CN110969210A - Small and slow target identification and classification method, device, equipment and storage medium - Google Patents
Small and slow target identification and classification method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN110969210A CN110969210A CN201911214450.XA CN201911214450A CN110969210A CN 110969210 A CN110969210 A CN 110969210A CN 201911214450 A CN201911214450 A CN 201911214450A CN 110969210 A CN110969210 A CN 110969210A
- Authority
- CN
- China
- Prior art keywords
- target
- image
- small
- slow
- processed
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000000605 extraction Methods 0.000 claims abstract description 54
- 230000003287 optical effect Effects 0.000 claims abstract description 24
- 238000003709 image segmentation Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims description 24
- 238000007781 pre-processing Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 8
- 238000012937 correction Methods 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 7
- 230000004044 response Effects 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000006872 improvement Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012067 mathematical method Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-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
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a method, a device and equipment for identifying and classifying small and slow targets and a computer readable storage medium, wherein the method comprises the following steps: acquiring an image to be processed, and performing image segmentation on the image to be processed; extracting optical flow field characteristic points and gray level vector characteristic points of small and slow targets in the image to be processed, and fusing to obtain target characteristic points; inputting the target feature points into a pre-trained target extraction model, and receiving the classification credibility probability output by the target extraction model; and determining the target type of the small and slow target according to the classification credibility probability. Therefore, the method can identify and classify the small and slow targets by fusing the optical flow field characteristic points and the gray level vector characteristic points and utilizing the target characteristic points, and because the target characteristic points fuse the characteristics of the optical flow field and the gray level vector, the identification and classification of the small and slow targets by utilizing the target characteristic points are more accurate, and the accuracy and the reliability of the identification and classification are improved.
Description
Technical Field
The present invention relates to the field of image processing, and in particular, to a method, an apparatus, a device, and a computer-readable storage medium for identifying and classifying small and slow targets.
Background
In recent years, with the rapid development of low-altitude detection aircrafts such as unmanned planes, sounding balloons and the like, low-altitude-to-ground video information extraction and analysis is becoming an important means for low-altitude reconnaissance and monitoring. The characteristic points of the small and slow targets in the image frame of the video are extracted to serve as independent identifiable characteristics for classifying the small and slow targets, and the small and slow targets are identified and classified by the characteristic points. Therefore, the accuracy and reliability of the extracted feature points directly affect the accuracy of the small and slow target recognition classification.
In the prior art, methods for extracting features of an object in an image include a background model method, an optical flow method, a frame difference method, and a neural network method based on machine learning. However, the background model method and the inter-frame difference method both require significant changes between the background and the target in the image, and since the small slow target from the air to the ground is acquired in the air long-distance motion, the pixel riddle of the small slow target in the image is extremely low, the inter-frame displacement is small, and the time length is extremely short, so that the robustness of extracting the small slow target is extremely poor, and further the identification and classification of the small slow target are wrong.
Therefore, how to improve the accuracy of feature extraction on the small and slow targets, and further improve the accuracy of identification and classification on the small and slow targets is a technical problem that needs to be solved by those skilled in the art at present.
Disclosure of Invention
In view of this, the present invention provides a method for identifying and classifying small and slow targets, which can improve the accuracy of extracting the features of the small and slow targets, and further improve the accuracy of identifying and classifying the small and slow targets; another object of the present invention is to provide a device, an apparatus and a computer-readable storage medium for identifying and classifying small and slow objects, all of which have the above advantages.
In order to solve the above technical problem, the present invention provides a method for identifying and classifying small and slow targets, comprising:
acquiring an image to be processed, and performing image segmentation on the image to be processed;
extracting optical flow field characteristic points and gray level vector characteristic points of small and slow targets in the image to be processed, and fusing to obtain target characteristic points;
inputting the target feature points into a pre-trained target extraction model, and receiving a classification credibility probability output by the target extraction model;
and determining the target type of the small and slow target according to the classification credibility probability.
Preferably, the process of obtaining the target extraction model comprises:
acquiring a sample image;
performing image segmentation on each sample image;
extracting optical flow field characteristic points and gray level vector characteristic points of the small and slow sample targets in each sample image, and fusing to obtain sample target characteristic points;
and inputting each sample target feature point into a learning network for learning training to obtain the target extraction model.
Preferably, after the step of inputting each sample target feature point into a learning network for learning training to obtain the target extraction model, the method further includes:
and performing secondary training on the target extraction model by using the error sample, and updating the target extraction model by using the secondarily trained model.
Preferably, further comprising:
and correcting the network parameters of the learning network in response to a correction instruction input by an operator.
Preferably, the learning network is specifically a YOLO V3 learning network.
Preferably, after the acquiring the image to be processed, the method further comprises:
and carrying out preprocessing operation on the image to be processed.
Preferably, the pre-processing the image to be processed specifically includes:
and carrying out noise filtering processing and gray scale lifting operation on the image to be processed.
In order to solve the above technical problem, the present invention further provides a device for identifying and classifying small and slow targets, comprising:
the acquisition module is used for acquiring an image to be processed and carrying out image segmentation on the image to be processed;
the extraction module is used for extracting the optical flow field characteristic points and the gray level vector characteristic points of the small and slow targets in the image to be processed, and fusing to obtain target characteristic points;
the input module is used for inputting the target feature points into a pre-trained target extraction model and receiving the classification credibility probability output by the target extraction model;
and the classification module is used for determining the target type of the small and slow target according to the classification credibility probability.
Preferably, further comprising:
and the secondary training module is used for carrying out secondary training on the target extraction model by using the error sample and updating the target extraction model by using the model after secondary training.
Preferably, further comprising:
and the correction module is used for responding to a correction instruction input by an operator and correcting the network parameters of the learning network.
Preferably, further comprising:
and the preprocessing module is used for preprocessing the image to be processed.
In order to solve the above technical problem, the present invention further provides a device for identifying and classifying small and slow targets, including:
a memory for storing a computer program;
and the processor is used for realizing the steps of any one of the small and slow target identification and classification methods when the computer program is executed.
In order to solve the above technical problem, the present invention further provides a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when being executed by a processor, the computer program implements the steps of any of the above methods for identifying and classifying small and slow targets.
The invention provides a method for identifying and classifying small and slow targets, which is characterized in that after an image to be processed is obtained and is subjected to image segmentation, target feature points are obtained by extracting optical flow field feature points and gray level vector feature points of the small and slow targets in the image to be processed and fusing; inputting the target feature points into a pre-trained target extraction model, and receiving the classification credibility probability output by the target extraction model; and determining the target type of the small and slow target according to the classification credibility probability. Therefore, the method can identify and classify the small and slow targets by fusing the optical flow field characteristic points and the gray level vector characteristic points and utilizing the target characteristic points, and because the target characteristic points fuse the characteristics of the optical flow field and the gray level vector, the identification and classification of the small and slow targets by utilizing the target characteristic points are more accurate, and the accuracy and the reliability of the identification and classification are improved.
In order to solve the technical problems, the invention also provides a device, equipment and a computer readable storage medium for identifying and classifying small and slow targets, which have the beneficial effects.
Drawings
In order to more clearly illustrate the embodiments or technical solutions of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a flowchart of a method for identifying and classifying small and slow targets according to an embodiment of the present invention;
FIG. 2 is a block diagram of an apparatus for identifying and classifying small and slow objects according to an embodiment of the present invention;
fig. 3 is a structural diagram of a device for identifying and classifying small and slow objects according to an embodiment of the present invention.
Detailed Description
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 only a part of the embodiments of the present invention, and not all of the embodiments. 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 core of the embodiment of the invention is to provide a method for identifying and classifying small and slow targets, which can improve the accuracy of extracting the characteristics of the small and slow targets, and further improve the accuracy of identifying and classifying the small and slow targets; another core of the present invention is to provide a device, an apparatus and a computer-readable storage medium for identifying and classifying small and slow objects, all having the above advantages.
In order that those skilled in the art will better understand the disclosure, the invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
Fig. 1 is a flowchart of a method for identifying and classifying small and slow targets according to an embodiment of the present invention. As shown in fig. 1, a method for identifying and classifying small and slow targets includes:
s10: and acquiring an image to be processed, and performing image segmentation on the image to be processed.
Firstly, an image to be processed which needs to be subjected to small and slow target recognition is acquired. The image to be processed may be an image obtained by directly shooting with a shooting device, or may be a frame image in a video, and the specific type of the image to be processed is not limited in this embodiment. It should be noted that the image to be processed includes a small slow target and background information, and in this step, the image to be processed is segmented, that is, the small slow target is positioned in the image to be processed, and the small slow target is separated from the image to be processed.
Specifically, image segmentation refers to describing small and slow targets in an image to be processed by a mathematical method; most image segmentation algorithms are based on the discontinuous and similar properties of gray-scale values, and image segmentation methods include image edge segmentation, image threshold segmentation, region-based segmentation and the like. The edge detection algorithm used for image edge segmentation comprises the following steps: roberts operators, Prewitt operators, Sobel operators, Marr-hilberth edge detectors, Canny edge detectors, and the like. In the image threshold segmentation method, the threshold processing method is intuitive, simple to implement and high in calculation speed, so that the threshold processing is in a core position in segmentation application. Region growing algorithms, which are processes that combine pixels or sub-regions into larger regions according to predefined growth criteria, and region splitting and aggregation are both region-based segmentation algorithms.
S20: and extracting the optical flow field characteristic points and the gray level vector characteristic points of the small and slow targets in the image to be processed, and fusing to obtain target characteristic points.
Specifically, the optical flow field is a two-dimensional instantaneous velocity field formed by all pixel points in an image, wherein a two-dimensional velocity vector is a projection of a three-dimensional velocity vector of a visible point in a scene on an imaging surface. The gray level vector characteristic points refer to the characteristics of a gray level histogram of a small slow target in an image to be processed. In this step, after extracting the optical flow field feature points and the gray level vector feature points of the small and slow target in the image to be processed, feature fusion is further performed on the optical flow field feature points and the gray level vector feature points to obtain target feature points.
S30: inputting the target feature points into a pre-trained target extraction model, and receiving the classification credibility probability output by the target extraction model;
s40: and determining the target type of the small and slow target according to the classification credibility probability.
Specifically, after target feature points are obtained, namely, independent identifiable features of the small and slow targets are obtained, then the target feature points are input into a pre-trained target extraction model, each feature vector of the target feature points of the small and slow targets is compared with a feature vector representing each target type in the target extraction model, classification credibility probabilities of the small and slow targets and each target type are obtained through comparison results, namely, the probability that the small and slow targets are each corresponding target type, namely, the target extraction model is calculated, a credibility probability representing the small and slow targets in a given type is distributed to the target feature points, and then the target type of the small and slow targets is determined according to the classification credibility probability. Specifically, when a certain classification reliability probability exceeds a preset threshold, the small and slow target may be determined as a target type corresponding to the classification reliability probability; or the maximum value of the classification credibility probabilities can be determined by comparing the classification credibility probabilities, and then the small and slow target is determined as the target type corresponding to the classification credibility probability.
The embodiment of the invention provides a method for identifying and classifying small and slow targets, which is characterized in that after an image to be processed is obtained and is subjected to image segmentation, target feature points are obtained by extracting optical flow field feature points and gray level vector feature points of the small and slow targets in the image to be processed and fusing; inputting the target feature points into a pre-trained target extraction model, and receiving the classification credibility probability output by the target extraction model; and determining the target type of the small and slow target according to the classification credibility probability. Therefore, the method can identify and classify the small and slow targets by fusing the optical flow field characteristic points and the gray level vector characteristic points and utilizing the target characteristic points, and because the target characteristic points fuse the characteristics of the optical flow field and the gray level vector, the identification and classification of the small and slow targets by utilizing the target characteristic points are more accurate, and the accuracy and the reliability of the identification and classification are improved.
On the basis of the foregoing embodiment, the present embodiment further describes and optimizes the technical solution, and specifically, in the present embodiment, after acquiring the image to be processed, the method further includes:
and carrying out preprocessing operation on the image to be processed.
It can be understood that, in practical operation, in order to reduce the interference of noise and clutter in the image to be processed on the identification and classification of the small slow target in the image to be processed, in this embodiment, after the image to be processed is acquired, a preprocessing operation is further performed on the image to be processed before the image segmentation operation is performed on the image to be processed.
As a preferred embodiment, the pre-processing operation performed on the image to be processed specifically includes:
and carrying out denoising processing and gray scale lifting operation on the image to be processed.
That is to say, in this embodiment, the preprocessing operation on the image to be processed includes denoising processing and gray scale enhancing operation, and the order of the denoising processing and the gray scale enhancing operation is not limited, and is set according to actual requirements. The denoising process refers to filtering out noise in an image to be processed, and the method includes gaussian filtering, mean filtering, median filtering, bilateral filtering, and the like. The gray scale improvement refers to the enhancement of the contrast between the small and slow target and the background image in the image to be processed, the method includes Adaptive Histogram Equalization (AHE), Adaptive Contrast Enhancement (ACE), and the like, and the specific gray scale improvement operation method adopted in the embodiment is not limited.
Therefore, the embodiment can reduce noise and clutter interference in the image to be processed and improve the utilization rate of image information in the image to be processed by further performing preprocessing operation on the image to be processed.
On the basis of the foregoing embodiment, this embodiment further describes and optimizes the technical solution, and specifically, in this embodiment, the process of obtaining the target extraction model includes:
acquiring a sample image;
carrying out image segmentation on each sample image;
extracting optical flow field characteristic points and gray level vector characteristic points of the small and slow sample targets in each sample image, and fusing to obtain sample target characteristic points;
and inputting the target characteristic points of each sample into a learning network for learning training to obtain a target extraction model.
Specifically, in the process of training the target extraction model, a sample image needs to be acquired first. It can be understood that the higher the quality and the larger the number of the sample images, the higher the accuracy of the trained target extraction model. Moreover, since the small slow target refers to a target obtained by shooting based on air-to-ground, that is, the acquired image to be processed is generally top and upper side information of the small slow target based on air-to-ground, in order to further improve the accuracy of identifying and classifying the small slow target in the image to be processed, the sample image in the embodiment is preferably a sample image based on the small slow target based on air-to-ground.
Specifically, after the sample image is acquired, the sample image may be further preprocessed to reduce the influence of noise and clutter interference in the sample image on the model training. Then, image segmentation is carried out on the sample images, optical flow field characteristic points and gray level vector characteristic points of the sample small and slow targets in each sample image are extracted, and the sample target characteristic points are obtained through fusion. These two steps are similar to the processing steps performed on the image to be processed in the previous embodiment, and are not described herein again.
And after sample target characteristic points corresponding to the sample images are obtained, inputting the sample target characteristic points into a learning network for learning training to obtain a target extraction model.
As a preferred embodiment, the learning network is specifically YOLO V3 learning network.
It should be noted that the YOLO learning network is an end-to-end deep learning network structure, and has the characteristics of higher efficiency and higher speed. In addition, compared with the previous version of the YOLO V3 learning network, the YOLO V3 learning network adds a multi-scale feature map similar to the SSD network for detection, and improves the detection precision of small and slow targets, so that the trained target extraction model can more accurately identify the target feature points of each image to be processed, and further improves the accuracy of identification and classification of the small and slow targets in the images to be processed. In addition, the backbone network in the YOLO V3 learning network is the pre-trained model network Darknet53, Darknet53 contains 53 convolutional layers, the method of residual network is used for reference, and a shortcut link is arranged between some layers, so that the efficiency of identification and classification can be further improved.
As a preferred embodiment, after inputting each sample target feature point into a learning network for learning training to obtain a target extraction model, the method further includes:
and performing secondary training on the target extraction model by using the error sample, and updating the target extraction model by using the model after the secondary training.
Specifically, in actual operation, an error sample may be further set, where the error sample refers to a sample opposite to a forward sample library used for training a model, and the target extraction model is subjected to secondary training by further using the error sample, and the target extraction model obtained by using the secondary training has higher robustness and can more accurately extract a small and slow target in the image to be processed, compared with the target extraction model in the previous embodiment.
As a preferred embodiment, further comprising:
and correcting the network parameters of the learning network in response to a correction instruction input by an operator.
In actual operation, in the process of observing the training model in real time, technicians can modify network parameters of the target extraction model according to actual training conditions so as to accelerate the network convergence speed and improve the robustness of the target extraction model.
The above detailed description is given to the embodiment of the method for identifying and classifying small and slow objects provided by the present invention, and the present invention further provides an apparatus, a device, and a computer-readable storage medium for identifying and classifying small and slow objects corresponding to the method.
Fig. 2 is a structural diagram of an apparatus for identifying and classifying a small and slow target according to an embodiment of the present invention, and as shown in fig. 2, the apparatus for identifying and classifying a small and slow target includes:
the acquisition module 21 is configured to acquire an image to be processed and perform image segmentation on the image to be processed;
the extraction module 22 is configured to extract optical flow field feature points and grayscale vector feature points of a small slow target in the image to be processed, and fuse the optical flow field feature points and the grayscale vector feature points to obtain target feature points;
the input module 23 is configured to input the target feature points into a pre-trained target extraction model, and receive a classification reliability probability output by the target extraction model;
and the classification module 24 is used for determining the target type of the small and slow target according to the classification credibility probability.
The device for identifying and classifying the small and slow targets provided by the embodiment of the invention has the beneficial effects of the method for identifying and classifying the small and slow targets.
As a preferred embodiment, further comprising:
and the secondary training module is used for carrying out secondary training on the target extraction model by using the error sample and updating the target extraction model by using the model after the secondary training.
As a preferred embodiment, further comprising:
and the correction module is used for responding to a correction instruction input by an operator and correcting the network parameters of the learning network.
As a preferred embodiment, further comprising:
and the preprocessing module is used for preprocessing the image to be processed.
Fig. 3 is a structural diagram of a device for identifying and classifying small and slow objects according to an embodiment of the present invention, and as shown in fig. 3, the device for identifying and classifying small and slow objects includes:
a memory 31 for storing a computer program;
a processor 32 for implementing the steps of the above method for identifying and classifying small and slow objects when executing the computer program.
The small and slow target identification and classification device provided by the embodiment of the invention has the beneficial effects of the small and slow target identification and classification method.
In order to solve the above technical problem, the present invention further provides a computer-readable storage medium, on which a computer program is stored, and the computer program, when executed by a processor, implements the steps of the above method for identifying and classifying small and slow objects.
The computer-readable storage medium provided by the embodiment of the invention has the beneficial effects of the small and slow target identification and classification method.
The method, the device, the equipment and the computer readable storage medium for identifying and classifying the small and slow targets provided by the invention are described in detail above. The principles and embodiments of the present invention are explained herein using specific examples, which are set forth only to help understand the method and its core ideas of the present invention. It should be noted that, for those skilled in the art, it is possible to make various improvements and modifications to the present invention without departing from the principle of the present invention, and those improvements and modifications also fall within the scope of the claims of the present invention.
The embodiments are described in a progressive manner in the specification, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. The device disclosed by the embodiment corresponds to the method disclosed by the embodiment, so that the description is simple, and the relevant points can be referred to the method part for description.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
Claims (10)
1. A method for identifying and classifying small and slow targets is characterized by comprising the following steps:
acquiring an image to be processed, and performing image segmentation on the image to be processed;
extracting optical flow field characteristic points and gray level vector characteristic points of small and slow targets in the image to be processed, and fusing to obtain target characteristic points;
inputting the target feature points into a pre-trained target extraction model, and receiving a classification credibility probability output by the target extraction model;
and determining the target type of the small and slow target according to the classification credibility probability.
2. The method of claim 1, wherein obtaining the target extraction model comprises:
acquiring a sample image;
performing image segmentation on each sample image;
extracting optical flow field characteristic points and gray level vector characteristic points of the small and slow sample targets in each sample image, and fusing to obtain sample target characteristic points;
and inputting each sample target feature point into a learning network for learning training to obtain the target extraction model.
3. The method of claim 2, wherein after the inputting each sample target feature point into a learning network for learning training to obtain the target extraction model, the method further comprises:
and performing secondary training on the target extraction model by using the error sample, and updating the target extraction model by using the secondarily trained model.
4. The method of claim 2, further comprising:
and correcting the network parameters of the learning network in response to a correction instruction input by an operator.
5. Method according to claim 2, characterized in that said learning network is in particular a YOLO V3 learning network.
6. The method according to any one of claims 1 to 5, further comprising, after said acquiring the image to be processed:
and carrying out preprocessing operation on the image to be processed.
7. The method according to claim 6, wherein the pre-processing the image to be processed specifically comprises:
and carrying out noise filtering processing and gray scale lifting operation on the image to be processed.
8. A device for identifying and classifying small and slow targets is characterized by comprising:
the acquisition module is used for acquiring an image to be processed and carrying out image segmentation on the image to be processed;
the extraction module is used for extracting the optical flow field characteristic points and the gray level vector characteristic points of the small and slow targets in the image to be processed, and fusing to obtain target characteristic points;
the input module is used for inputting the target feature points into a pre-trained target extraction model and receiving the classification credibility probability output by the target extraction model;
and the classification module is used for determining the target type of the small and slow target according to the classification credibility probability.
9. A device for identifying and classifying small and slow targets, comprising:
a memory for storing a computer program;
a processor for implementing the steps of the method for the identification and classification of small and slow objects according to any one of claims 1 to 7 when executing said computer program.
10. A computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, which computer program, when being executed by a processor, carries out the steps of the method for the identification and classification of small and slow objects according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911214450.XA CN110969210A (en) | 2019-12-02 | 2019-12-02 | Small and slow target identification and classification method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911214450.XA CN110969210A (en) | 2019-12-02 | 2019-12-02 | Small and slow target identification and classification method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110969210A true CN110969210A (en) | 2020-04-07 |
Family
ID=70032588
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911214450.XA Pending CN110969210A (en) | 2019-12-02 | 2019-12-02 | Small and slow target identification and classification method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110969210A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112734931A (en) * | 2020-12-31 | 2021-04-30 | 罗普特科技集团股份有限公司 | Method and system for assisting point cloud target detection |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104504365A (en) * | 2014-11-24 | 2015-04-08 | 闻泰通讯股份有限公司 | System and method for smiling face recognition in video sequence |
| CN106503617A (en) * | 2016-09-21 | 2017-03-15 | 北京小米移动软件有限公司 | Model training method and device |
| CN107330384A (en) * | 2017-06-19 | 2017-11-07 | 北京协同创新研究院 | The method and device of motion target tracking in a kind of video |
| CN109409170A (en) * | 2017-08-18 | 2019-03-01 | 广州极飞科技有限公司 | The identifying pest method and apparatus of crops |
| CN109636771A (en) * | 2018-10-23 | 2019-04-16 | 中国船舶重工集团公司第七0九研究所 | Airbound target detection method and system based on image procossing |
| CN110287911A (en) * | 2019-06-28 | 2019-09-27 | 广东工业大学 | A kind of content identification method of invoice, device, equipment and storage medium |
-
2019
- 2019-12-02 CN CN201911214450.XA patent/CN110969210A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104504365A (en) * | 2014-11-24 | 2015-04-08 | 闻泰通讯股份有限公司 | System and method for smiling face recognition in video sequence |
| CN106503617A (en) * | 2016-09-21 | 2017-03-15 | 北京小米移动软件有限公司 | Model training method and device |
| CN107330384A (en) * | 2017-06-19 | 2017-11-07 | 北京协同创新研究院 | The method and device of motion target tracking in a kind of video |
| CN109409170A (en) * | 2017-08-18 | 2019-03-01 | 广州极飞科技有限公司 | The identifying pest method and apparatus of crops |
| CN109636771A (en) * | 2018-10-23 | 2019-04-16 | 中国船舶重工集团公司第七0九研究所 | Airbound target detection method and system based on image procossing |
| CN110287911A (en) * | 2019-06-28 | 2019-09-27 | 广东工业大学 | A kind of content identification method of invoice, device, equipment and storage medium |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112734931A (en) * | 2020-12-31 | 2021-04-30 | 罗普特科技集团股份有限公司 | Method and system for assisting point cloud target detection |
| CN112734931B (en) * | 2020-12-31 | 2021-12-07 | 罗普特科技集团股份有限公司 | Method and system for assisting point cloud target detection |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109272509B (en) | Target detection method, device and equipment for continuous images and storage medium | |
| CN109460754B (en) | A kind of water surface foreign matter detecting method, device, equipment and storage medium | |
| CN110781836A (en) | Human body recognition method and device, computer equipment and storage medium | |
| CN112598713A (en) | Offshore submarine fish detection and tracking statistical method based on deep learning | |
| CN106960178B (en) | Training method of insulator recognition model and insulator recognition and positioning method | |
| CN111833306A (en) | Defect detection method and model training method for defect detection | |
| CN110298281B (en) | Video structuring method and device, electronic equipment and storage medium | |
| CN109242032B (en) | Target detection method based on deep learning | |
| CN113869449A (en) | Model training method, image processing method, device, equipment and storage medium | |
| CN112417955B (en) | Method and device for processing tour inspection video stream | |
| CN111368758A (en) | Face ambiguity detection method and device, computer equipment and storage medium | |
| CN111539957A (en) | Image sample generation method, system and detection method for target detection | |
| CN113989604A (en) | Tire DOT information identification method based on end-to-end deep learning | |
| CN115841633A (en) | Power tower and power line associated correction power tower and power line detection method | |
| KR101690050B1 (en) | Intelligent video security system | |
| CN110969210A (en) | Small and slow target identification and classification method, device, equipment and storage medium | |
| CN113065379A (en) | Image detection method and device fusing image quality and electronic equipment | |
| CN115713633A (en) | Visual SLAM method, system and storage medium based on deep learning in dynamic scene | |
| CN115311680A (en) | Human body image quality detection method and device, electronic equipment and storage medium | |
| CN110751163A (en) | Target positioning method and device, computer readable storage medium and electronic equipment | |
| CN113554685A (en) | Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium | |
| CN114511591B (en) | Track tracking method and device, electronic equipment and storage medium | |
| CN106875369B (en) | Real-time dynamic target tracking method and device | |
| CN115294478B (en) | Aerial unmanned aerial vehicle target detection method applied to modern photoelectric platform | |
| CN117523428B (en) | Ground target detection method and device based on aircraft platform |
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: 20200407 |
|
| RJ01 | Rejection of invention patent application after publication |