transmission line
CN103399257A (en) * 2013-07-31 2013-11-20 武汉大学 Ferromagnetic resonance failure detection method of neutral point ungrounded system
CN104915969A (en) * 2015-05-21 2015-09-16 云南大学 Template matching tracking method based on particle swarm optimization
CN106786662A (en) * 2016-11-15 2017-05-31 国家电网公司 A kind of PSS4B parameter tuning methods based on Hybrid Particle Swarm Optimization

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
ZHIMIN CHEN 等: "A Particle Filter Algorithm Based on Chaos Particle Swarm Optimization and Its Application to Radar Target Tracking", 《JOURNAL OF COMPUTATIONAL INFORMATION SYSTEMS》 *
刘明彦 主编: "《高职教育创新与社会服务 2006-2016年高等职业教育成果荟萃》", 31 August 2017, 北京：北京邮电大学出版社 *
李人厚 主编: "《智能控制理论和方法(第二版)》", 28 February 2013, 西安：西安电子科技大学出版社 *
李娅 等: "全局混沌杂交的粒子群优化算法及应用", 《湖南师范大学自然科学学报》 *
郭巳秋 等: "改进的粒子群优化目标跟踪方法", 《中国光学》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109146922A (en) * 2018-07-11 2019-01-04 哈尔滨工程大学 A kind of Forward-looking Sonar Method for Underwater Target Tracking based on adaptive particle swarm optimization
CN109146922B (en) * 2018-07-11 2021-07-06 哈尔滨工程大学 Forward-looking sonar underwater target tracking method based on adaptive particle swarm optimization
CN109326006A (en) * 2018-09-30 2019-02-12 百度在线网络技术（北京）有限公司 Map amalgamation method and device

Similar Documents

Publication Publication Date Title
CN108090443B (en) Scene text detection method and system based on deep reinforcement learning
CN107766794B (en) Image semantic segmentation method with learnable feature fusion coefficient
CN106127776B (en) It is identified based on multiple features space-time context robot target and moves decision-making technique
CN112132856B (en) Twin network tracking method based on self-adaptive template updating
CN106874914A (en) A kind of industrial machinery arm visual spatial attention method based on depth convolutional neural networks
CN105513093B (en) A kind of method for tracking target represented based on low-rank matrix
JP2020126624A (en) Method for recognizing face using multiple patch combination based on deep neural network and improving fault tolerance and fluctuation robustness
CN110728751A (en) Construction method of indoor 3D point cloud semantic map
CN111125403B (en) Aided design drawing method and system based on artificial intelligence
CN107767396A (en) Motion target tracking method, system, equipment and storage medium
Jiang et al. Dfnet: Semantic segmentation on panoramic images with dynamic loss weights and residual fusion block
Jin et al. Memory-based semantic segmentation for off-road unstructured natural environments
CN106779055A (en) Image characteristic extracting method and device
CN115578416A (en) Unmanned aerial vehicle target tracking method, system, medium and electronic equipment
CN113870312B (en) Single target tracking method based on twin network
CN110321818A (en) A kind of pedestrian detection method in complex scene
CN116258944B (en) Remote sensing image classification model sample increment learning method based on double networks
CN117576149A (en) Single-target tracking method based on attention mechanism
CN116912804A (en) Efficient anchor-frame-free 3-D target detection and tracking method and model
CN116309700A (en) Target tracking general robust anti-attack method based on twin network
CN116843896A (en) Lightweight semantic segmentation method and system based on Transformer
CN113793472B (en) Image type fire detector pose estimation method based on feature depth aggregation network
Schieber et al. Deep sensor fusion with pyramid fusion networks for 3D semantic segmentation
CN115953625A (en) Vehicle detection method based on characteristic diagram double-axis Transformer module
CN114419356A (en) Detection method, system, equipment and storage medium for densely-arranged power equipment

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20180306

RJ01 Rejection of invention patent application after publication