CN104239845B - The identification of double vision intelligent image and object localization method based on infrared thermal imaging technique - Google Patents
The identification of double vision intelligent image and object localization method based on infrared thermal imaging technique Download PDFInfo
- Publication number
- CN104239845B CN104239845B CN201310246835.0A CN201310246835A CN104239845B CN 104239845 B CN104239845 B CN 104239845B CN 201310246835 A CN201310246835 A CN 201310246835A CN 104239845 B CN104239845 B CN 104239845B
- Authority
- CN
- China
- Prior art keywords
- target
- infrared
- visible images
- identification
- image
- 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.)
- Active
Links
Abstract
A kind of identification of double vision intelligent image and object localization method based on infrared thermal imaging technique, comprise the following steps:It is infrared to be matched with visible ray visual field, initial target positions, and To Template data preserve, goal-setting, target identification, positioning correcting, carry out the target identification in image respectively to infrared image and visible images by computer software, and target is positioned using double vision image recognition complementarity method.The present invention compensate for the deficiency using single image identification and object localization method, greatly increase the degree of accuracy to monitoring objective identification and the reliability of failure or defect inspection.
Description
Technical field
The present invention relates to thermal imaging monitoring technical field, more particularly, to a kind of double vision intelligence based on infrared thermal imaging technique
Can image recognition and object localization method.
Background technology
At present, numerous industries by infrared thermal imaging and thermometry be applied to equipment fault or defect monitoring, temperature rise it is pre-
Alert, technical field of security and protection, its implementation are mainly taken the photograph using thermal infrared imager and visible camera combination as complementary monitor
Picture platform, the infrared ray that thermal infrared imager is launched by receiving testee, the infra-red radiation on testee surface is transformed into
Thermal energy signal;Again temperature value and thermal image are converted to through special-purpose software;Temperature value or Temperature Distribution change in image
Carry out the failure or defect of monitoring objective;And assist in identifying target by visible images.
In monitoring is applied, monitoring objective usually can be by environmental condition (such as:Wind, expand with heat and contract with cold) or head movement deviation
Influence is subjected to displacement, and the monitor of fixed installation is offset monitoring objective, so as to cause the something lost to failure or defect inspection
Leakage, causes economic loss.At present, infrared camera with visible image capturing head mostly with parallel installation, infrared image and visible ray
The visual field of image can not match, and there is very big gap position of the target on infrared image with the position on visible images,
Therefore infrared and visible images can not be effectively utilized simultaneously to same target identification.It is existing red due to target difference position
The target identification and localization method of external monitor system mostly use and single image (infrared or visible ray) are identified.Work as environmental colors
Or the color around target and monitored target color it is very close when, it is seen that light image recognizer just can not effectively be known
Other target;When the temperature of temperature and monitoring objective around environment temperature or target is very close, infrared image recognizer
Target can not be just efficiently identified, so the situation of None- identified target, drop may usually occur using single image identification
Low target discrimination, the omission of failure or defect inspection may be caused in detection, cause great economic loss, therefore use
The complementary identification of infrared and visible ray and position monitor goal approach, it can greatly improve the degree of accuracy of object recognition rate.
The content of the invention
It is of the invention by thermal-induced imagery and visible for insufficient existing for existing image recognition and Technology for Target Location
Light image identification and the technology complementation of target positioning, open a kind of brand-new double vision intelligent image identification and target positioning side
Method, on the basis of being adjusted to the camera physics angle of visual field, identified by double vision with view field image and correcting algorithm, realization monitor mesh
Target identifies and positioning.The present invention can greatly improve the degree of accuracy of monitoring objective discrimination and positioning, avoid unnecessary people
Member and economic loss.Present invention could apply to the identification of any dual camera monitor monitoring image and target positioning.
To achieve the above object, the present invention provides a kind of double vision intelligent image identification based on infrared thermal imaging technique and mesh
Localization method is marked, this method includes:A. infrared image and visible images visual field Matching and modification;B. infrared image and visible is preserved
Initial target shape and situation template in light image;C. by contrasting To Template, identify in infrared image and visible images
Target location;D. the current location of target and initial positional deviation in infrared image are calculated;E. target in visible images is calculated
Current location and initial positional deviation;F. by double vision complementary target bearing calibration, the reality of current goal is gone out with pattern identification
Border position.
According to an embodiment of the invention, infrared figure is made by adjusting scaling multiple that visible images show and position skew
As visual field matches with visible images visual field.
According to an embodiment of the invention, with pattern identification initial target, identification and preserve in figure target signature data and
Identification maximum position deviation (Δ MAX) is specified as recognition template in targeted graphical position.
According to an embodiment of the invention, target in infrared image and visible images is distinguished by the To Template of preservation
Carry out shape and border is found and identification, and calculate the current location of closest target and initial positional deviation value in infrared image
The current location of closest target and initial positional deviation value (Δ CCD) in (Δ IR) and visible images.
According to an embodiment of the invention, by judging relation between Δ IR and Δ MAX and Δ CCD and Δ MAX, identification is calculated
Target current location.
According to an embodiment of the invention, judge whether target can be identified in current infrared image, if identification position is inclined
Difference, which is more than the target for specifying maximum deflection difference value (Δ IR > Δ MAX) to will be regarded as in current infrared image, to be identified, infrared figure
The target location deviation of picture will not participate in target positioning correcting.
According to an embodiment of the invention, judge whether target can be identified in current visible images, if identification position
Deviation, which is more than, specifies maximum deflection difference value (Δ CCD > Δ MAX), and will be regarded as the target in current visible images can not be identified,
The target location deviation of visible images will not participate in target positioning correcting.
According to an embodiment of the invention, (Δ IR≤Δ MAX, Δ the CCD > when target can be identified in only infrared image
Δ MAX), with the deviation amendment of infrared image it is original mark targeted graphical coordinate value, with the targeted graphical of renewal infrared
With the target that displacement is identified on visible images.
According to an embodiment of the invention, when target can be all identified in infrared and visible images (Δ IR≤Δ MAX,
Δ CCD≤Δ MAX), with the coordinate value of the original mark targeted graphical of infrared and visible images deviation intermediate value amendments, with renewal
Targeted graphical the target of displacement is identified on infrared and visible images.
According to an embodiment of the invention, (Δ IR > Δ MAX, Δ CCD when target can be identified in only visible images
≤ Δ MAX), with the coordinate value of the original mark targeted graphical of the deviation amendment of visible images, existed with the targeted graphical of renewal
The target of displacement is identified on infrared and visible images.
According to an embodiment of the invention, (the Δ IR > Δs when target can not be all identified in infrared and visible images
MAX, Δ CCD > Δ MAX), the mark targeted graphical of initial position is marked on infrared and visible images, and prompt image
None- identified.
According to an embodiment of the invention, it is real by repeating to call the complementary identification of double vision intelligent image and object localization method
Target identification and positioning now are carried out to multiple targets in same image.
According to the present invention, the identification of double vision intelligent image and the repeatable calling of target positioning software, realize in same image
Multiple targets be identified and position.
Brief description of the drawings
The camera angle of visual field that Fig. 1 is the present invention positions schematic diagram;
Fig. 2 is the identification of double vision intelligent image and the initial alignment flow chart of target positioning software of the present invention;
Fig. 3 is the identification of double vision intelligent image and the workflow diagram of target positioning software of the present invention.
Embodiment
In order that the purpose of embodiments of the invention, technical scheme, advantage are more clearly understood, with reference to embodiment and
Accompanying drawing, embodiments of the invention are described in detail.Here, the schematic description and description of the present invention is only used
It is of the invention but not as a limitation of the invention in explaining.
At present, the infrared and visible image capturing head of double vision monitor be substantially it is parallel installation (a), its infrared image with can
See that light image is substantially all beyond monitoring range with visual field distance, it is simultaneously right that infrared and visible images can not be effectively utilized
Same target carries out contrasting identification and positioning.In view of the angle of visual field of visible camera is typically much deeper than the visual field of infrared video camera
Angle, the present invention is on the basis of using camera angle of visual field method of adjustment (b), there is provided one kind is based on the complementary identification of double vision intelligent image
And object localization method.
The double vision intelligent image identification of the present invention and target positioning software pass through infrared image and visible images visual field
Read with module (1), initial target mark and locating module (2), To Template data storage module (3), target identification template
Module (4), target identification and deviant computing module (5), target identification judge module (6-8), object recognition and detection straightening die
Block (10-12) collective effect, realize the mesh for effectively and accurately being monitored infrared and visible ray to same target with visual field
's.
As Figure 2-3,
According to an embodiment of the invention, visible ray figure is adjusted by the person-machine dialogue function in software module (1), user
The scaling of picture is (such as:Amplification, reduce) and displacement is (such as:Upper and lower, left and right) visible images, make local in visible images
Image matches with infrared image, and intercepts the visible images part, is shown as visible images, realize visible images with
Infrared image matching display.
According to an embodiment of the invention, matched in visible images with infrared image on display base, pass through software module
(2) the person-machine dialogue function in, user with same figure (such as:Point, line, circle, rectangle, polygon) come while identify infrared
Target (the note of same position in image and visible images:Because the matching of infrared and visible images, user only need to infrared or
Pattern identification target is drawn on visible images, software module draws the same position on visible ray or infrared image together automatically
Master drawing shape identifies target), target in figure is identified by image recognition algorithm for software module, and software module (3) will
Characteristics of image, infrared image, visible images preserve in the lump as To Template data in graph style, graph position, figure
Into computer file system or database.User can identify and preserve multiple targets on same image.
According to an embodiment of the invention, when user or system (such as:Monitoring task) need to specify in image frame to one
When positioning is identified in target, software module (4) will read its corresponding image template data, specify maximum identification position deviation
It is worth (ΔMAX), as target identification and locating reference datum.
According to an embodiment of the invention, software module (5) is known by the target signature of template data, graphics field, image
Other algorithm, on current infrared and visible images identification most similar to mesh target area, and calculate the regional center position with
Position deviation value (the Δ in initial target centre of figure regionIR, ΔCCD)。
According to an embodiment of the invention, module (6) judges whether target can be identified in current infrared image.If target
The position deviation of identification is more than user and specifies maximum deflection difference value (ΔIR> ΔsMAX), show environment temperature or target surrounding objects
The temperature of temperature and monitored target is very close, causes module can not efficiently identify the target in current infrared image, red
The graph position deviation of outer image will not participate in target positioning correcting.If the position deviation of identification is small or is specified equal to user
Maximum deflection difference value (ΔIR≤ΔMAX), show that module recognizes the target in current infrared image, the graph position of infrared image is inclined
Difference will participate in target positioning correcting.
According to an embodiment of the invention, module (7-8) judges whether target can be identified in current visible images.If
The position deviation of target identification is more than user and specifies maximum deflection difference value (ΔCCD> ΔsMAX), show environmental colors or target ambient
The color of body color and monitored target is very close, causes module can not efficiently identify the mesh in current visible images
Mark, it is seen that the graph position deviation of light image will not participate in target positioning correcting.If the position deviation of identification is small or is equal to
User specifies maximum deflection difference value (ΔCCD≤ΔMAX), show that module recognizes the target in current visible images, it is seen that light figure
The graph position deviation of picture will participate in target positioning correcting.
According to an embodiment of the invention, (the Δ when the target in only infrared image is identifiedIR≤ΔMAX, ΔCCD>
ΔMAX), the figure deviation (Δ that software passes through module (9) infrared imageIR) amendment initial marking target figure coordinate
Value, and with the figure after more new position it is infrared with visible images on identify displacement after target.
According to an embodiment of the invention, (the Δ when the target in infrared image and visible ray figure is all identifiedIR≤
ΔMAX, ΔCCD≤ΔMAX), software is by module (10) with infrared and visible images figure deviation intermediate value ([ΔsCCD+ΔIR]/
2) coordinate value of the figure of initial marking target is corrected, and is identified with the figure after more new position on infrared and visible images
The target gone out after displacement.
According to an embodiment of the invention, (the Δ when the target in only visible images is identifiedIR> ΔsMAX, ΔCCD≤
ΔMAX), the figure deviation (Δ that software passes through module (11) visible imagesCCD) amendment initial marking target figure
Coordinate value, and the target after identifying displacement on infrared and visible images with the figure after more new position.
According to an embodiment of the invention, (the Δ when target can not be all identified in infrared and visible imagesIR> ΔsMAX,
ΔCCD> ΔsMAX), software module (12) does not do positioning correcting, by the targeted graphical mark of initial marking in infrared and visible ray figure
As upper.
According to an embodiment of the invention, user or the repeatable module (4-12) of system are entered to multiple targets on same image
Row identification and positioning.
Although being had been described in detail above in conjunction with accompanying drawing to the specific illustrative embodiment of the present invention, it is to be understood that
, within the spirit and scope of the principle of present disclosure, those skilled in the art can derive many other completely
Change and embodiment.Especially, can in the disclosure, accompanying drawing and scope of the following claims to component and/or annex and/
Or the setting combination of software carries out a variety of changes and improvements.In addition to the changes and improvements of component and/or annex and/or software,
Other selectable applications are for art technology and obvious.
Claims (7)
1. a kind of identification of double vision intelligent image and object localization method based on infrared thermal imaging technique, including:
A. infrared image and visible images visual field Matching and modification;
B. initial target shape and situation template in infrared image and visible images are preserved;
C. To Template data corresponding to reading, identification maximum position deviation Δ is specifiedMAX;
D. by contrasting To Template, target location in infrared image and visible images is identified;
E. the current location of closest target and initial positional deviation value Δ in infrared image are calculatedIR;
F. the current location of closest target and initial positional deviation value Δ in visible images are calculatedCCD;
G. by double vision complementary target bearing calibration, the physical location of current goal is gone out with pattern identification;
Configure one kind in following four kinds of correcting modes or four kinds are used as target correction method to carry out target correction:
I) when target can be identified in only infrared image, i.e. ΔIR≤ΔMAXAnd ΔCCD> ΔsMAXWhen, it is inclined with infrared image
The coordinate value of the original mark targeted graphical of difference amendment, displacement is identified on infrared and visible images with the targeted graphical of renewal
Target;
Ii) when target can be all identified in infrared and visible images, i.e. ΔIR≤ΔMAXAnd ΔCCD≤ΔMAXWhen, use is infrared
With the deviation intermediate value of visible images, i.e. [ΔCCD+ΔIR]/2, the coordinate value of original mark targeted graphical is corrected, with the mesh of renewal
Shape of marking on a map identifies the target of displacement on infrared and visible images;
Iii) when target can be identified in only visible images, i.e. ΔIR> ΔsMAXAnd ΔCCD≤ΔMAXWhen, with visible ray figure
The coordinate value of the original mark targeted graphical of deviation amendment of picture, with the targeted graphical of renewal in infrared and visible images subscript
Know the target of displacement;
Iv) when target can not be all identified in infrared and visible images, i.e. ΔIR> ΔsMAXAnd ΔCCD> ΔsMAXWhen, will be just
The beginning mark targeted graphical of position marks on infrared and visible images, and prompts image None- identified.
2. according to the method for claim 1, by adjust scaling multiple that visible images shows and position offset make it is red
Outer image f iotaeld-of-view matches with visible images visual field.
3. according to the method for claim 1, identify and preserve target signature data in figure with pattern identification initial target
With targeted graphical position as recognition template, and specify identification maximum position deviation ΔMAX。
4. according to the method for claim 1, by the To Template of preservation to target in infrared image and visible images
Shape is carried out respectively and border is found and identification, and it is inclined to calculate the current location of closest target and initial position in infrared image
Difference DELTAIRWith the current location of closest target in visible images and initial positional deviation value ΔCCD。
5. according to the method for claim 4, it further, judges whether target can be identified in current infrared image, such as
It is Δ that fruit identification position deviation, which is more than specified maximum deflection difference value,IR> ΔsMAX, will be regarded as target in current infrared image can not be by
Identification, the target location deviation of infrared image will not participate in target positioning correcting.
6. according to the method for claim 4, it further, judges whether target can be identified in current visible images,
If it is Δ to identify that position deviation is more than specified maximum deflection difference valueCCD> ΔsMAX, will be regarded as target in current visible images without
Method is identified, it is seen that the target location deviation of light image will not participate in target positioning correcting.
7. according to the method for claim 1, it further, the complementary identification of double vision intelligent image and target is called by repetition
Localization method, realize and target identification and positioning are carried out to multiple targets in same image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310246835.0A CN104239845B (en) | 2013-06-15 | 2013-06-15 | The identification of double vision intelligent image and object localization method based on infrared thermal imaging technique |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310246835.0A CN104239845B (en) | 2013-06-15 | 2013-06-15 | The identification of double vision intelligent image and object localization method based on infrared thermal imaging technique |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104239845A CN104239845A (en) | 2014-12-24 |
CN104239845B true CN104239845B (en) | 2017-12-15 |
Family
ID=52227874
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310246835.0A Active CN104239845B (en) | 2013-06-15 | 2013-06-15 | The identification of double vision intelligent image and object localization method based on infrared thermal imaging technique |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104239845B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2018200252A (en) * | 2017-05-29 | 2018-12-20 | アズビル株式会社 | Thermal image processing device and off-set amount calculation method |
CN109239669B (en) * | 2018-08-16 | 2020-11-03 | 厦门大学 | Self-evolution radar target detection algorithm based on deep learning |
CN109934296B (en) * | 2019-03-18 | 2023-04-07 | 江苏科技大学 | Method for identifying water surface personnel in multi-environment based on infrared and visible light images |
CN112649570A (en) * | 2020-12-11 | 2021-04-13 | 河海大学 | Tail gas detection device and method based on infrared thermal imaging double vision and ultrasonic positioning |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI237799B (en) * | 2003-11-11 | 2005-08-11 | Chung Shan Inst Of Science | Infrared image recognition system for remote building and method thereof |
CN101477631A (en) * | 2009-01-20 | 2009-07-08 | 深圳先进技术研究院 | Method, equipment for extracting target from image and human-machine interaction system |
CN102005054A (en) * | 2010-11-24 | 2011-04-06 | 中国电子科技集团公司第二十八研究所 | Real-time infrared image target tracking method |
US8355554B2 (en) * | 2009-04-14 | 2013-01-15 | Sonosite, Inc. | Systems and methods for adaptive volume imaging |
CN103065131A (en) * | 2012-12-28 | 2013-04-24 | 中国航天时代电子公司 | Method and system of automatic target recognition tracking under complex scene |
-
2013
- 2013-06-15 CN CN201310246835.0A patent/CN104239845B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI237799B (en) * | 2003-11-11 | 2005-08-11 | Chung Shan Inst Of Science | Infrared image recognition system for remote building and method thereof |
CN101477631A (en) * | 2009-01-20 | 2009-07-08 | 深圳先进技术研究院 | Method, equipment for extracting target from image and human-machine interaction system |
US8355554B2 (en) * | 2009-04-14 | 2013-01-15 | Sonosite, Inc. | Systems and methods for adaptive volume imaging |
CN102005054A (en) * | 2010-11-24 | 2011-04-06 | 中国电子科技集团公司第二十八研究所 | Real-time infrared image target tracking method |
CN103065131A (en) * | 2012-12-28 | 2013-04-24 | 中国航天时代电子公司 | Method and system of automatic target recognition tracking under complex scene |
Non-Patent Citations (1)
Title |
---|
基于红外与可见光图像融合的目标跟踪;陈少华;《中国优秀硕士学位论文全文数据库 信息科技辑(月刊)》;20130415;I138-939 * |
Also Published As
Publication number | Publication date |
---|---|
CN104239845A (en) | 2014-12-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110044486B (en) | Method, device and equipment for avoiding repeated alarm of human body inspection and quarantine system | |
WO2018076732A1 (en) | Method and apparatus for merging infrared image and visible light image | |
CN107992857A (en) | A kind of high-temperature steam leakage automatic detecting recognition methods and identifying system | |
US9432633B2 (en) | Visual command processing | |
WO2021008252A1 (en) | Method and apparatus for recognizing position of person in image, computer device and storage medium | |
CN104239845B (en) | The identification of double vision intelligent image and object localization method based on infrared thermal imaging technique | |
CN110142785A (en) | A kind of crusing robot visual servo method based on target detection | |
CN108510545B (en) | Space positioning method, space positioning apparatus, space positioning system, and computer-readable storage medium | |
US11600112B2 (en) | Bio-security system based on multi-spectral sensing | |
CN104253482A (en) | Image data base and inspection robot-based equipment trouble detection method | |
CN104573713B (en) | A kind of transformer Infrared image recognition based on image texture characteristic | |
CN111144207B (en) | Human body detection and tracking method based on multi-mode information perception | |
CN108731587A (en) | A kind of the unmanned plane dynamic target tracking and localization method of view-based access control model | |
CN107576269B (en) | Power transmission line forest fire positioning method | |
CN112836634B (en) | Multi-sensor information fusion gate anti-trailing method, device, equipment and medium | |
CN102446355B (en) | Method for detecting target protruding from plane based on double viewing fields without calibration | |
CN104966108A (en) | Visible light and infrared image fusion method based on gradient transfer | |
CN113688817A (en) | Instrument identification method and system for automatic inspection | |
CN110210292A (en) | A kind of target identification method based on deep learning | |
CN116310678A (en) | Fire source identification and positioning method by combining solid-state laser radar with thermal imaging vision | |
CN110320523A (en) | Follow the target locating set and method of robot | |
Chowdhury et al. | Robust human detection and localization in security applications | |
CN117314986A (en) | Unmanned aerial vehicle cross-mode power distribution equipment inspection image registration method based on semantic segmentation | |
CN113792684B (en) | Multi-mode visual flame detection method for fire-fighting robot under weak alignment condition | |
CN115830719B (en) | Building site dangerous behavior identification method based on image processing |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |