CN113554685A - Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium - Google Patents

Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium Download PDF

Info

Publication number
CN113554685A
CN113554685A CN202110881304.3A CN202110881304A CN113554685A CN 113554685 A CN113554685 A CN 113554685A CN 202110881304 A CN202110881304 A CN 202110881304A CN 113554685 A CN113554685 A CN 113554685A
Authority
CN
China
Prior art keywords
gaussian distribution
video frame
frame image
image
video
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
CN202110881304.3A
Other languages
Chinese (zh)
Inventor
吴昱舟
郭强
王海鹏
许立科
刘传辉
赵凌业
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.)
School Of Aeronautical Combat Service Naval Aeronautical University Of People's Liberation Army
Original Assignee
School Of Aeronautical Combat Service Naval Aeronautical University Of People's Liberation Army
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 School Of Aeronautical Combat Service Naval Aeronautical University Of People's Liberation Army filed Critical School Of Aeronautical Combat Service Naval Aeronautical University Of People's Liberation Army
Priority to CN202110881304.3A priority Critical patent/CN113554685A/en
Publication of CN113554685A publication Critical patent/CN113554685A/en
Pending legal-status Critical Current

Links

Images

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
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G06T7/251—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/254—Analysis of motion involving subtraction of images

Landscapes

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

Abstract

The invention relates to a method and a device for detecting a moving target of a remote sensing satellite, electronic equipment and a storage medium, wherein the method comprises the following steps: acquiring a satellite video; the method comprises the steps of preprocessing a video to obtain a video frame image, processing the video frame image by adopting a multimodal Gaussian distribution model to obtain a first foreground image, processing the video frame image by adopting a three-frame difference method to obtain a second foreground image, and performing AND operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, wherein parameters of the multimodal Gaussian distribution model are updated based on a matching result of pixel values of pixel points in the video frame image and Gaussian distribution in the process of processing the video frame image by adopting the multimodal Gaussian distribution model. The method utilizes the advantage of high operation efficiency of the frame difference method and the characteristics of high stability and accuracy of mixed Gaussian background modeling, reduces the false alarm rate, and realizes reliable detection rate under different satellite video data.

Description

Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium
Technical Field
The invention relates to the field of computer vision and digital image processing, in particular to a method and a device for detecting a moving target of a remote sensing satellite, electronic equipment and a storage medium.
Background
Satellite video moving object detection has many differences: one is satellite video with high frame rate and large breadth, and the requirement of high-speed real-time processing is difficult to meet for the huge data volume; secondly, a plurality of moving targets exist in the satellite video, and the target loss is easily generated in the detection process; thirdly, the number of pixel points of each target is small, the texture features are small, and the detection difficulty is increased, so that the problems in the moving target detection process of the satellite video need to be effectively solved.
The classical satellite video moving object detection method mainly comprises a frame difference method, a background modeling method and an optical flow method, wherein the accuracy and the effect of the background modeling method are better, and the method is one of main methods researched and applied by people. The classical background modeling method comprises a single Gaussian background modeling method, a mixed Gaussian background modeling method, a Bayesian model method, a pixel self-adaptive segmentation method and the like. The main idea of the methods is to establish a background model by using a mathematical statistics principle, then segment foreground pixels and background pixels by model comparison, and update the background model according to the segmentation result. The classical detection algorithm has good effect in the research and application of ground video.
However, for satellite videos, many punctate moving targets exist, and planar rigid moving targets have no texture, different satellite video data have differences in factors such as pixels, background, target state, and the like, and some improvements to background modeling, such as a threshold algorithm of a fixed threshold, may cause a phenomenon of a decrease in detection rate. The characteristics lead to weak applicability of a single moving target detection algorithm and need targeted improvement.
Disclosure of Invention
The invention aims to provide a method and a device for detecting a moving target of a remote sensing satellite, electronic equipment and a storage medium, which are used for solving the problems in the prior art.
In a first aspect, the invention provides a method for detecting a moving target of a remote sensing satellite, which comprises the following steps:
acquiring a satellite video;
the method comprises the steps of preprocessing a video to obtain a video frame image, processing the video frame image by adopting a multimodal Gaussian distribution model to obtain a first foreground image, processing the video frame image by adopting a three-frame difference method to obtain a second foreground image, and performing AND operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, wherein parameters of the multimodal Gaussian distribution model are updated based on a matching result of pixel values of pixel points in the video frame image and Gaussian distribution in the process of processing the video frame image by adopting the multimodal Gaussian distribution model.
Further, the updating the parameters of the multimodal gaussian distribution model based on the matching result of the pixel values of the pixels in the video frame image and the gaussian distribution in the process of processing the video frame image by using the multimodal gaussian distribution model includes:
judging whether the Euclidean distance between the pixel value of a pixel point in the video frame image and the mean value of a first Gaussian distribution is smaller than 2.5 times of the standard deviation of the first Gaussian distribution, wherein the first Gaussian distribution is any one Gaussian distribution in a subset in the multi-peak Gaussian distribution model;
and when the judgment result is yes, updating the mean value and the covariance of the first Gaussian distribution, and when the judgment result is no, updating the weighting coefficients of all the Gaussian distributions in the subset of the multimodal Gaussian distribution model.
Further, the subset in the multi-peak Gaussian distribution model is formed by taking a preset number of Gaussian distributions according to the weighting coefficients in a descending order.
Further, the processing the video frame image by using the multimodal gaussian distribution model to obtain the first foreground image comprises:
and when the judgment result is negative, judging that the pixel point is a component of the first foreground image, thereby obtaining the first foreground image.
Further, the preprocessing the video to obtain a video frame image includes:
and carrying out Gaussian filtering processing on the frame image in the video, and graying to obtain the video frame image.
Further, the processing the video frame image by using a three-frame difference method to obtain a second foreground image includes:
reading continuous three frames of images from the video frame image, respectively performing difference on every two adjacent frames to obtain a difference image, performing binarization processing on the difference image to obtain a binarized image, and performing AND operation on the binarized image to obtain the second foreground image.
Further, still include:
and performing morphological filtering on the preliminary target detection result to obtain a final target detection result.
In a second aspect, the present invention provides a remote sensing satellite moving object detection device, including:
the video acquisition module is used for acquiring a satellite video;
the target detection module is used for preprocessing the video to obtain a video frame image, processing the video frame image by adopting a multimodal Gaussian distribution model to obtain a first foreground image, processing the video frame image by adopting a three-frame difference method to obtain a second foreground image, and performing AND operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, wherein parameters of the multimodal Gaussian distribution model are updated based on a matching result of pixel values of pixel points in the video frame image and Gaussian distribution in the process of processing the video frame image by adopting the multimodal Gaussian distribution model.
In a third aspect, the present invention provides an electronic device, comprising a memory, a processor and a computer program stored on the memory and operable on the processor, wherein the processor implements the steps of the method for detecting a moving object of a remote sensing satellite according to the first aspect when executing the program.
In a fourth aspect, the invention provides a non-transitory computer readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method for remote sensing satellite moving object detection according to the first aspect.
According to the technical scheme, the method, the device, the electronic equipment and the storage medium for detecting the remote sensing satellite moving target provided by the invention have the advantages of high operation efficiency and high detection rate of a frame difference method and the characteristics of good stability and high accuracy of mixed Gaussian background modeling, so that reasonable background parameters can be updated in time, and the false alarm rate is reduced. Meanwhile, the video preprocessing is added to perform noise reduction processing on the video, so that stable and reliable detection rates under different satellite video data are realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of a method for detecting a moving object of a remote sensing satellite according to an embodiment of the invention;
FIG. 2 is a flow chart of a remote sensing satellite moving object detection algorithm according to an embodiment of the invention;
FIG. 3 is a schematic structural diagram of a remote sensing satellite moving object detection device according to an embodiment of the invention;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention provides a satellite video target detection method combining an interframe difference method and a mixed Gaussian background modeling method. By utilizing the advantages of high operation efficiency and high detection rate of the frame difference method and the characteristics of good stability and high accuracy of Gaussian mixture background modeling, reasonable background parameter updating is carried out in time, and the false alarm rate is reduced. And simultaneously, methods of video preprocessing and morphological filtering are added to perform noise reduction processing on the video. And the stable and reliable detection rate under different satellite video data is realized.
Aiming at the particularity of the satellite video moving target detection, the characteristics of high operation efficiency of an interframe difference method, good robustness of a background modeling method and the like are considered, a three-frame difference method which is more accurate than a two-frame difference method is selected, and the satellite video moving target detection method combining the interframe difference method and the background modeling method is provided. The method mainly comprises the steps of video preprocessing, background modeling video moving object detection, three-frame difference video moving object detection, model parameter updating, foreground object extraction, morphological filtering and the like.
Fig. 1 is a flowchart of a method for detecting a moving object of a remote sensing satellite according to an embodiment of the present invention, and referring to fig. 1, the method for detecting a moving object of a remote sensing satellite according to an embodiment of the present invention includes:
step 110, acquiring a satellite video;
step 120, preprocessing the video to obtain a video frame image, processing the video frame image by using a multi-peak gaussian distribution model to obtain a first foreground image, processing the video frame image by using a three-frame difference method to obtain a second foreground image, and performing and operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, wherein parameters of the multi-peak gaussian distribution model are updated based on a matching result of pixel values of pixels in the video frame image and gaussian distribution in the process of processing the video frame image by using the multi-peak gaussian distribution model.
In this embodiment, it should be noted that after the video preprocessing, the frame image after the gaussian filtering is grayed and converted into a grayscale image, the information of each frame image is stored, and then the target of each frame image is sequentially detected. And (3) performing target matching detection by using a self-adaptive model, detecting the foreground by using a three-frame difference method, performing logical AND operation on the two results, and finally obtaining the final detection result by morphological filtering.
Due to the limitation of the frame difference method principle, a large number of false targets can be generated, and the false targets can be detected as moving targets even if the background slightly changes, so that a large number of false targets are removed through logical AND operation, and the stability and the accuracy of detection are improved.
In the algorithm, a multi-peak Gaussian distribution model is adopted, each pixel point of a frame image can be weighted through a certain number of Gaussian distributions, and the needed model is obtained through combination on the basis. The gaussian distributions and the states of the relevant color information have a certain corresponding relationship, and the weights of the distributions also change in the time change process. When data processing is performed on a full-color video frame image, R, G, B channels of all pixels are assumed to have the same variance and the channels are independent of each other.
During detection, if the information of one pixel point can be described by using the corresponding Gaussian distribution, the formula | z-mu is satisfiedk，tIf the value of less than 2.5 sigma is determined as the background and the background model is updated, otherwise, the value is determined as the foreground, wherein z is the pixel value, muk，tσ is the standard deviation of the first gaussian distribution, which is the mean of each gaussian distribution. And (3) carrying out 'AND' on a foreground image obtained by background modeling and a foreground image obtained by multi-frame difference, wherein the obtained result is a primary target detection result of the frame of image. And then applying a digital morphological method to reduce noise. In the method, a 3X 3 'probe' is selected according to actual conditions, and then opening operation processing is carried out before extracting the target. Finally, storing the detection result into the newly-built image sequence, and sequentially circulatingAnd looping all the frames to finally obtain the video with the same time sequence as the original video and the detected moving target.
Fig. 2 is a flowchart of a remote sensing satellite moving object detection algorithm according to an embodiment of the present invention, and referring to fig. 2, the remote sensing satellite moving object detection algorithm provided by the embodiment of the present invention includes:
updating the mean value mu of each Gaussian distribution in the modelk，tCovariance ∑ k, t and corresponding weighting factor ωk，t. If the pixel value xtWith the kth Gaussian distribution GkIs less than 2.5 times GkThen the pixel value x can be considered as the standard deviation oftAnd GkAnd (6) matching.
According to xtWhether the obtained pixel value is successfully matched with the established Gaussian mixture model is divided into the following conditions:
(1) assuming the obtained pixel value xtAt least matching with a certain Gaussian distribution in the model is successful, and the parameters are updated according to the following rules.
For Gaussian distribution without matching, its mean value mu is maintainedk，tAnd the covariance ∑ k, t is unchanged. If the pixel value xtAnd Gaussian distribution GkMatching, then according to the formula (5-1) and the formula (5-2) to GkMean value μk，tAnd the covariance Σ k, t.
μk，t＝(1-ρ)·μk-1，t+ρ·Xt (5-1)
∑k,t＝(1-ρ)·∑k,t-1+ρ·diag[(Xt-μk,t)T(Xt-μk,t)] (5-2)
Wherein:
ρ＝α·Gk(Xt|μk,t-1,Σk,t-1) (5-3)
α is a learning factor for parameter estimation, and the learning factor reflects the degree of importance of the current pixel value.
(2) If the current pixel value x is obtainedtIf the matching with all the distributions fails, the parameter value G of the background with the minimum probability is adjustedj,j＝mini{ωk,t-1K is 1. I.e. replacing omega in K models by the following parametersi/|ΣiParameter of | is given.
ωj，t＝ω0，μj，t＝Xt，∑j，t＝V0 (5-4)
Wherein ω is0Is a predetermined small positive value;
Figure BDA0003192451990000071
i is a unit array, and I is a unit array,
Figure BDA0003192451990000072
is the original variance.
Thereafter, the weighting coefficients ω of all the K Gaussian distributions are updated according to equation (5-5)k，t。
ωk，t＝(1-α)·ωk,t-1+α·(Mk,t) (5-5)
Wherein, alpha is a learning factor, and the magnitude of alpha represents the speed of the change of the model parameters. Pixel value x at time ttAnd Gaussian distribution GkMatch, then order Mk,tIf not, let Mk,tThis means that their corresponding weights will decay when there is no match. And meanwhile, the adaptive capacity required in the model updating is corresponded. However, it is not preferable that the larger the value is, but the more the influence of the false target is given to the increase of the adaptivity, and the accuracy of the target detection is lowered. Usually, a proper value is selected through continuous verification according to a corresponding scene. However, this determination method cannot be dynamically adjusted according to the light change in the video image, and thus cannot be well applied in an environment where the light changes in a complex manner.
The frame difference can quickly reflect the change of the illumination condition, so the algorithm in the chapter judges the change condition of the light by using the frame difference method, thereby adjusting the value of the learning factor in due time. The principle is that a certain small-range area (pixel part of a pure background) in a target area is selected, whether certain pixel points continuously change along with time or not is judged, and therefore the self-adaptability of the algorithm is improved by judging illumination factors.
By adopting the learning factor mechanism, some situations of unmoving objects can be misjudged. If a new target pixel element appears temporarily, the weight value corresponding to Gaussian distribution of the new target pixel element is continuously reduced, and finally the new target pixel element enters a video frame after being processed by other pixels.
Aiming at the particularity of the satellite video moving target detection, the invention provides the satellite video moving target detection method combining the interframe difference method and the background modeling method by selecting the three-frame difference method which is more accurate than the two-frame difference method in consideration of the characteristics of high operation efficiency of the interframe difference method, good robustness of the background modeling method and the like. The method mainly comprises the steps of video preprocessing, background modeling video moving object detection, three-frame difference video moving object detection, model parameter updating, foreground object extraction, morphological filtering and the like.
After video preprocessing, graying the frame image after Gaussian filtering processing, converting the frame image into a grayscale image, storing the image information of each frame, and then sequentially detecting the target of each frame image. And (3) performing target matching detection by using a self-adaptive model, detecting the foreground by using a three-frame difference method, performing logical AND operation on the two results, and finally obtaining the final detection result by morphological filtering.
Due to the limitation of the frame difference method principle, a large number of false targets can be generated, and the false targets can be detected as moving targets even if the background slightly changes, so that a large number of false targets are removed through logical AND operation, and the stability and the accuracy of detection are improved.
According to the method for detecting the moving target of the remote sensing satellite, provided by the embodiment of the invention, reasonable background parameter updating is carried out in time by utilizing the advantages of high operation efficiency and high detection rate of a frame difference method and the characteristics of good stability and high accuracy of mixed Gaussian background modeling, so that the false alarm rate is reduced. And simultaneously, methods of video preprocessing and morphological filtering are added to perform noise reduction processing on the video. And the stable and reliable detection rate under different satellite video data is realized.
The advantages of the present invention can be verified by the following experiments.
Three indexes of accuracy (Correctness), integrity (complexity) and Quality index (Quality) are selected to analyze the detection accuracy, and the calculation formula is as follows:
Correctness＝TP/(TP+FP)
Completeness＝TP/(TP+FN)
Quality＝TP/(TP+FP+FN)
where TP indicates the number of correctly detected foregrounds, FP indicates the number of detected false foregrounds, and FN indicates the number of undetected foregrounds.
Counting target detection results of three algorithms such as a three-frame difference method, a single Gaussian background modeling method and a mixed background modeling method combined with multi-frame difference to obtain evaluation indexes of the three algorithms, wherein the results are as follows:
TABLE 1 comparison of three methods
Figure BDA0003192451990000091
As can be seen from table 1, the integrity of the multi-frame difference method is high, but the accuracy and quality index are low. Therefore, although most of the foreground can be detected when the multi-frame difference method is applied to the satellite video, a large false alarm rate is generated due to background shaking, inter-frame jumping and the like. The single Gaussian background modeling method has high detection accuracy, but low integrity and quality indexes. Namely, the foreground detected by the single Gaussian background modeling method has high accuracy, but a large number of targets are not detected, and the detection rate is low. The algorithm provided by the invention achieves higher levels on three indexes, and multiple experiments show that the method can be suitable for more complex scene transformation and various foreground types, and the more backward the time sequence is, the more accurate the background extraction is, and the importance of updating background model parameters can be seen. In addition, the method has the best effect of detecting the shape of the target and is greatly helpful for judging the type of the target.
Fig. 3 is a schematic diagram of a remote sensing satellite moving object detection device provided in an embodiment of the present invention, and as shown in fig. 3, the remote sensing satellite moving object detection device provided in the embodiment of the present invention includes:
a video acquisition module 310, configured to acquire a satellite video;
the target detection module 320 is configured to pre-process the video to obtain a video frame image, process the video frame image using a multi-peak gaussian distribution model to obtain a first foreground image, process the video frame image using a three-frame difference method to obtain a second foreground image, and perform an and operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, where parameters of the multi-peak gaussian distribution model are updated based on a matching result of a pixel value of a pixel point in the video frame image and a gaussian distribution in a process of processing the video frame image using the multi-peak gaussian distribution model.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
The remote sensing satellite moving object detection device provided by the embodiment of the invention can be used for executing the remote sensing satellite moving object detection method in the embodiment, and the working principle and the beneficial effect are similar, so detailed description is omitted here, and specific contents can be referred to the introduction of the embodiment.
Fig. 4 illustrates a physical structure diagram of an electronic device, which may include, as shown in fig. 4: a processor (processor)410, a communication Interface 420, a memory (memory)430 and a communication bus 440, wherein the processor 410, the communication Interface 420 and the memory 430 are communicated with each other via the communication bus 440. Processor 410 may invoke logic instructions in memory 430 to perform a telemetry satellite moving object detection method comprising: acquiring a satellite video; the method comprises the steps of preprocessing a video to obtain a video frame image, processing the video frame image by adopting a multimodal Gaussian distribution model to obtain a first foreground image, processing the video frame image by adopting a three-frame difference method to obtain a second foreground image, and performing AND operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, wherein parameters of the multimodal Gaussian distribution model are updated based on a matching result of pixel values of pixel points in the video frame image and Gaussian distribution in the process of processing the video frame image by adopting the multimodal Gaussian distribution model.
In addition, the logic instructions in the memory 430 may be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the method for moving object detection of a remote sensing satellite provided by the above methods, the method comprising: acquiring a satellite video; the method comprises the steps of preprocessing a video to obtain a video frame image, processing the video frame image by adopting a multimodal Gaussian distribution model to obtain a first foreground image, processing the video frame image by adopting a three-frame difference method to obtain a second foreground image, and performing AND operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, wherein parameters of the multimodal Gaussian distribution model are updated based on a matching result of pixel values of pixel points in the video frame image and Gaussian distribution in the process of processing the video frame image by adopting the multimodal Gaussian distribution model.
In yet another aspect, the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, the computer program being implemented by a processor to perform the method for detecting a moving object of a remote sensing satellite provided in the above aspects, the method including: acquiring a satellite video; the method comprises the steps of preprocessing a video to obtain a video frame image, processing the video frame image by adopting a multimodal Gaussian distribution model to obtain a first foreground image, processing the video frame image by adopting a three-frame difference method to obtain a second foreground image, and performing AND operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, wherein parameters of the multimodal Gaussian distribution model are updated based on a matching result of pixel values of pixel points in the video frame image and Gaussian distribution in the process of processing the video frame image by adopting the multimodal Gaussian distribution model.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims (10)

1. A remote sensing satellite moving target detection method is characterized by comprising the following steps:
acquiring a satellite video;
the method comprises the steps of preprocessing a video to obtain a video frame image, processing the video frame image by adopting a multimodal Gaussian distribution model to obtain a first foreground image, processing the video frame image by adopting a three-frame difference method to obtain a second foreground image, and performing AND operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, wherein parameters of the multimodal Gaussian distribution model are updated based on a matching result of pixel values of pixel points in the video frame image and Gaussian distribution in the process of processing the video frame image by adopting the multimodal Gaussian distribution model.
2. The method for detecting the remote sensing satellite moving target according to claim 1, wherein the updating parameters of the multimodal Gaussian distribution model based on the matching result of the pixel values of the pixel points in the video frame image and the Gaussian distribution in the process of processing the video frame image by using the multimodal Gaussian distribution model comprises:
judging whether the Euclidean distance between the pixel value of a pixel point in the video frame image and the mean value of a first Gaussian distribution is smaller than 2.5 times of the standard deviation of the first Gaussian distribution, wherein the first Gaussian distribution is any one Gaussian distribution in a subset in the multi-peak Gaussian distribution model;
and when the judgment result is yes, updating the mean value and the covariance of the first Gaussian distribution, and when the judgment result is no, updating the weighting coefficients of all the Gaussian distributions in the subset of the multimodal Gaussian distribution model.
3. The method for detecting the moving target of the remote sensing satellite according to claim 2, wherein the subset in the multi-peak Gaussian distribution model is formed by taking a preset number of Gaussian distributions according to the weighting coefficients in a descending order.
4. The method for detecting the remote sensing satellite moving object according to claim 2, wherein the processing the video frame image by using the multimodal Gaussian distribution model to obtain the first foreground image comprises:
and when the judgment result is negative, judging that the pixel point is a component of the first foreground image, thereby obtaining the first foreground image.
5. The method for remotely sensing a moving object as claimed in claim 1, wherein said preprocessing the video to obtain a video frame image comprises:
and carrying out Gaussian filtering processing on the frame image in the video, and graying to obtain the video frame image.
6. The method for detecting the remote sensing satellite moving target according to claim 1, wherein the processing the video frame image by adopting a three-frame difference method to obtain a second foreground image comprises:
reading continuous three frames of images from the video frame image, respectively performing difference on every two adjacent frames to obtain a difference image, performing binarization processing on the difference image to obtain a binarized image, and performing AND operation on the binarized image to obtain the second foreground image.
7. The method for detecting the moving target of the remote sensing satellite according to claim 1, further comprising:
and performing morphological filtering on the preliminary target detection result to obtain a final target detection result.
8. A remote sensing satellite moving object detection device is characterized by comprising:
the video acquisition module is used for acquiring a satellite video;
the target detection module is used for preprocessing the video to obtain a video frame image, processing the video frame image by adopting a multimodal Gaussian distribution model to obtain a first foreground image, processing the video frame image by adopting a three-frame difference method to obtain a second foreground image, and performing AND operation on the first foreground image and the second foreground image to obtain a preliminary target detection result, wherein parameters of the multimodal Gaussian distribution model are updated based on a matching result of pixel values of pixel points in the video frame image and Gaussian distribution in the process of processing the video frame image by adopting the multimodal Gaussian distribution model.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the steps of the method of moving object detection by a remote sensing satellite according to any of claims 1 to 7 are implemented when the program is executed by the processor.
10. A non-transitory computer-readable storage medium, on which a computer program is stored, wherein the computer program, when being executed by a processor, implements the steps of the method for moving object detection of a remote sensing satellite according to any one of claims 1 to 7.
CN202110881304.3A 2021-08-02 2021-08-02 Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium Pending CN113554685A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110881304.3A CN113554685A (en) 2021-08-02 2021-08-02 Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110881304.3A CN113554685A (en) 2021-08-02 2021-08-02 Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium

Publications (1)

Publication Number Publication Date
CN113554685A true CN113554685A (en) 2021-10-26

Family

ID=78105126

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110881304.3A Pending CN113554685A (en) 2021-08-02 2021-08-02 Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN113554685A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115471502A (en) * 2022-11-01 2022-12-13 山东土之素生物技术有限公司 Animal manure rapid detection method based on computer vision

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102568005A (en) * 2011-12-28 2012-07-11 江苏大学 Moving object detection method based on Gaussian mixture model
CN105354791A (en) * 2015-08-21 2016-02-24 华南农业大学 Improved adaptive Gaussian mixture foreground detection method
WO2017054455A1 (en) * 2015-09-30 2017-04-06 深圳大学 Motion target shadow detection method and system in monitoring video
CN107452005A (en) * 2017-08-10 2017-12-08 中国矿业大学(北京) A kind of moving target detecting method of jointing edge frame difference and gauss hybrid models
CN107507221A (en) * 2017-07-28 2017-12-22 天津大学 With reference to frame difference method and the moving object detection and tracking method of mixed Gauss model
CN108701015A (en) * 2017-11-30 2018-10-23 深圳市大疆创新科技有限公司 For the arithmetic unit of neural network, chip, equipment and correlation technique
CN109949337A (en) * 2019-04-11 2019-06-28 新疆大学 Moving target detecting method and device based on Gaussian mixture model-universal background model
CN111383250A (en) * 2020-03-20 2020-07-07 内蒙古工业大学 Moving target detection method and device based on improved Gaussian mixture model
CN112184759A (en) * 2020-09-18 2021-01-05 深圳市国鑫恒运信息安全有限公司 Moving target detection and tracking method and system based on video

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102568005A (en) * 2011-12-28 2012-07-11 江苏大学 Moving object detection method based on Gaussian mixture model
CN105354791A (en) * 2015-08-21 2016-02-24 华南农业大学 Improved adaptive Gaussian mixture foreground detection method
WO2017054455A1 (en) * 2015-09-30 2017-04-06 深圳大学 Motion target shadow detection method and system in monitoring video
CN107507221A (en) * 2017-07-28 2017-12-22 天津大学 With reference to frame difference method and the moving object detection and tracking method of mixed Gauss model
CN107452005A (en) * 2017-08-10 2017-12-08 中国矿业大学(北京) A kind of moving target detecting method of jointing edge frame difference and gauss hybrid models
CN108701015A (en) * 2017-11-30 2018-10-23 深圳市大疆创新科技有限公司 For the arithmetic unit of neural network, chip, equipment and correlation technique
US20200285446A1 (en) * 2017-11-30 2020-09-10 SZ DJI Technology Co., Ltd. Arithmetic device for neural network, chip, equipment and related method
CN109949337A (en) * 2019-04-11 2019-06-28 新疆大学 Moving target detecting method and device based on Gaussian mixture model-universal background model
CN111383250A (en) * 2020-03-20 2020-07-07 内蒙古工业大学 Moving target detection method and device based on improved Gaussian mixture model
CN112184759A (en) * 2020-09-18 2021-01-05 深圳市国鑫恒运信息安全有限公司 Moving target detection and tracking method and system based on video

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
吴昱舟: "基于帧差和背景建模的卫星视频目标检测", 海军航空工程学院学报, vol. 33, no. 5, pages 441 - 446 *
廖武;赵新民;钱盛友;聂义;: "结合三帧差分与改进混合高斯模型的运动目标", 电脑与信息技术, no. 01, pages 15 - 17 *
魏建猛;陈松;庞首颜;: "改进的混合高斯模型视频运动目标检测算法", 重庆交通大学学报(自然科学版), no. 02, pages 365 - 368 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115471502A (en) * 2022-11-01 2022-12-13 山东土之素生物技术有限公司 Animal manure rapid detection method based on computer vision
CN115471502B (en) * 2022-11-01 2023-03-24 山东土之素生物技术有限公司 Animal manure rapid detection method based on computer vision

Similar Documents

Publication Publication Date Title
WO2021184894A1 (en) Deblurred face recognition method and system and inspection robot
CN108229526B (en) Network training method, network training device, image processing method, image processing device, storage medium and electronic equipment
CN108230291B (en) Object recognition system training method, object recognition method, device and electronic equipment
CN109272509B (en) Target detection method, device and equipment for continuous images and storage medium
CN109271958B (en) Face age identification method and device
CN112597864B (en) Monitoring video anomaly detection method and device
CN109685045B (en) Moving target video tracking method and system
CN110084130B (en) Face screening method, device, equipment and storage medium based on multi-target tracking
CN109840467A (en) A kind of in-vivo detection method and system
CN111723644A (en) Method and system for detecting occlusion of surveillance video
CN110543848B (en) Driver action recognition method and device based on three-dimensional convolutional neural network
US9183671B2 (en) Method for accelerating Monte Carlo renders
CN111160229A (en) Video target detection method and device based on SSD (solid State disk) network
KR20180109658A (en) Apparatus and method for image processing
CN111369521A (en) Image filtering method based on image quality and related device
CN111814690A (en) Target re-identification method and device and computer readable storage medium
CN114724218A (en) Video detection method, device, equipment and medium
CN113554685A (en) Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium
CN113361567A (en) Image processing method, image processing device, electronic equipment and storage medium
CN117409083A (en) Cable terminal identification method and device based on infrared image and improved YOLOV5
CN112818774A (en) Living body detection method and device
CN116994175A (en) Space-time combination detection method, device and equipment for depth fake video
CN116612355A (en) Training method and device for face fake recognition model, face recognition method and device
CN110889817A (en) Image fusion quality evaluation method and device
CN114663347B (en) Unsupervised object instance detection method and unsupervised object instance detection device

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