CN114494819A - Anti-interference infrared target identification method based on dynamic Bayesian network - Google Patents
Anti-interference infrared target identification method based on dynamic Bayesian network Download PDFInfo
- Publication number
- CN114494819A CN114494819A CN202111200199.9A CN202111200199A CN114494819A CN 114494819 A CN114494819 A CN 114494819A CN 202111200199 A CN202111200199 A CN 202111200199A CN 114494819 A CN114494819 A CN 114494819A
- Authority
- CN
- China
- Prior art keywords
- bayesian network
- feature
- time slice
- descriptor
- interference
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 230000003068 static effect Effects 0.000 claims abstract description 22
- 239000013598 vector Substances 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims description 4
- 238000004088 simulation Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 abstract description 5
- 238000013461 design Methods 0.000 abstract description 3
- 230000005855 radiation Effects 0.000 abstract description 2
- 230000007704 transition Effects 0.000 description 6
- 238000012360 testing method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Aiming, Guidance, Guns With A Light Source, Armor, Camouflage, And Targets (AREA)
Abstract
The invention relates to an anti-interference infrared target identification method based on a dynamic Bayesian network, which is used for constructing and updating Bayesian feature vectors. The Bayesian feature vector is mainly formed by combining the radiation intensity of the target and the interference, the appearance contour and other feature abstract representations, so that the characteristic definition of the target and the interference is realized, and a foundation is laid for accurate identification. Feature vector timing association and matching. And on the basis of the completion of the construction of the feature vector, constructing a static Bayesian network by taking the features as a core, and describing the association among the features. Meanwhile, a dynamic Bayesian network framework is constructed by taking the static Bayesian network as an initial network and aiming at the same node characteristics at different moments. And reasoning and identifying under the Bayesian network. In the interference projection process, the target and the interference characteristics are continuously changed, and the characteristic difference between the front moment and the rear moment is small. On the basis of the construction of the current dynamic Bayesian network, target information is identified through characteristic time sequence variation reasoning, so that the identification precision is improved, and the architecture design of the target identification method is completed.
Description
Technical Field
The invention belongs to the field of information processing, and relates to an anti-interference infrared target identification method based on a dynamic Bayesian network.
Background
With the development of modern air combat weaponry, the air combat environment becomes increasingly complex. Only for the infrared air-to-air missile, the target interferes the missile guidance system by releasing artificial baits such as point sources and surface sources which are similar to the target characteristic information, so that the guidance system cannot make accurate and effective judgment. Therefore, the resistance to the artificial interference in the complex scene is the core problem faced by the current infrared air-to-air missile, and the anti-interference capability is of great significance to the infrared guidance system.
Most of the traditional aerial infrared target identification methods firstly extract features, use the extracted features as templates and then verify the extracted features through template matching. Typical features include shape features, local texture features, edge features, and the like. Firstly, people express target information through a single feature, later people fuse and process a plurality of features, and the method can comprehensively analyze and process image information. However, with the increasingly complex environmental background and the emergence of infrared interference, the traditional method is difficult to deal with the contemporary air battle environment.
The application requirements of actual air battlefield countermeasures cannot be met only by considering the traditional feature fusion method, and the time sequence correlation among multiple feature frames needs to be comprehensively considered. In the process of throwing the bait into the target, the bait gradually covers the target in a half or full way from the target position and then is separated. In the process, the real-time change of the features is accelerated, and the complex change relation between the features cannot be expressed by simple feature fusion. Therefore, a multi-feature inter-frame time-series related dynamic bayesian network was introduced as part of the classification system for identifying targets. The dynamic Bayesian network introduces a time dimension on the basis of static Bayes, the characteristics of the target on each time node are represented by a group of variables, the relation among the variables is used for expressing the law that the characteristics of the target change constantly along with the time, and the target is modeled in such a way, so that the stability of target identification is greatly improved.
Therefore, on the basis of deep research on the anti-interference target identification of the traditional infrared air-to-air missile, the anti-interference infrared target identification method based on the dynamic Bayesian network is provided, and the target identification and anti-interference tracking capabilities of the seeker are improved.
Disclosure of Invention
Technical problem to be solved
In order to avoid the defects of the prior art, the invention provides an anti-interference infrared target identification method based on a dynamic Bayesian network, which is used for realizing stable target tracking in an infrared point source interference strong shielding environment.
Technical scheme
An anti-interference infrared target identification method based on a dynamic Bayesian network is characterized by comprising the following steps:
step 1: constructing a Bayesian feature vector of target interference:
step 2: constructing a static Bayesian network, which consists of a directed acyclic graph and a conditional probability table, wherein the directed acyclic graph qualitatively represents the dependency relationship among the variables, and the conditional probability table quantitatively describes the relationship strength among the variables; each node of the network corresponds to a characteristic variable, and a time slice comprises 17 characteristic nodes; according to the relationship between the characteristic nodes and the characteristics and the conditional probability table, the Bayesian network obtains the joint probability distribution of all the characteristic nodes;
establishing an anti-interference simulation data set of the infrared air-to-air missile, and respectively extracting positive and negative sample characteristics to form a characteristic vector Ai={X1,X2,X3,…,X17Forming a positive and negative sample library S+={A1,A2,A3,…,APAnd S-={A1,A2,A3,…,AQThe static bayesian network formed by 17 features is:
constructing a dynamic Bayesian network according to the time sequence correlation of the constructed static Bayesian network: the method comprises the steps that a static Bayesian network is used as an initial network, connection relations of various random variables among time slices are determined through directed arcs, and the initial network is expanded to a dynamic Bayesian network;
training an anti-interference recognition classifier by adopting a dynamic Bayesian network;
and step 3: and inputting the received infrared image into a classifier, and carrying out target identification on the infrared image by the classifier.
The static bayesian network structure of said step 2 is shown in fig. 3.
The dynamic bayesian network structure of step 2 is shown in fig. 4.
Advantageous effects
The invention provides an anti-interference infrared target identification method based on a dynamic Bayesian network, which is used for constructing and updating Bayesian feature vectors. The Bayesian feature vector is mainly formed by combining the radiation intensity of the target and the interference, the appearance outline and other feature abstract representations, so that the target and the interference are defined in a characteristic manner, and a foundation is laid for accurate identification. And (5) associating and matching the feature vector timing. And on the basis of the completion of the construction of the feature vector, constructing a static Bayesian network by taking the features as a core, and describing the association among the features. Meanwhile, a dynamic Bayesian network framework is constructed by taking the static Bayesian network as an initial network and aiming at the same node characteristics at different moments. And reasoning and identification under the Bayesian network. In the interference projection process, the target and the interference characteristics are continuously changed, and the characteristic difference between the front moment and the rear moment is small. On the basis of the construction of the current dynamic Bayesian network, target information is identified through characteristic time sequence variation reasoning, so that the identification precision is improved, and the architecture design of the target identification method is completed.
The invention has the advantages and beneficial effects that: therefore, the problem of identification of a strong shielding target is solved by constructing a dynamic Bayesian network based on the characteristic difference of target interference, the real-time requirement of a guidance system is met, the anti-interference capability of the infrared air-air missile is improved, and the accurate guidance and attack capability is effectively improved.
Drawings
FIG. 1: a flow chart of an anti-interference infrared target identification method based on a dynamic Bayesian network;
FIG. 2: an anti-interference infrared target identification method test chart based on a dynamic Bayesian network;
a: 160 ° left turn (# 52);
b: 160 ° left turn (# 225);
c: 160 ° left turn (# 446);
d: 160 ° left turn (# 545);
e: 100 degree jump (#114)
f: 100 degree jump (#377)
g: 100 degree jump (#609)
h: 100 degree jump (#720)
FIG. 3: static bayesian network structure diagram;
FIG. 4: dynamic bayesian network architecture.
Detailed Description
The invention will now be further described with reference to the following examples and drawings:
according to the method, the target and the interference characteristic vector are constructed mainly through the dynamic Bayesian network, the position of the target under the strong shielding condition is predicted in real time by utilizing the characteristic difference, and the identification method based on the dynamic Bayesian network is obtained, so that the problem of accurate identification of the air combat target is solved. The invention is further described by the following specific implementation method with the attached drawings:
the method comprises the following specific steps:
the method comprises the following steps: through analysis of target interference characteristics, variables capable of distinguishing the target and the interference characteristics are selected to be combined to form a characteristic vector. Seventeen features of length-width ratio, perimeter, energy, area, circularity, gray average value, entropy, Fourier descriptor and the like are selected as feature vectors to represent the characteristics of distinguishing targets from interference, and specific description of the features is shown in table 1.
TABLE 1 characteristics of the process used
Step two: and after the characteristics representing the target and the interference are determined, constructing a static Bayesian network according to the selected characteristics. And then constructing a dynamic Bayesian network according to the time sequence correlation of the constructed static Bayesian network.
Construction of static Bayesian networks
The Bayesian network is composed of a directed acyclic graph and a conditional probability table, wherein the directed acyclic graph qualitatively shows the dependency relationship among variables, and the conditional probability table quantitatively describes the relationship strength among the variables.
In this context, each node of the network corresponds to a feature variable, and a time slice contains 17 feature nodes. According to the feature nodes, the relationship among the features and the conditional probability table, the Bayesian network can obtain the joint probability distribution of all the feature nodes.
Establishing a simulation data set, and respectively extracting positive and negative sample characteristics to form a characteristic vector Ai={X1,X2,X3,…,X17Forming a positive and negative sample library S+={A1,A2,A3,…,APAnd S-={A1,A2,A3,…,AQ}. The static Bayesian network formed by the 17 characteristics of the method is shown in FIG. 3.
Dynamic Bayesian network structure and parameter learning
The dynamic Bayesian network is generally composed of an initial network and a transfer network, the static Bayesian network is used as the initial network in the method, the directed arcs are used for determining how various random variables are related among time slices, the initial network is extended to the dynamic Bayesian network to model the classifier based on the anti-interference recognition, and the structure of the dynamic Bayesian network is shown in FIG. 4.
On the basis, a parameter learning process needs to be completed, and the method adopts a maximum likelihood estimation method.
Let us assume an arbitrary characteristic variable Xi(0. ltoreq. i. ltoreq.N) has riValue case is set, its parent node pa (X)i) The variable has qiIf the value is taken, the parameter vector of the characteristic variable is theta={θijk|i=1,...,N;j=1,...,ri;k=1,...,qiTherein ofThe probability sum of the values of any characteristic variable is 1. Will satisfy p (X)i|pa(Xi) K) is represented as θi,k={θijk|j=1,...,ri}. The parameters to be estimated are:
θijk=p(Xi=j|pa(Xi)=k)(i=1,...,N;j=1,...,ri;k=1,...,qi)
from the log-likelihood estimation method, the following formula can be obtained, where NijkRepresents satisfying thetaijkThe number of samples.
According to the parameter learning process, each feature node X can be calculated respectively1,X2,X3,…,X17The conditional probability table of (2) is shown in table 3, and the conditional probability table is an area node as an example.
TABLE 2 Intra-chip conditional probability table of area feature nodes
TABLE 3 transition probability table for area-characterized nodes
Note: the parent node of the on-chip probabilistic surface area is the perimeter; the parent node of the inter-slice transition probability table area is the perimeter of the current time and the area of the previous time. The transition probability table of the area feature node is a three-dimensional probability table, and table 3 selects the transition probability table of the parent node corresponding to the fifth feature interval of the area at the current time.
Step three: after the dynamic Bayesian network is constructed, the parameters in the transfer network are not changed, so that the network is expanded to the Tth time slice according to the initial distribution and the conditional distribution between the adjacent time slices, and the identification of the target is realized.
Since the discrete dynamic bayesian network conforms to the conditional independence assumption, the probability distribution of the initial time of the dynamic bayesian network can be obtained as follows:
The conditional probability table between two adjacent time slice variables is distributed as follows, wherein,for the ith node on the t-th time slice,is composed ofThe parent node of (2).t > 0 represents the conditional probability distribution of each node in the next time slice. Node pointOf parent nodeWithin either of two time slices.
From the homogeneous assumption, a joint probability distribution across multiple time slices can be derived:
wherein T is the number of time slices, and m is the number of nodes in a single time slice. From this, the bayesian formula of the dynamic bayesian network is:
wherein the content of the first and second substances,represents the degree to which the test data matches class C, and p (C) is the prior probability of class C.
For a set of eigenvaluesAccording to Table 2, Table 3Calculating the intra-chip conditional probability table and the inter-chip transition probability table of each attribute nodePrior probability P (C) of each classi) It can be calculated from training samples, i.e., P (C ═ 1) ═ 0.5334, and P (C ═ 0) ═ 0.4666. And then according to dynamic Bayesian structure calculation (Q is the category of the first time slice):
the same can be calculated:
The sample belongs to the target and otherwise to the interference.
Therefore, the design flow of the target anti-interference identification method based on the dynamic bayesian network is shown in fig. 1:
1)Tipreprocessing and connected region marking are carried out on the frame image, the number of connected regions of the current frame is counted, and the characteristics of the connected regions are extracted to obtain a characteristic value set.
2) And respectively calculating the probability of each class of the characteristic value of each connected region according to the conditional probability table in the time slice and the transition probability table among the time slices obtained by training.
3) And comparing to obtain the probability of each type of the single connected region, wherein the type with the highest probability is the type of the connected region.
Examples effects of implementation
The invention can well construct an anti-interference infrared target identification method based on the dynamic Bayesian network, and can identify the target and the interference with high probability, thereby determining the position of the target, improving the anti-interference performance of the infrared air-to-air missile and realizing accurate attack.
The invention is tested based on the image sequence after background suppression, and the test result is shown in fig. 2.
Claims (3)
1. An anti-interference infrared target identification method based on a dynamic Bayesian network is characterized by comprising the following steps:
step 1: constructing a Bayesian feature vector of target interference:
step 2: constructing a static Bayesian network, which consists of a directed acyclic graph and a conditional probability table, wherein the directed acyclic graph qualitatively represents the dependency relationship among the variables, and the conditional probability table quantitatively describes the relationship strength among the variables; each node of the network corresponds to a characteristic variable, and a time slice comprises 17 characteristic nodes; according to the relationship between the characteristic nodes and the characteristics and the conditional probability table, the Bayesian network obtains the joint probability distribution of all the characteristic nodes;
establishing an anti-interference simulation data set of the infrared air-to-air missile, and respectively extracting positive and negative sample characteristics to form a characteristic vector Ai={X1,X2,X3,…,X17Forming a positive and negative sample library S+={A1,A2,A3,…,APAnd S-={A1,A2,A3,…,AQThe static bayesian network formed by 17 features is:
constructing a dynamic Bayesian network according to the time sequence correlation of the constructed static Bayesian network: the method comprises the steps that a static Bayesian network is used as an initial network, connection relations of various random variables among time slices are determined through directed arcs, and the initial network is expanded to a dynamic Bayesian network;
training an anti-interference recognition classifier by adopting a dynamic Bayesian network;
and step 3: and inputting the received infrared image into a classifier, and carrying out target identification on the infrared image by the classifier.
2. The method for recognizing the anti-interference infrared target based on the dynamic Bayesian network as recited in claim 1, wherein: the static Bayesian network structure of the step 2 is as follows:
a. the initial node is the length-width ratio, and the child nodes are circularity;
b. the circularity child nodes are perimeter, descriptor 1 and descriptor 2;
c. the perimeter sub-node is the area, the area sub-node is the energy, the energy sub-node is the average gray level, and the average gray level sub-node is the entropy;
d. the child node of the descriptor 1 is a descriptor 3, and the child nodes of the descriptor 3 are a descriptor 4 and a descriptor 5;
e. the descriptor 5 child nodes are a descriptor 6 and a descriptor 7;
f. the descriptor 6 child node is a descriptor 8, and the descriptor 8 child node is a descriptor 10;
g. the descriptor 7 child node is the descriptor 9.
3. The method for recognizing the anti-interference infrared target based on the dynamic Bayesian network as recited in claim 1, wherein: the dynamic Bayesian network structure of the step 2 is as follows:
the network structure in the first time slice and the second time slice is the same as the static Bayesian network structure, the network connection mode between the first time slice and the second time slice is corresponding feature connection, such as the aspect ratio feature in the first time slice is connected with the aspect ratio feature in the second time slice, the circularity feature in the first time slice is connected with the circularity feature in the second time slice, the descriptor feature in the first time slice is connected with the descriptor feature in the second time slice, the perimeter feature in the first time slice is connected with the perimeter feature in the second time slice, the area feature in the first time slice is connected with the area feature in the second time slice, the energy feature in the first time slice is connected with the energy feature in the second time slice, the average gray feature in the first time slice is connected with the average gray feature in the second time slice, and the entropy feature in the first time slice is connected with the entropy feature in the second time slice.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111200199.9A CN114494819B (en) | 2021-10-14 | 2021-10-14 | Anti-interference infrared target identification method based on dynamic Bayesian network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111200199.9A CN114494819B (en) | 2021-10-14 | 2021-10-14 | Anti-interference infrared target identification method based on dynamic Bayesian network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114494819A true CN114494819A (en) | 2022-05-13 |
CN114494819B CN114494819B (en) | 2024-03-08 |
Family
ID=81492125
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111200199.9A Active CN114494819B (en) | 2021-10-14 | 2021-10-14 | Anti-interference infrared target identification method based on dynamic Bayesian network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114494819B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115169252A (en) * | 2022-09-07 | 2022-10-11 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Structured simulation data generation system and method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103488886A (en) * | 2013-09-13 | 2014-01-01 | 清华大学 | State threat assessment method based on fuzzy dynamic Bayesian network |
CN110489898A (en) * | 2019-08-26 | 2019-11-22 | 北京航空航天大学 | A kind of dynamic multilayer grade system modelling and trend prediction method based on mixing cognition |
CN112016239A (en) * | 2020-07-07 | 2020-12-01 | 中国科学院西安光学精密机械研究所 | Method for rapidly solving system reliability based on dynamic Bayesian network |
LU102399B1 (en) * | 2019-12-16 | 2021-06-18 | Shandong Technology And Business Univ | Gpr image denoising method and system based on bayesian inference |
-
2021
- 2021-10-14 CN CN202111200199.9A patent/CN114494819B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103488886A (en) * | 2013-09-13 | 2014-01-01 | 清华大学 | State threat assessment method based on fuzzy dynamic Bayesian network |
CN110489898A (en) * | 2019-08-26 | 2019-11-22 | 北京航空航天大学 | A kind of dynamic multilayer grade system modelling and trend prediction method based on mixing cognition |
LU102399B1 (en) * | 2019-12-16 | 2021-06-18 | Shandong Technology And Business Univ | Gpr image denoising method and system based on bayesian inference |
CN112016239A (en) * | 2020-07-07 | 2020-12-01 | 中国科学院西安光学精密机械研究所 | Method for rapidly solving system reliability based on dynamic Bayesian network |
Non-Patent Citations (1)
Title |
---|
唐政;高晓光;: "基于离散动态贝叶斯网络的辐射源目标识别研究", 系统仿真学报, no. 01, 31 January 2009 (2009-01-31) * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115169252A (en) * | 2022-09-07 | 2022-10-11 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Structured simulation data generation system and method |
Also Published As
Publication number | Publication date |
---|---|
CN114494819B (en) | 2024-03-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Liu et al. | Human memory update strategy: a multi-layer template update mechanism for remote visual monitoring | |
CN106960214B (en) | Object recognition method based on image | |
CN107330396B (en) | Pedestrian re-identification method based on multi-attribute and multi-strategy fusion learning | |
Chen et al. | Enhancing detection model for multiple hypothesis tracking | |
CN108596211B (en) | Shielded pedestrian re-identification method based on centralized learning and deep network learning | |
Zheng et al. | Background learnable cascade for zero-shot object detection | |
Bi et al. | Rethinking camouflaged object detection: Models and datasets | |
CN106897390A (en) | Target precise search method based on depth measure study | |
CN113033520B (en) | Tree nematode disease wood identification method and system based on deep learning | |
CN110097060B (en) | Open set identification method for trunk image | |
CN105975932B (en) | Gait Recognition classification method based on time series shapelet | |
CN110533069B (en) | Two-dimensional foil strip distribution characteristic identification method based on support vector machine algorithm | |
CN103778441B (en) | A kind of sequence Aircraft Target Recognition based on DSmT and HMM | |
CN107256017B (en) | Route planning method and system | |
CN109840518B (en) | Visual tracking method combining classification and domain adaptation | |
CN112270286A (en) | Shadow interference resistant monochrome video target tracking method | |
CN111539422A (en) | Flight target cooperative identification method based on fast RCNN | |
Huang et al. | Hybrid bio-inspired lateral inhibition and imperialist competitive algorithm for complicated image matching | |
CN114818853B (en) | Intention recognition method based on bidirectional gating circulating unit and conditional random field | |
CN116109950A (en) | Low-airspace anti-unmanned aerial vehicle visual detection, identification and tracking method | |
CN111444816A (en) | Multi-scale dense pedestrian detection method based on fast RCNN | |
CN114494819A (en) | Anti-interference infrared target identification method based on dynamic Bayesian network | |
CN111144497B (en) | Image significance prediction method under multitasking depth network based on aesthetic analysis | |
Liu et al. | Deep learning of directional truncated signed distance function for robust 3D object recognition | |
JP2017117025A (en) | Pattern identification method, device thereof, and program thereof |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |