CN107578432A - Merge visible ray and the target identification method of infrared two band images target signature - Google Patents
Merge visible ray and the target identification method of infrared two band images target signature Download PDFInfo
- Publication number
- CN107578432A CN107578432A CN201710700308.0A CN201710700308A CN107578432A CN 107578432 A CN107578432 A CN 107578432A CN 201710700308 A CN201710700308 A CN 201710700308A CN 107578432 A CN107578432 A CN 107578432A
- Authority
- CN
- China
- Prior art keywords
- target
- image
- far infrared
- visible ray
- infrared
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The present invention disclose a kind of target identification method for merging visible ray and infrared two band images target signature, and methods described is included to visible ray with, the band image of far infrared two carries out the target registration stage and to visible ray with, the band image of the far infrared two progress target identification stage.This method is a kind of using target overall region as starting point, and visible ray and the target identification method of infrared two band images target signature are merged with being realized on the basis of the infrared same target area of two band images in registering visible ray.
Description
Technical field
Visible ray and the target identification method of infrared two band images target signature are merged the present invention relates to a kind of, belongs to mesh
Identify other technical field.
Background technology
Modernized war active demand round-the-clock can identify Research on Target from complex background.Aircraft, rocket
The heat wave length of the targets of military importance such as bullet, cruise missile has focused largely on 3~5 microns of middle infrared band and 8~10 micro-
In the far infrared band of rice;Under different atmospheric environments, the transmissivity that middle infrared band and far infrared band radiate is different.Can
See that light image has abundant colouring information, texture information, the details of Research on Target can be caught.Infrared image can it is round-the-clock into
Picture, remains the marginal information of target well, and structural information is complete.Middle infrared image is richer than the grain details of far infrared image
Information that is rich, including is more, the high temperature heat radiation concentrated area especially in target;Infrared image is integrally inclined in far infrared image ratio
It is bright.For higher-quality round-the-clock identification target, fusion visible ray and the target identification of infrared two band images target signature
As key technology.
At present, the research of multi-source image fusion is mostly the Pixel-level fusion of feature based Point matching, when different source images
When quantity increases and image background complicates, Feature Points Matching effect is poor, substantial amounts of error hiding characteristic point pair be present, ultimately results in
There is ghost phenomena in Pixel-level fused images, be unfavorable for the target identification in later stage.
The content of the invention
For above-mentioned technical problem, the present invention is intended to provide one kind is using target overall region as starting point, it is visible in registration
Light merges visible ray and infrared two band images target signature with being realized on the basis of the infrared same target area of two band images
Target identification method.
To achieve the above object, the present invention adopts the following technical scheme that:
A kind of target identification method for merging visible ray and infrared two band images target signature, methods described includes pair can
See light with, the band image of far infrared two carry out the target registration stage and to visible ray with, the band image of far infrared two carry out
The target identification stage.
It is described to visible ray with, the band image of far infrared two carry out the target registration stage, first to wave band carry out rough segmentation
Essence registration is carried out after class;
First it will be seen that light image carries out medium filtering, in infrared, far infrared image carry out histogram after first medium filtering
Equilibrium, target area imaging model is established, extract image target area, the spatial gradation histogram for then extracting target area is special
Sign, the target area is carried out to carry out rough sort under its affiliated light source;Secondly according to all kinds of destination numbers included and
Notable feature difference between class and class judges which visible images target class, middle infrared image target class and far infrared image
Target class belongs to same class target;When certain class number of targets be more than or equal to 2 when, extract such each target visible ray, in it is red
Spatial edge direction histogram feature in outer and far infrared image, calculate target visible light image and target far infrared image it
Between similarity, and the similarity in target between infrared image and target far infrared image, respective maximum similarity is corresponding
Two kinds of light source image targets be same target, realize similar target in 3 kinds of light by intermediate variable of target far infrared image
Source hypograph target registration.
It is described to visible ray with, the band image of far infrared two carry out the target identification stage, based on target registration result,
Target visible light and infrared two band images Sample Storehouse are established, fusion feature is calculated, is entered into softmax graders
In, realize fusion visible ray with, the target identification of the band image target signature of far infrared two.
There are the graders such as Logistic, SVM, K-Means, Random Forest, KNN in addition.Softmax graders
The classification of identification is not limited to two classes, and multi-class targets identification can be achieved.Softmax graders are using cross-entropy loss function as cost letter
Number, the probability that test sample belongs to each classification is calculated, classification corresponding to maximum probability is the classification of the test sample.
What Softmax graders were seen is the accuracy rate of classification, higher to the classification sensitivity of mistake, and its loss function value is deposited all the time
, be a The Insatiable grader.Therefore, what in deep learning field, we used is more Softmax graders.
It is described to be to visible ray and the detailed process in infrared two band image progress target registration stage:
Step 1.1, to image preprocessing, it is seen that light image carries out medium filtering, in infrared, far infrared image carry out intermediate value
Filtering and histogram equalization;
Step 1.2, according to visible ray, in infrared and far infrared image imaging characteristicses, determine image target area and background
The feature difference in region, image target area imaging model is established, extract the target area in multi-source image;
Step 1.3, for each single source images, the spatial gradation histogram feature of extraction each of which target area, calculate every
Similitude between individual target area, by each single source images target area rough sort;
Step 1.4, the number of targets that all kinds of targets include in visible images, middle infrared image and far infrared image is compared
Amount, contained number of targets identical visible images target class, middle infrared image target class and far infrared image object class is corresponding
Get up, referred to as the visible images target of same type, middle infrared image target and far infrared image object.If each single source images
The destination number that middle inhomogeneity target includes is consistent, then needs by the big target signature (such as area) of discrimination between all kinds of
To determine which class visible images target, middle infrared image target and far infrared image object are to belong to same type of.
Step 1.5, when in rough sort certain class destination number be more than or equal to 2 when, extract such each target visible ray,
In spatial edge direction histogram feature in infrared and far infrared image;First carry out target far infrared image with it is infrared in target
Image compares, then carries out target far infrared image compared with target visible light image;
Step 1.6, for two kinds of light source image targets of same type, the spatial edge direction histogram for calculating target is similar
Degree.Using far infrared image object as intermediate variable, middle infrared image target corresponding to maximum similarity and the far infrared image mesh
Mark is same target, realizes similar target mid and far infrared image object and middle infrared image target registration;Maximum similarity pair
The visible images target answered and the far infrared image object are same targets, realize similar target mid and far infrared image object
With visible images target registration.Finally realize visible images target in similar target, middle infrared image target and far infrared
Image object registration.
It is described to visible ray with, the band image of far infrared two carry out the target identification stage detailed process be:
Step 2.1, obtained in the step 1.6 registration cross target on the basis of, target establish target visible light with
In, the band image Sample Storehouse of far infrared two, half is as training set, and half is as test set;
Step 2.2, visible ray is automatically extracted with, effective spy of the band image target of far infrared two with convolutional neural networks
Sign, obtains fusion feature, final to realize fusion visible ray and infrared two wave bands figure as the input of softmax graders
As the target identification of target signature.
The present invention has following innovation:
1st, merged target visible ray, in infrared target identification, fully profit are carried out with the band image feature of far infrared three
With the complementary information between target difference source images.
2nd, in visible ray and infrared two band images target registration stage, propose first to carry out target rough segmentation on single source images
Class, then progress visible images target, middle infrared image target are smart registering with far infrared image object in similar target, can
Improve target registration speed.Meanwhile the target rough sort stage is carried out in single source images, different source images imagings can be reduced
Influence caused by characteristic difference.
3rd, the target identification stage has used convolutional neural networks, and it can be automatically extracted in target visible light image, target
The fusion feature of infrared image and target far infrared image, the error of artificial extraction characteristic strip can be reduced.
Brief description of the drawings
Fig. 1 is visible ray of the present invention and infrared two band images target registration method flow block diagram;
Fig. 2 is present invention fusion visible ray and the target identification method FB(flow block) of infrared two band images target signature.
Embodiment
This method is divided into visible ray and infrared two band images target registration and visible ray and infrared two band images target
Two parts are identified, are comprised the following steps that:
(1) visible ray and infrared two band images target registration
When destination number is very big, visible ray and infrared two band images target registration method based on blind search mechanism
Time complexity is also big, so the present invention is proposed based on the visible ray of essence registration and infrared two band images mesh after first rough sort
Mark method for registering:
1. image preprocessing, it is seen that light image carries out medium filtering, in infrared, far infrared image carry out medium filtering and straight
Side's figure is balanced;
2. study visible ray, in infrared and far infrared image imaging characteristicses, specify image target area and background area
Feature difference, establish image target area imaging model;
3. according to visible ray, in infrared and far infrared image target area imaging model, extract multi-source image in target
Region;
4. being directed to each single source images, the spatial gradation histogram feature of extraction each of which target area, each target is calculated
Similitude between region, it is believed that similarity is same class target more than 0.5, realizes the rough segmentation of each single source images target area
Class (target elder generation rough sort can reduce influence of the difference of different source images image-forming principles to registration accuracy in single source images);
5. compare the destination number that all kinds of targets include in visible images, middle infrared image and far infrared image, will
Contained number of targets identical visible images target class, middle infrared image target class and far infrared image object class are mapped,
Referred to as the visible images target of same type, middle infrared image target and far infrared image object.If in each single source images not
The destination number that similar target includes is consistent, then needs by the big target signature (such as area) of discrimination between all kinds of come really
Fixed which class visible images target, middle infrared image target and far infrared image object are to belong to same type of.
6. when in rough sort certain class destination number be more than or equal to 2 when, extract such each target visible ray, in it is infrared
With the spatial edge direction histogram feature in far infrared image;First carry out target far infrared image and infrared image ratio in target
Compared with, then target far infrared image is carried out compared with target visible light image;
7. for same type far infrared image object and middle infrared image target, calculate some far infrared image object with
The spatial edge direction histogram similarity of all middle infrared image targets, middle infrared image target corresponding to maximum similarity with
The far infrared image object is same target, realizes that similar target mid and far infrared image object and middle infrared image target are matched somebody with somebody
It is accurate.Similarly, similar target mid and far infrared image object and visible images target registration are realized, finally realizes quick visible ray
With infrared two band images target registration.
Far infrared image is first carried out compared with middle infrared image, then carries out far infrared image compared with visible images, will
Far infrared image is as intermediate quantity.Here visible images target class, middle infrared image target corresponding to referring to more than or equal to 2
The destination number that class and far infrared image object class include is more than or equal to 2, just needs to utilize target image in this case
Spatial edge direction histogram carries out further registration, if the target number that corresponding three source images target class includes is 1,
Visible images, middle infrared image and far infrared image are just registered upper so corresponding to this target, it is not necessary to enter one
The utilization space edge orientation histogram of step carries out registration.Each target include visible ray, in it is infrared with 3 kinds of light sources of far infrared
Image.Comparison is all based on the comparison that two kinds of light sources belong between same type target, is all two two and compares, pair of 3 kinds of light sources
Than being realized by regarding far infrared image as intermediate quantity.
(2) visible ray and infrared two band images target identification
Completed visible ray with infrared two band image same target it is registering, extract on this basis same
The visible ray of target is used for target identification with infrared two band images feature.Visible ray and infrared two band images target identification
Concretely comprise the following steps:
1. establishing target visible light and infrared two band images Sample Storehouse, half is as training set, and half is as test set;
2. automatically extracting the validity feature of visible ray and infrared two band images target with convolutional neural networks, merged
Feature, it is final to realize fusion visible ray and infrared two band images target signature as the input of softmax graders
Target identification.The present embodiment is using softmax graders, also Logistic, SVM, K-Means, Random in addition
The graders such as Forest, KNN.The classification of Softmax graders identification is not limited to two classes, and multi-class targets identification can be achieved.
Softmax graders calculate the probability that test sample belongs to each classification, maximum probability using cross-entropy loss function as cost function
Corresponding classification is the classification of the test sample.What Softmax graders were seen is the accuracy rate of classification, to wrong classification
Sensitivity is higher, and its loss function value exists all the time, is the grader of a The Insatiable.Therefore, led in deep learning
It is more Softmax graders that we, which use, in domain.
The thinking has feasibility, the inventive method is used to study among human body performance-based objective identification process, to data
Handled, being input to convolutional neural networks in convolutional neural networks as input data carries out Local Features Analysis, obtains spy
Levy output item, be directly inputted in Softmax graders, can recognize that walk, runnings, stair activity, 5 kinds of standings etc. act, knowledge
Not rate is up to 84.8%.Equally, the inventive method can also be applied to Finger print identification field, the ability of feature extraction from
A large amount of finger print data learnings are input to Softmax graders, fingerprint image classification are obtained, in International Publication to line type feature
The classification accuracy of line type four is measured on finger print data collection NISTDB4 up to 94.2%.In addition, apply the inventive method to clap on the spot
For the visible ray taken the photograph with, in the band image target identification of far infrared two, target includes car, motorcycle, boy student and schoolgirl, target
Discrimination is up to 96.8%
Described above is only the preferred embodiment of the present invention, it should be pointed out that:For the ordinary skill people of the art
For member, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvements and modifications also should
It is considered as protection scope of the present invention.
Claims (8)
1. a kind of merge visible ray and the target identification method of infrared two band images target signature, it is characterised in that the side
Method include to visible ray with, the band image of far infrared two carry out the target registration stage and to visible ray with, the ripple of far infrared two
Section image carries out the target identification stage.
A kind of 2. target identification side for merging visible ray and infrared two band images target signature according to claim 1
Method, it is characterised in that it is described to visible ray with, the band image of far infrared two carry out the target registration stage, first wave band is entered
Essence registration is carried out after row rough sort;
First it will be seen that light image carries out medium filtering, in infrared, far infrared image carry out histogram equalization after first medium filtering,
Target area imaging model is established, extracts image target area, then extracts the spatial gradation histogram feature of target area,
The target area is carried out to carry out rough sort under its affiliated light source;Secondly according to all kinds of destination numbers included and class with
Notable feature difference between class judges which visible images target class, middle infrared image target class and far infrared image object
Class belongs to same class target;Extract certain a kind of each target visible ray, in spatial edge side in infrared and far infrared image
To histogram feature, the similarity between target visible light image and target far infrared image, and infrared figure in target are calculated
Picture and the similarity between target far infrared image, two kinds of light source image targets corresponding to respective maximum similarity are same mesh
Mark, realizes similar target in 3 kinds of light source hypograph target registrations by intermediate variable of target far infrared image.
A kind of 3. target identification side for merging visible ray and infrared two band images target signature according to claim 1
Method, it is characterised in that it is described to visible ray with, the band image of far infrared two carry out the target identification stage, matched somebody with somebody based on target
Quasi- result, target visible light and infrared two band images Sample Storehouse are established, fusion feature is calculated, is entered into grader
In, realize fusion visible ray with, the target identification of the band image target signature of far infrared two.
A kind of 4. target identification side for merging visible ray and infrared two band images target signature according to claim 2
Method, it is characterised in that described to be to visible ray and the detailed process in infrared two band image progress target registration stage:
Step 1.1, to image preprocessing, it is seen that light image carries out medium filtering, in infrared, far infrared image carry out medium filtering
And histogram equalization;
Step 1.2, according to visible ray, in infrared and far infrared image imaging characteristicses, determine image target area and background area
Feature difference, establish image target area imaging model, extract the target area in multi-source image;
Step 1.3, for each single source images, the spatial gradation histogram feature of extraction each of which target area, each mesh is calculated
The similitude between region is marked, by each single source images target area rough sort;
Step 1.4, the destination number that all kinds of targets include in visible images, middle infrared image and far infrared image is compared,
Contained number of targets identical visible images target class, middle infrared image target class and far infrared image object class are corresponded to
Come, referred to as the visible images target of same type, middle infrared image target and far infrared image object;
Step 1.5, when in rough sort certain class destination number be more than or equal to 2 when, extract such each target visible ray, in it is red
Spatial edge direction histogram feature in outer and far infrared image, first carry out target far infrared image and infrared image in target
Compare, then carry out target far infrared image compared with target visible light image;
Step 1.6, for two kinds of light source image targets of same type, the spatial edge direction histogram similarity of target is calculated;
Using far infrared image object as intermediate variable, middle infrared image target corresponding to maximum similarity is with the far infrared image object
Same target, realize similar target mid and far infrared image object and middle infrared image target registration;Corresponding to maximum similarity
Visible images target and the far infrared image object are same targets, realize similar target mid and far infrared image object and can
See light pattern objects registration, finally realize visible images target in similar target, middle infrared image target and far infrared image
Target registration.
A kind of 5. target identification side for merging visible ray and infrared two band images target signature according to claim 3
Method, it is characterised in that it is described to visible ray with, the band image of far infrared two carry out the target identification stage detailed process
For:
Step 2.1, on the basis of the target that the registration obtained in the step 1.6 is crossed, target establish target visible light with, it is remote
Infrared two band images Sample Storehouse, half is as training set, and half is as test set;
Step 2.2, visible ray is automatically extracted with, the validity feature of the band image target of far infrared two with convolutional neural networks,
Fusion feature is obtained, it is final to realize fusion visible ray and infrared two band images target signature as the input of grader
Target identification.
6. a kind of fusion visible ray and the target identification of infrared two band images target signature according to claim 4 or 5
Method, it is characterised in that
It is special by the big target of discrimination between all kinds of if the destination number that inhomogeneity target includes in each single source images is consistent
Levy to determine which class visible images target, middle infrared image target and far infrared image object are to belong to same type of.
7. a kind of fusion visible ray and the mesh of infrared two band images target signature according to any one of claim 2 to 5
Mark recognition methods, it is characterised in that
The target signature includes what is automatically extracted by convolutional neural networks, and target signature automatically extracts process:Sample will be trained
Originally classification logotype is carried out, with the deep learning method training sample based on convolutional neural networks, obtains disaggregated model, when there is test
When sample inputs, the disaggregated model that convolutional neural networks can train according to this automatically extracts feature.
8. a kind of fusion visible ray and the mesh of infrared two band images target signature according to any one of claim 2 to 5
Mark recognition methods, it is characterised in that the grader includes softmax, Logistic, SVM, K-Means, Random
Forest, KNN grader.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710700308.0A CN107578432B (en) | 2017-08-16 | 2017-08-16 | Target identification method fusing visible light and infrared two-band image target characteristics |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710700308.0A CN107578432B (en) | 2017-08-16 | 2017-08-16 | Target identification method fusing visible light and infrared two-band image target characteristics |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107578432A true CN107578432A (en) | 2018-01-12 |
CN107578432B CN107578432B (en) | 2020-08-14 |
Family
ID=61033966
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710700308.0A Active CN107578432B (en) | 2017-08-16 | 2017-08-16 | Target identification method fusing visible light and infrared two-band image target characteristics |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107578432B (en) |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108229440A (en) * | 2018-02-06 | 2018-06-29 | 北京奥开信息科技有限公司 | One kind is based on Multi-sensor Fusion indoor human body gesture recognition method |
CN108428224A (en) * | 2018-01-09 | 2018-08-21 | 中国农业大学 | Animal body surface temperature checking method and device based on convolutional Neural net |
CN108681710A (en) * | 2018-05-16 | 2018-10-19 | 北京理工大学 | Ship identification method and device under sky and ocean background based on broadband-EO-1 hyperion infrared image fusion method |
CN108710910A (en) * | 2018-05-18 | 2018-10-26 | 中国科学院光电研究院 | A kind of target identification method and system based on convolutional neural networks |
CN108961180A (en) * | 2018-06-22 | 2018-12-07 | 理光软件研究所(北京)有限公司 | infrared image enhancing method and system |
CN109063701A (en) * | 2018-08-08 | 2018-12-21 | 合肥英睿***技术有限公司 | Labeling method, device, equipment and the storage medium of target in a kind of infrared image |
CN109271921A (en) * | 2018-09-12 | 2019-01-25 | 合刃科技(武汉)有限公司 | A kind of intelligent identification Method and system of multispectral imaging |
CN109308483A (en) * | 2018-07-11 | 2019-02-05 | 南京航空航天大学 | Double source image characteristics extraction and fusion identification method based on convolutional neural networks |
CN109360179A (en) * | 2018-10-18 | 2019-02-19 | 上海海事大学 | A kind of image interfusion method, device and readable storage medium storing program for executing |
CN109376750A (en) * | 2018-06-15 | 2019-02-22 | 武汉大学 | A kind of Remote Image Classification merging medium-wave infrared and visible light |
CN110243769A (en) * | 2019-07-30 | 2019-09-17 | 南阳理工学院 | A kind of the high spectrum sub-pixel target identification system and method for multi-source information auxiliary |
CN110428455A (en) * | 2019-04-19 | 2019-11-08 | 中国航空无线电电子研究所 | A kind of visible images and far infrared image object method for registering |
CN110738100A (en) * | 2019-09-04 | 2020-01-31 | 沈阳工业大学 | camouflage military target identification method and system based on deep learning |
CN110775467A (en) * | 2019-10-31 | 2020-02-11 | 哈尔滨工业大学(深圳) | Garbage storage system and method based on intelligent recognition and voice prompt |
CN110930311A (en) * | 2018-09-19 | 2020-03-27 | 杭州萤石软件有限公司 | Method and device for improving signal-to-noise ratio of infrared image and visible light image fusion |
CN111401107A (en) * | 2019-01-02 | 2020-07-10 | 上海大学 | Multi-mode face recognition method based on feature fusion neural network |
CN111899288A (en) * | 2020-06-01 | 2020-11-06 | 上海大学 | Tunnel leakage water area detection and identification method based on infrared and visible light image fusion |
CN112132753A (en) * | 2020-11-06 | 2020-12-25 | 湖南大学 | Infrared image super-resolution method and system for multi-scale structure guide image |
CN112907624A (en) * | 2021-01-27 | 2021-06-04 | 湖北航天技术研究院总体设计所 | Target positioning and tracking method and system based on multi-band information fusion |
CN113313023A (en) * | 2021-05-28 | 2021-08-27 | 的卢技术有限公司 | Method and system for guaranteeing driving safety when pet is intelligently sensed in vehicle |
CN115082968A (en) * | 2022-08-23 | 2022-09-20 | 天津瑞津智能科技有限公司 | Behavior identification method based on infrared light and visible light fusion and terminal equipment |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101835032A (en) * | 2010-04-27 | 2010-09-15 | 沈阳瑗玛信息技术有限公司 | Multi-camera crowd-gathered message statistic device and method |
US20150309707A1 (en) * | 2013-12-18 | 2015-10-29 | Flir Systems Ab | Processing an infrared (ir) image based on swipe gestures |
CN106096604A (en) * | 2016-06-02 | 2016-11-09 | 西安电子科技大学昆山创新研究院 | Multi-spectrum fusion detection method based on unmanned platform |
CN106485245A (en) * | 2015-08-24 | 2017-03-08 | 南京理工大学 | A kind of round-the-clock object real-time tracking method based on visible ray and infrared image |
CN106960202A (en) * | 2017-04-11 | 2017-07-18 | 广西师范大学 | A kind of smiling face's recognition methods merged based on visible ray with infrared image |
CN106960428A (en) * | 2016-01-12 | 2017-07-18 | 浙江大立科技股份有限公司 | Visible ray and infrared double-waveband image co-registration Enhancement Method |
-
2017
- 2017-08-16 CN CN201710700308.0A patent/CN107578432B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101835032A (en) * | 2010-04-27 | 2010-09-15 | 沈阳瑗玛信息技术有限公司 | Multi-camera crowd-gathered message statistic device and method |
US20150309707A1 (en) * | 2013-12-18 | 2015-10-29 | Flir Systems Ab | Processing an infrared (ir) image based on swipe gestures |
CN106485245A (en) * | 2015-08-24 | 2017-03-08 | 南京理工大学 | A kind of round-the-clock object real-time tracking method based on visible ray and infrared image |
CN106960428A (en) * | 2016-01-12 | 2017-07-18 | 浙江大立科技股份有限公司 | Visible ray and infrared double-waveband image co-registration Enhancement Method |
CN106096604A (en) * | 2016-06-02 | 2016-11-09 | 西安电子科技大学昆山创新研究院 | Multi-spectrum fusion detection method based on unmanned platform |
CN106960202A (en) * | 2017-04-11 | 2017-07-18 | 广西师范大学 | A kind of smiling face's recognition methods merged based on visible ray with infrared image |
Non-Patent Citations (2)
Title |
---|
JING JIANG ET AL.: "Visible and Infrared Image Automatic Registration Algorithm Using Mutual Information", 《CHINESE CONTROL AND DECISION CONFERENCE》 * |
张鹏辉: "红外、微光/可见光图像融合算法研究", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (33)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108428224A (en) * | 2018-01-09 | 2018-08-21 | 中国农业大学 | Animal body surface temperature checking method and device based on convolutional Neural net |
CN108428224B (en) * | 2018-01-09 | 2020-05-22 | 中国农业大学 | Animal body surface temperature detection method and device based on convolutional neural network |
CN108229440A (en) * | 2018-02-06 | 2018-06-29 | 北京奥开信息科技有限公司 | One kind is based on Multi-sensor Fusion indoor human body gesture recognition method |
CN108681710A (en) * | 2018-05-16 | 2018-10-19 | 北京理工大学 | Ship identification method and device under sky and ocean background based on broadband-EO-1 hyperion infrared image fusion method |
CN108681710B (en) * | 2018-05-16 | 2020-11-27 | 北京理工大学 | Ship identification method and device under sea-sky background based on broadband-hyperspectral infrared image fusion method |
CN108710910A (en) * | 2018-05-18 | 2018-10-26 | 中国科学院光电研究院 | A kind of target identification method and system based on convolutional neural networks |
CN108710910B (en) * | 2018-05-18 | 2020-12-04 | 中国科学院光电研究院 | Target identification method and system based on convolutional neural network |
CN109376750A (en) * | 2018-06-15 | 2019-02-22 | 武汉大学 | A kind of Remote Image Classification merging medium-wave infrared and visible light |
CN108961180B (en) * | 2018-06-22 | 2020-09-25 | 理光软件研究所(北京)有限公司 | Infrared image enhancement method and system |
CN108961180A (en) * | 2018-06-22 | 2018-12-07 | 理光软件研究所(北京)有限公司 | infrared image enhancing method and system |
CN109308483A (en) * | 2018-07-11 | 2019-02-05 | 南京航空航天大学 | Double source image characteristics extraction and fusion identification method based on convolutional neural networks |
CN109308483B (en) * | 2018-07-11 | 2021-09-17 | 南京航空航天大学 | Dual-source image feature extraction and fusion identification method based on convolutional neural network |
CN109063701A (en) * | 2018-08-08 | 2018-12-21 | 合肥英睿***技术有限公司 | Labeling method, device, equipment and the storage medium of target in a kind of infrared image |
CN109271921B (en) * | 2018-09-12 | 2021-01-05 | 合刃科技(武汉)有限公司 | Intelligent identification method and system for multispectral imaging |
CN109271921A (en) * | 2018-09-12 | 2019-01-25 | 合刃科技(武汉)有限公司 | A kind of intelligent identification Method and system of multispectral imaging |
CN110930311A (en) * | 2018-09-19 | 2020-03-27 | 杭州萤石软件有限公司 | Method and device for improving signal-to-noise ratio of infrared image and visible light image fusion |
CN110930311B (en) * | 2018-09-19 | 2023-04-25 | 杭州萤石软件有限公司 | Method and device for improving signal-to-noise ratio of infrared image and visible light image fusion |
CN109360179A (en) * | 2018-10-18 | 2019-02-19 | 上海海事大学 | A kind of image interfusion method, device and readable storage medium storing program for executing |
CN111401107A (en) * | 2019-01-02 | 2020-07-10 | 上海大学 | Multi-mode face recognition method based on feature fusion neural network |
CN111401107B (en) * | 2019-01-02 | 2023-08-18 | 上海大学 | Multi-mode face recognition method based on feature fusion neural network |
CN110428455A (en) * | 2019-04-19 | 2019-11-08 | 中国航空无线电电子研究所 | A kind of visible images and far infrared image object method for registering |
CN110428455B (en) * | 2019-04-19 | 2022-11-04 | 中国航空无线电电子研究所 | Target registration method for visible light image and far infrared image |
CN110243769A (en) * | 2019-07-30 | 2019-09-17 | 南阳理工学院 | A kind of the high spectrum sub-pixel target identification system and method for multi-source information auxiliary |
CN110738100A (en) * | 2019-09-04 | 2020-01-31 | 沈阳工业大学 | camouflage military target identification method and system based on deep learning |
CN110775467B (en) * | 2019-10-31 | 2021-11-23 | 哈尔滨工业大学(深圳) | Garbage storage system and method based on intelligent recognition and voice prompt |
CN110775467A (en) * | 2019-10-31 | 2020-02-11 | 哈尔滨工业大学(深圳) | Garbage storage system and method based on intelligent recognition and voice prompt |
CN111899288A (en) * | 2020-06-01 | 2020-11-06 | 上海大学 | Tunnel leakage water area detection and identification method based on infrared and visible light image fusion |
CN112132753B (en) * | 2020-11-06 | 2022-04-05 | 湖南大学 | Infrared image super-resolution method and system for multi-scale structure guide image |
CN112132753A (en) * | 2020-11-06 | 2020-12-25 | 湖南大学 | Infrared image super-resolution method and system for multi-scale structure guide image |
CN112907624A (en) * | 2021-01-27 | 2021-06-04 | 湖北航天技术研究院总体设计所 | Target positioning and tracking method and system based on multi-band information fusion |
CN113313023A (en) * | 2021-05-28 | 2021-08-27 | 的卢技术有限公司 | Method and system for guaranteeing driving safety when pet is intelligently sensed in vehicle |
CN115082968A (en) * | 2022-08-23 | 2022-09-20 | 天津瑞津智能科技有限公司 | Behavior identification method based on infrared light and visible light fusion and terminal equipment |
CN115082968B (en) * | 2022-08-23 | 2023-03-28 | 天津瑞津智能科技有限公司 | Behavior identification method based on infrared light and visible light fusion and terminal equipment |
Also Published As
Publication number | Publication date |
---|---|
CN107578432B (en) | 2020-08-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107578432A (en) | Merge visible ray and the target identification method of infrared two band images target signature | |
Ramesh et al. | Plant disease detection using machine learning | |
Minetto et al. | Hydra: An ensemble of convolutional neural networks for geospatial land classification | |
CN110348319B (en) | Face anti-counterfeiting method based on face depth information and edge image fusion | |
Feng et al. | CNN-based multilayer spatial–spectral feature fusion and sample augmentation with local and nonlocal constraints for hyperspectral image classification | |
CN110363140B (en) | Human body action real-time identification method based on infrared image | |
CN107368831B (en) | English words and digit recognition method in a kind of natural scene image | |
CN109064502A (en) | The multi-source image method for registering combined based on deep learning and artificial design features | |
Mahapatra et al. | Retinal image quality classification using saliency maps and CNNs | |
CN104361313B (en) | A kind of gesture identification method merged based on Multiple Kernel Learning heterogeneous characteristic | |
CN104463250B (en) | A kind of Sign Language Recognition interpretation method based on Davinci technology | |
CN109684922B (en) | Multi-model finished dish identification method based on convolutional neural network | |
CN109101938B (en) | Multi-label age estimation method based on convolutional neural network | |
CN106529499A (en) | Fourier descriptor and gait energy image fusion feature-based gait identification method | |
CN110298291A (en) | Ox face and ox face critical point detection method based on Mask-RCNN | |
CN109308483A (en) | Double source image characteristics extraction and fusion identification method based on convolutional neural networks | |
CN107292246A (en) | Infrared human body target identification method based on HOG PCA and transfer learning | |
CN113420607A (en) | Multi-scale target detection and identification method for unmanned aerial vehicle | |
CN109543632A (en) | A kind of deep layer network pedestrian detection method based on the guidance of shallow-layer Fusion Features | |
CN106909902A (en) | A kind of remote sensing target detection method based on the notable model of improved stratification | |
CN106897681A (en) | A kind of remote sensing images comparative analysis method and system | |
Xu et al. | Robust self-ensembling network for hyperspectral image classification | |
CN113158943A (en) | Cross-domain infrared target detection method | |
CN109086803B (en) | Deep learning and personalized factor-based haze visibility detection system and method | |
CN113221655B (en) | Face spoofing detection method based on feature space constraint |
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 | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20230309 Address after: 211599 No.8, Huyue East Road, Longchi street, Liuhe District, Nanjing City, Jiangsu Province Patentee after: Jiangsu Jingtian Technology Co.,Ltd. Address before: No. 29, Qinhuai District, Qinhuai District, Nanjing, Jiangsu Patentee before: Nanjing University of Aeronautics and Astronautics |