CN109711256A - A kind of low latitude complex background unmanned plane target detection method - Google Patents
A kind of low latitude complex background unmanned plane target detection method Download PDFInfo
- Publication number
- CN109711256A CN109711256A CN201811424123.2A CN201811424123A CN109711256A CN 109711256 A CN109711256 A CN 109711256A CN 201811424123 A CN201811424123 A CN 201811424123A CN 109711256 A CN109711256 A CN 109711256A
- Authority
- CN
- China
- Prior art keywords
- target
- unmanned plane
- image
- background
- filtering
- 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)
- Image Processing (AREA)
Abstract
The invention belongs to technical field of image detection, and in particular to a kind of low latitude complex background unmanned plane target detection method.The Filtering Template that this method is used in image filtering aims at the design of sky background Weak target, the contrast that target and neighborhood background are significantly improved after filtering, the selection of gray threshold, Edge Following threshold value when being more advantageous to the target detection of low signal-to-noise ratio (SNR) images and extracting target；Utilize suspected target neighborhood characteristics primary system calculating method, typical context interference can effectively be rejected, do not need by consecutive frame image be registrated and difference, it is not high to searching turntable stability requirement, it is low compared to the Algorithms T-cbmplexity based on image registration, be more conducive to realized in real time embedded system；The present invention make full use of unmanned plane and fractus, atural object, flying bird kinetic characteristic difference, judge whether suspected target is unmanned plane target by the requirement of limitation and orientation consistency to target trajectory displacement, false alarm rate is lower.
Description
Technical field
The invention belongs to technical field of image detection, and in particular to a kind of low latitude based on panorama infrared search system is complicated
Background unmanned plane target detection method.
Background technique
In recent years, with the fast development of unmanned air vehicle technique and extensive use, unmanned plane target Detection Techniques, which have become, to be worked as
Preceding hot research problem.The object detection method based on panorama infrared search system of mainstream is mostly based on back both at home and abroad at present
The means such as scape difference extract target, need to consume a large amount of hardware resource storage background image, to the lasting accuracy of search platform
It is higher with the performance requirement of registration Algorithm, it is not easy to realize by the real time embedded system of core of DSP.
Summary of the invention
(1) technical problems to be solved
The present invention proposes a kind of low latitude complex background unmanned plane target detection method, to solve how effectively to exclude intricately
Object background, flying bird and cloud noise, accurate the technical issues of detecting unmanned plane target in sky background.
(2) technical solution
In order to solve the above-mentioned technical problem, the present invention proposes a kind of low latitude complex background unmanned plane target detection method, should
Detection method includes the following steps:
S1, image filtering: doing convolution using following Filtering Template and original image, relatively uniform in original image after filtering
Sky background be compressed into lower gray level, and day in the air relatively bright Small object, the bright background in ground edge still protect
Hold higher gray level；
- 1, -2, -3, -2, -1
- 2,3,4,3, -2
- 3,4,4,4, -3
- 2,3,4,3, -2
- 1, -2, -3, -2, -1
S2, it extracts suspected target: gray threshold, Edge Following threshold value, size threshold is set to filtered image data,
The suspected target for meeting gray scale, size threshold requirement in image is extracted using Edge Following method；
S3, reject background interference: the neighborhood bright pixel number of suspected target in difference statistic procedure S2 exceeds bright pixel number
The suspected target of threshold range is determined as that atural object or cloud background interference are rejected, and remaining target is then the aerial doubtful small mesh in day
Mark；
S4, time-domain information correlation establish targetpath: by judging that doubtful Small object motion profile further confirms that nobody
Machine target.
Further, step S4 is specifically included: after finding in a certain size neighborhood of previous frame image suspected target
With the immediate suspected target of its gray scale in one frame image, it is then considered same target if it exists, according to front and back two field pictures
Target position calculates moving displacement and the direction of target；If certain suspected target two continuous frames direction of motion is consistent and displacement is certain
Unmanned plane target is then determined that it is in range.
(3) beneficial effect
Complex background unmanned plane target detection method in low latitude proposed by the present invention, including image filtering, extract suspected target,
Reject that background interference is related to time-domain information establishes four steps of targetpath.Wherein, the Filtering Template used when image filtering
The design of sky background Weak target is aimed at, the contrast of target and neighborhood background is significantly improved after filtering, is more advantageous to low noise
Than the target detection of image, while being conducive to the selection of gray threshold, Edge Following threshold value when extraction target；Utilize suspected target
Neighborhood characteristics primary system calculating method, can effectively reject the interference of the typical contexts such as most atural objects, cloud layer, and this method does not need to pass through
Consecutive frame image registration and difference, it is not high to searching turntable stability requirement, it is complicated compared to the algorithm time based on image registration
Spend it is low, be more conducive to realized in real time embedded system；The present invention makes full use of unmanned plane and fractus, atural object, flying bird movement spy
The difference of property, by the requirement of limitation and orientation consistency to target trajectory displacement judge suspected target whether be
Unmanned plane target, false alarm rate are lower.
Detailed description of the invention
Fig. 1 is the unmanned plane target detection method flow chart of the embodiment of the present invention；
Fig. 2 is contour-tracking algorithm flow chart in the embodiment of the present invention；
Fig. 3 is target neighborhood schematic diagram in the embodiment of the present invention；
Fig. 4 a is the remote 80m high unmanned plane infrared image of 700m of embodiment 1；
Fig. 4 b is Fig. 4 a filtered image in embodiment 1；
Fig. 4 c is that Fig. 4 b extracts suspected target result in embodiment 1；
Fig. 4 d is that Fig. 4 c removes the object detection results after background process in embodiment 1；
Fig. 4 e is that Fig. 4 a is superimposed object detection results rectangular window result in embodiment 1；
Fig. 5 a is the remote 70m high unmanned plane infrared image of 1.2km of embodiment 2；
Fig. 5 b is that local area image enhances result where Fig. 5 a target in embodiment 2；
Fig. 5 c is Fig. 5 a filtered image in embodiment 2；
Fig. 5 d is that Fig. 5 c removes the object detection results after background process in embodiment 2；
Fig. 5 e is that Fig. 5 b is superimposed object detection results rectangular window result in embodiment 2；
Fig. 6 a has the remote 80m high unmanned plane infrared image of the 400m of cloud background for embodiment 3；
Fig. 6 b is Fig. 6 a filtered image in embodiment 3；
Fig. 6 c is that Fig. 6 b extracts suspected target result in embodiment 3；
Fig. 6 d is that Fig. 6 c removes the object detection results after background process in embodiment 3；
Fig. 6 e is that Fig. 6 a is superimposed object detection results rectangular window result in embodiment 3.
Specific embodiment
To keep the purpose of the present invention, content and advantage clearer, with reference to the accompanying drawings and examples, to tool of the invention
Body embodiment is described in further detail.
The present embodiment proposes that a kind of low latitude complex background unmanned plane target detection method, process are as shown in Figure 1.The target
Detection method includes the following steps:
S1, image filtering: doing convolution using following Filtering Template and original image, relatively uniform in original image after filtering
Sky background be compressed into lower gray level, and day in the air relatively bright Small object, the bright background in ground edge still protect
Higher gray level is held, to be more advantageous to the extraction of suspected target in step S2.
- 1, -2, -3, -2, -1
- 2,3,4,3, -2
- 3,4,4,4, -3
- 2,3,4,3, -2
- 1, -2, -3, -2, -1
S2, it extracts suspected target: reasonable gray threshold, Edge Following threshold value, ruler is set to filtered image data
Very little threshold value extracts the suspected target for meeting gray scale, size threshold requirement in image using Edge Following method.Contour-tracking algorithm stream
Journey, as shown in Figure 2.
S3, reject background interference: since real goal is in more uniform sky background, its neighborhood is (such as Fig. 3 institute after filtering
Showing) number is less for bright pixel (pixel that gray value is greater than certain threshold value is denoted as bright pixel), and atural object, the filtered neighbour of cloud background
Domain bright pixel number is more, is distinguished using this difference to aerial target and surface feature background interference, respectively in statistic procedure S2
Bright pixel number is determined as atural object or cloud background interference beyond the suspected target of threshold range by the neighborhood bright pixel number of suspected target
It is rejected, remaining target is then the aerial doubtful Small object in day.
S4, time-domain information correlation establish targetpath: still having flying bird, dotted broken after step S3, in doubtful Small object
The interference such as cloud, individual ground scenery, slower than flying bird, dotted fractus and individual ground scenery phases using unmanned plane lateral movement velocity
To the features such as static, by judging that doubtful Small object motion profile further confirms that unmanned plane target, i.e., in previous frame image
It finds in a certain size neighborhood of suspected target in a later frame image with the immediate suspected target of its gray scale, then thinks if it exists
It is same target, moving displacement and the direction of target is calculated according to the target position of front and back two field pictures.Since unmanned plane is short
(adjacent 3 frame) substantially linear motion state, direction of displacement are consistent in time, if therefore certain suspected target two continuous frames
The direction of motion is consistent and displacement then determines that it is unmanned plane target in a certain range.
It is illustrated below by way of specific embodiment
Using certain panorama infrared search system acquire 14 low latitude unmanned plane infrared images, image resolution ratio be 640 ×
512, unmanned plane is big boundary 300mm wheelbase quadrotor drone, and flying height is 50~120m, 0.1~1.2km of horizontal distance.
Embodiment 1
Fig. 4 a is the remote 80m high unmanned plane infrared image of 700m, has the surface feature backgrounds such as tree crown, building top in figure.Utilize mould
Plate is to image filtering, as a result as shown in Figure 4 b.Using Fig. 2 contour-tracking algorithm, detection threshold value TH1=300~500, TH2 are set
=200~300, which obtain suspected target, extracts result as illustrated in fig. 4 c.Target neighborhood size is set as 8~15 pixels, bright pixel is grey
Spending decision threshold is 120~140, and atural object, cloud background bright pixel number judgment threshold are 12~18, is carried out to the suspected target of extraction
It goes atural object to handle, obtains object detection results as fig 5d, be superimposed testing result rectangular window result on the original image as schemed
Shown in 4e.
Embodiment 2
Fig. 5 a is the remote 70m high unmanned plane infrared image of 1.2km, the atural objects such as live wire bar, tree crown and building top back in figure
Scape.Fig. 5 b is target region image enhancement result.Using template to image filtering, as a result as shown in Figure 5 c.Using the side Fig. 2
Edge tracing algorithm, setting detection threshold value TH1=300~400, TH2=150~300, target neighborhood size are 8~15 pixels, bright
Pixel grey scale decision threshold is 120~140, and atural object, cloud background bright pixel number judgment threshold are 12~18, is carried out to image doubtful
Objective extraction goes atural object to handle, and obtains object detection results as fig 5d, and testing result rectangular window result is superimposed on Fig. 5 b
As depicted in fig. 5e.
Embodiment 3
Fig. 6 a is the remote 80m high unmanned plane infrared image of 400m, and there is more cloud background in day in the air.Image is filtered using template
Wave, as a result as shown in Figure 6 b.Using Fig. 2 contour-tracking algorithm, detection threshold value TH1=300~500, TH2=200~300 are set
It obtains suspected target and extracts result as fig. 6 c.Target neighborhood size is set as 8~15 pixels, bright pixel gray scale decision threshold
It is 120~140, atural object, cloud background bright pixel number judgment threshold are 12~18, are carried out at atural object to the suspected target of extraction
Reason, obtains object detection results as shown in fig 6d, is superimposed testing result rectangular window result as shown in fig 6e on the original image.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, without departing from the technical principles of the invention, several improvement and deformations can also be made, these improvement and deformations
Also it should be regarded as protection scope of the present invention.
Claims (2)
1. a kind of low latitude complex background unmanned plane target detection method, which is characterized in that the detection method includes the following steps:
S1, image filtering: doing convolution using following Filtering Template and original image, relatively uniform day in original image after filtering
Empty background is compressed into lower gray level, and day in the air relatively bright Small object, the bright background in ground edge still keep compared with
High gray level；
- 1, -2, -3, -2, -1
- 2,3,4,3, -2
- 3,4,4,4, -3
- 2,3,4,3, -2
- 1, -2, -3, -2, -1
S2, it extracts suspected target: gray threshold, Edge Following threshold value, size threshold being set to filtered image data, used
Edge Following method extracts the suspected target for meeting gray scale, size threshold requirement in image；
S3, reject background interference: bright pixel number is exceeded threshold value by the neighborhood bright pixel number of suspected target in difference statistic procedure S2
The suspected target of range is determined as that atural object or cloud background interference are rejected, and remaining target is then the aerial doubtful Small object in day；
S4, time-domain information correlation establish targetpath: by judging that doubtful Small object motion profile further confirms that unmanned plane mesh
Mark.
2. object detection method as described in claim 1, which is characterized in that the step S4 is specifically included: in former frame figure
As found in a certain size neighborhood of certain suspected target in a later frame image with the immediate suspected target of its gray scale, if it exists then
It is considered same target, moving displacement and the direction of target is calculated according to the target position of front and back two field pictures；If certain doubtful mesh
The mark two continuous frames direction of motion is consistent and displacement then determines that it is unmanned plane target in a certain range.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811424123.2A CN109711256B (en) | 2018-11-27 | 2018-11-27 | Low-altitude complex background unmanned aerial vehicle target detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811424123.2A CN109711256B (en) | 2018-11-27 | 2018-11-27 | Low-altitude complex background unmanned aerial vehicle target detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109711256A true CN109711256A (en) | 2019-05-03 |
| CN109711256B CN109711256B (en) | 2022-12-09 |
Family
ID=66254413
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811424123.2A Active CN109711256B (en) | 2018-11-27 | 2018-11-27 | Low-altitude complex background unmanned aerial vehicle target detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109711256B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110796689A (en) * | 2019-10-28 | 2020-02-14 | 咪咕视讯科技有限公司 | Video processing method, electronic equipment and storage medium |
| CN111210422A (en) * | 2020-01-13 | 2020-05-29 | 北京科技大学 | Air target detection method based on infrared image |
| CN111624590A (en) * | 2020-05-13 | 2020-09-04 | 飒铂智能科技有限责任公司 | Unmanned aerial vehicle target confirmation method and system |
| CN112070786A (en) * | 2020-07-17 | 2020-12-11 | 中国人民解放军63892部队 | Alert radar PPI image target/interference extraction method |
| CN112435249A (en) * | 2020-11-30 | 2021-03-02 | 天津津航技术物理研究所 | Dynamic small target detection method based on periodic scanning infrared search system |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6496592B1 (en) * | 1998-07-13 | 2002-12-17 | Oerlikon Contraves Ag | Method for tracking moving object by means of specific characteristics |
| US20090262977A1 (en) * | 2008-04-18 | 2009-10-22 | Cheng-Ming Huang | Visual tracking system and method thereof |
| CN101604383A (en) * | 2009-07-24 | 2009-12-16 | 哈尔滨工业大学 | A kind of method for detecting targets at sea based on infrared image |
| US20120106799A1 (en) * | 2009-07-03 | 2012-05-03 | Shenzhen Taishan Online Technology Co., Ltd. | Target detection method and apparatus and image acquisition device |
| US20160048952A1 (en) * | 2014-08-15 | 2016-02-18 | Nikon Corporation | Algorithm and device for image processing |
| JP2017104025A (en) * | 2015-12-07 | 2017-06-15 | 株式会社東和電機製作所 | Fishing support system using unmanned aircraft |
| CN107944497A (en) * | 2017-12-06 | 2018-04-20 | 天津大学 | Image block method for measuring similarity based on principal component analysis |
| US20180128895A1 (en) * | 2016-11-08 | 2018-05-10 | Dedrone Holdings, Inc. | Systems, Methods, Apparatuses, and Devices for Identifying, Tracking, and Managing Unmanned Aerial Vehicles |
-
2018
- 2018-11-27 CN CN201811424123.2A patent/CN109711256B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6496592B1 (en) * | 1998-07-13 | 2002-12-17 | Oerlikon Contraves Ag | Method for tracking moving object by means of specific characteristics |
| US20090262977A1 (en) * | 2008-04-18 | 2009-10-22 | Cheng-Ming Huang | Visual tracking system and method thereof |
| US20120106799A1 (en) * | 2009-07-03 | 2012-05-03 | Shenzhen Taishan Online Technology Co., Ltd. | Target detection method and apparatus and image acquisition device |
| CN101604383A (en) * | 2009-07-24 | 2009-12-16 | 哈尔滨工业大学 | A kind of method for detecting targets at sea based on infrared image |
| US20160048952A1 (en) * | 2014-08-15 | 2016-02-18 | Nikon Corporation | Algorithm and device for image processing |
| JP2017104025A (en) * | 2015-12-07 | 2017-06-15 | 株式会社東和電機製作所 | Fishing support system using unmanned aircraft |
| US20180128895A1 (en) * | 2016-11-08 | 2018-05-10 | Dedrone Holdings, Inc. | Systems, Methods, Apparatuses, and Devices for Identifying, Tracking, and Managing Unmanned Aerial Vehicles |
| CN107944497A (en) * | 2017-12-06 | 2018-04-20 | 天津大学 | Image block method for measuring similarity based on principal component analysis |
Non-Patent Citations (2)
| Title |
|---|
| 赵英海等: "可见光遥感图像中舰船目标检测方法", 《光电工程》 * |
| 赵高鹏等: "一种基于特征匹配的目标识别跟踪方法", 《火炮发射与控制学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110796689A (en) * | 2019-10-28 | 2020-02-14 | 咪咕视讯科技有限公司 | Video processing method, electronic equipment and storage medium |
| CN111210422A (en) * | 2020-01-13 | 2020-05-29 | 北京科技大学 | Air target detection method based on infrared image |
| CN111624590A (en) * | 2020-05-13 | 2020-09-04 | 飒铂智能科技有限责任公司 | Unmanned aerial vehicle target confirmation method and system |
| CN111624590B (en) * | 2020-05-13 | 2023-07-21 | 飒铂智能科技有限责任公司 | Unmanned aerial vehicle target confirmation method and system |
| CN112070786A (en) * | 2020-07-17 | 2020-12-11 | 中国人民解放军63892部队 | Alert radar PPI image target/interference extraction method |
| CN112070786B (en) * | 2020-07-17 | 2023-11-24 | 中国人民解放军63892部队 | Method for extracting warning radar PPI image target and interference |
| CN112435249A (en) * | 2020-11-30 | 2021-03-02 | 天津津航技术物理研究所 | Dynamic small target detection method based on periodic scanning infrared search system |
| CN112435249B (en) * | 2020-11-30 | 2024-04-16 | 天津津航技术物理研究所 | Dynamic small target detection method based on circumferential scanning infrared search system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109711256B (en) | 2022-12-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109711256A (en) | A kind of low latitude complex background unmanned plane target detection method | |
| CN101739686B (en) | Moving object tracking method and system thereof | |
| CN107133969B (en) | A kind of mobile platform moving target detecting method based on background back projection | |
| CN104200485B (en) | Video-monitoring-oriented human body tracking method | |
| CN106056053B (en) | The human posture's recognition methods extracted based on skeleton character point | |
| CN105574855B (en) | Infrared small target detection method under cloud background based on template convolution and false alarm rejection | |
| CN110956094A (en) | RGB-D multi-mode fusion personnel detection method based on asymmetric double-current network | |
| WO2022027931A1 (en) | Video image-based foreground detection method for vehicle in motion | |
| CN109145708B (en) | Pedestrian flow statistical method based on RGB and D information fusion | |
| CN104978567B (en) | Vehicle checking method based on scene classification | |
| WO2017054314A1 (en) | Building height calculation method and apparatus, and storage medium | |
| CN102214309B (en) | Special human body recognition method based on head and shoulder model | |
| CN104835145B (en) | Foreground detection method based on adaptive Codebook background models | |
| CN102867177B (en) | A kind of demographic method based on gradation of image coupling | |
| CN103927520A (en) | Method for detecting human face under backlighting environment | |
| CN103824070A (en) | Rapid pedestrian detection method based on computer vision | |
| CN105160297A (en) | Masked man event automatic detection method based on skin color characteristics | |
| CN108804992B (en) | Crowd counting method based on deep learning | |
| CN110197494B (en) | Pantograph contact point real-time detection algorithm based on monocular infrared image | |
| CN105043395B (en) | A kind of real-time Dynamic Location method of aircraft menology soft landing | |
| CN103049765A (en) | Method for judging crowd density and number of people based on fish eye camera | |
| CN111241943B (en) | Scene recognition and loopback detection method based on background target and triple loss | |
| CN108471497A (en) | A kind of ship target real-time detection method based on monopod video camera | |
| CN105913464A (en) | Multi-body target online measurement method based on videos | |
| CN104299234B (en) | The method and system that rain field removes in video data |
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 |