CN109472826A - Localization method and device based on binocular vision - Google Patents
Localization method and device based on binocular vision Download PDFInfo
- Publication number
- CN109472826A CN109472826A CN201811259210.7A CN201811259210A CN109472826A CN 109472826 A CN109472826 A CN 109472826A CN 201811259210 A CN201811259210 A CN 201811259210A CN 109472826 A CN109472826 A CN 109472826A
- Authority
- CN
- China
- Prior art keywords
- view
- visible light
- saliency maps
- component
- left view
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 27
- 230000004807 localization Effects 0.000 title claims abstract description 17
- 238000012545 processing Methods 0.000 claims abstract description 11
- 230000011218 segmentation Effects 0.000 claims description 34
- 230000009466 transformation Effects 0.000 claims description 13
- 238000009499 grossing Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 5
- 238000003709 image segmentation Methods 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 7
- 238000004458 analytical method Methods 0.000 abstract description 4
- 238000003384 imaging method Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 239000012491 analyte Substances 0.000 description 2
- 230000004888 barrier function Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 239000003595 mist Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002329 infrared spectrum Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 229910052710 silicon Inorganic materials 0.000 description 1
- 239000010703 silicon Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
Abstract
The invention discloses a kind of localization method and device based on binocular vision, is related to technical field of image processing.This method includes；Acquire the left view and right view of scene；Conspicuousness is carried out to the left view and right view respectively to calculate to obtain and the corresponding visible light Saliency maps of left view and near-infrared Saliency maps corresponding with the right view；Conspicuousness mean value is carried out to visible light Saliency maps and near-infrared Saliency maps respectively to compare, and the area to be targeted in left view and the right view is determined according to comparison result；The area to be targeted of area to be targeted and right view to left view is matched to determine target area in left view and the right view；Disparity computation is carried out to determine the position of target area to target area.Localization method and device provided by the invention based on binocular vision can be accurately detected the position of target, and only carry out the matching analysis to small part region in the view of left and right, and computation complexity is effectively reduced, and detection efficiency is high.
Description
Technical field
The present invention relates to technical field of image processing, more particularly, to a kind of localization method and dress based on binocular vision
It sets.
Background technique
Currently, binocular stereo vision has become a research hotspot of computer vision processing, it is based on human eye binocular
Parallax mode calculates correspondence of the spatial scene in different images by matching the Same Scene searched out in binocular image
Relationship obtains the D coordinates value of the point.
However, existing method would become hard to accomplish to be accurately positioned, leads to measurement inaccuracy, instrument under substation is read
For several and device temperature detection, the position of target area how is accurately detected out to realize to meter reading and equipment temperature
The detection of degree is a urgent problem to be solved in the prior art.
Summary of the invention
To solve the above-mentioned problems, the invention proposes a kind of localization methods based on binocular vision, comprising the following steps:
Step S1, the left view and right view of scene are acquired；
Step S2, the left view and right view progress conspicuousness are calculated respectively corresponding with the left view to obtain
Saliency maps picture and Saliency maps picture corresponding with the right view；
Step S3, respectively to the corresponding Saliency maps picture of the left view and Saliency maps picture corresponding with the right view
It carries out conspicuousness mean value to compare, and the area to be targeted in the left view and right view is determined according to comparison result；
Step S4, the area to be targeted to the left view and the area to be targeted of the right view are matched with determination
Target area in the left view and the right view；
Step S5, disparity computation is carried out to determine the position of target area to the target area in the left view and right view
It sets.
Preferably, any view is visible light view, another view in the left view and right view acquired in the step S1
Figure is near-infrared view.
Preferably, the step S2 is specifically included:
Step S2.1 is smoothed the visible light view by difference of Gaussian filtering；After smoothing processing
Visible light view transformation is the visible light view based on Lab color model；According to the visible light view based on Lab color model
The color mean value computation of the L * component of figure, a component and b component obtains the visible light Saliency maps；
Step S2.2, by the near-infrared view publishing at the mutually disjoint multiple segmentation blocks being made of 7*7 pixel
P, and using the pixel value of the central pixel point in each segmentation block as the pixel value of its place segmentation block；According to difference point
Cut the distance between the distance between the pixel value of block location of pixels, the diversity of more different segmentation blocks；By choose with
The immediate K pixel of current pixel point calculates the conspicuousness of the current pixel point in the near-infrared view, and circulation is held
Row is until obtain the near-infrared Saliency maps.
Preferably, the color mean value of the L * component, a component and b component passes through following formula respectively and obtains:
Wherein, m_L, m_a, m_b are respectively the color mean value of L * component, a component and b component, Iv_L, Iv_a, Iv_b difference
For the L * component, a component and b component of the visible light view, N is the line number of the visible light view, and M is visible light view
The columns of figure；
The visible light Saliency maps are obtained by following formula:
Sv(i, j)=[Iv_L (i, j)-m_L]2+[Iv_a(i,j)-m_a]2+[Iv_b(i,j)-m_b]2
Wherein, Sv(i, j) is the visible light conspicuousness of a pixel, and it is significant that the conspicuousness of all pixels point constitutes visible light
Property figure.
Preferably, the step S3 is specifically included:
Step S3.1 carries out conspicuousness mean value to the visible light Saliency maps and the near-infrared Saliency maps and compares
Compared with；
Step S3.2, according to the comparison result root to the visible light Saliency maps and the near-infrared Saliency maps
Image segmentation is carried out to obtain the area to be targeted in the left view and the right view.
Preferably, the step S5 is specifically included:
Step S5.1, the target area chosen in the left view and right view carry out disparity computation, obtain target area
Distance apart from collection point；
The visible light view transformation is the visible light view based on RGB color model by step S5.2；
Step S5.3 spatially to HVS by the visible light view transformation based on RGB color model extracts V component,
And the edge feature of the V component is matched to obtain the shape of the target area.
Further, this method further includes step S7, is analyzed the distance and shape of the target area to determine
The width for stating target area is dominant to be dominant with height.
The invention also provides a kind of positioning devices based on binocular vision, comprising:
Image collecting device, for acquiring the left view and right view of scene；
Conspicuousness image collection module, for the left view and right view carry out conspicuousness calculate with obtain with it is described
The corresponding Saliency maps of left view and Saliency maps picture corresponding with the right view；
Area to be targeted determining module, for the corresponding Saliency maps picture of the left view and corresponding with the right view
Saliency maps picture carry out conspicuousness mean value and compare, and determined according to comparison result to be positioned in the left view and right view
Region；
Target area determining module, the area to be targeted for area to be targeted and the right view to the left view
It is matched with the target area in the determination left view and the right view；
Target-region locating module, for carrying out disparity computation to the target area in the left view and right view with true
Set the goal the position in region.
Preferably, any view is visible light view in the left view and right view, and another view is near-infrared view.
Preferably, shown conspicuousness image collection module is filtered by difference of Gaussian and is carried out smoothly to the visible light view
Processing；It is the visible light view based on Lab color model by the visible light view transformation after smoothing processing；It is based on according to described
The color mean value computation of the L * component of the visible light view of Lab color model, a component and b component obtains the visible light conspicuousness
Figure；By the near-infrared view publishing at the mutually disjoint multiple segmentation block p being made of 7*7 pixel, and with described each
Divide pixel value of the pixel value of the central pixel point in block as segmentation block where it；According to it is different segmentation blocks pixel values it
Between distance and the distance between location of pixels, the diversities of more different segmentation blocks；By choosing and the near-infrared view
The middle immediate K pixel of current pixel point calculates the conspicuousness of the current pixel point, and circulation executes until obtaining described
Near-infrared Saliency maps.
The present invention has the advantage that and the utility model has the advantages that
The device of the invention structure is simple, versatile, and to good environmental adaptability, because near-infrared is in mist and low illumination
Imaging effect is better than visible light in degree.Above method step of the invention is simple, it is easy to accomplish, various barriers can be effectively treated
Hinder analyte detection problem.
Localization method and device provided by the invention based on binocular vision can be accurately detected the position of target, and
The matching analysis only is carried out to small part region in the view of left and right, computation complexity is effectively reduced, detection efficiency is high.
Detailed description of the invention
Attached drawing described herein is used to provide to further understand the embodiment of the present invention, constitutes one of the application
Point, do not constitute the restriction to the embodiment of the present invention.In the accompanying drawings:
Fig. 1 is the flow chart of the localization method provided in an embodiment of the present invention based on binocular vision；
Fig. 2 is the functional block diagram of the positioning device provided in an embodiment of the present invention based on binocular vision.
Description of symbols:
Positioning device of the 600- based on binocular vision；610- image collecting device；620- conspicuousness image collection module；
The area to be targeted 630- determining module；The target area 640- determining module；650- target-region locating module.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, below with reference to embodiment and attached drawing, to this
Invention is described in further detail, and exemplary embodiment of the invention and its explanation for explaining only the invention, are not made
For limitation of the invention.
The component of embodiments of the present invention, which are generally described and illustrated herein in the accompanying drawings can be come with a variety of different configurations
Arrangement and design.
Therefore, the detailed description of the embodiment of the present invention provided in the accompanying drawings is not intended to limit below claimed
The scope of the present invention, but be merely representative of selected embodiment of the invention.Based on the embodiments of the present invention, this field is common
Technical staff's every other embodiment obtained without creative efforts belongs to the model that the present invention protects
It encloses.
It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, does not then need that it is further defined and explained in subsequent attached drawing.
Term " first ", " second ", " third " etc. are only used for distinguishing description, are not understood to indicate or imply relatively heavy
The property wanted.
In addition, term " parallel ", " vertical " etc. are not offered as requiring component absolute parallel or vertical, but can slightly incline
Tiltedly.It is not to indicate that the structure has to put down completely if " parallel " only refers to that its direction is more parallel with respect to for " vertical "
Row, but can be slightly tilted.
In the description of the present invention, it is also necessary to which explanation is unless specifically defined or limited otherwise, term " setting ",
" installation ", " connected ", " connection " shall be understood in a broad sense, for example, it may be fixedly connected, may be a detachable connection or one
Connect to body；It can be directly connected, the connection inside two elements can also be can be indirectly connected through an intermediary.
For the ordinary skill in the art, the concrete meaning of above-mentioned term in the present invention can be understood with concrete condition.
As shown in Figure 1, the flow chart of the localization method based on binocular vision for the embodiment of the present invention.The embodiment of the present invention
The localization method based on binocular vision, comprising the following steps:
Step S101, acquires the left view and right view of scene, and in the present embodiment, the left view is visible images,
The right view is near-infrared image.For example, passing through hybrid binocular vision (left view and right view of scene) acquisition system
Acquire the left and right view of scene.The acquisition system is that visible light and near infrared spectrum imaging device is respectively adopted in left and right visual angle, is had
Parallel optical axis imaging can be used in body imaging device.As current general silicon CCD can incude near infrared band, it is possible to filter
Except the optical filtering that visible light remains near-infrared loads the acquisition of the realization near-infrared image before general black and white camera.
It is corresponding with the left view to obtain to carry out conspicuousness calculating to the left view and right view respectively by step S102
Visible light Saliency maps and near-infrared Saliency maps corresponding with the right view.
Generally, for calculating visible light Saliency maps, because visible images have colouring information abundant,
Therefore it can be estimated according to it in current real-time image acquisition relative to background color region more outstanding as approximate human eye
Saliency maps, be denoted as Sv, and then distinguish the salient region in visible images as possible area to be targeted, be denoted as
Rv。
Specifically, calculating visible light Saliency maps in the following manner: being filtered by difference of Gaussian to the left view
It is smoothed.Then it is the left view based on Lab color model by the left view transformation after smoothing processing, i.e., becomes image
Change to the L*a*b color space for being more in line with human vision rule.The last left view based on Lab color model according to
The color mean value computation of L * component, a component and b component obtains the visible light Saliency maps, i.e., according in L*a*b color space
Under, each component is worth variation to be used as color conspicuousness component color, calculates final color Saliency maps.
In this embodiment, the color mean value of L * component, a component and b component passes through following formula respectively and obtains:
Wherein, m_L, m_a, m_b are respectively the color mean value of L * component, a component and b component, Iv_L, Iv_a, Iv_b difference
For the L * component, a component and b component of the visible images, N is the line number of the visible images, and M is the visible light figure
The columns of picture；I and j is respectively the position of corresponding pixel.
The visible light Saliency maps are obtained by following formula:
Sv(i, j)=[Iv_L (i, j)-m_L]2+[Iv_a(i,j)-m_a]2+[Iv_b(i,j)-m_b]2
Wherein, Sv(i, j) is the visible light conspicuousness of a pixel, and it is significant that the conspicuousness of all pixels point constitutes visible light
Property figure.
In another example of the invention, for calculating near-infrared Saliency maps because its there are also good high frequencies
Detailed information, it is possible to salient region is detected according to the similitude of each pixel and peripheral region, is denoted as Snir,
Important region is found out from Snir as area to be targeted, is denoted as Rnir.
Specifically, near-infrared Saliency maps are calculated in the following manner:
The right view is divided into the mutually disjoint multiple segmentation blocks being made of 7*7 pixel, and with described each
Divide pixel value of the pixel value of the central pixel point in block as segmentation block where it, i.e., takes 7* centered on each pixel
Representative of 7 adjacent domain as the point.The distance between pixel value then according to different segmentation blocks is between location of pixels
Distance, the diversities of more different segmentation blocks, wherein the distance between pixel value of different segmentation block is with each segmentation block
The Euclidean distance d between pixel vector is formed by columnv(pi,pj), the distance between described location of pixels is with different central pixel points
Between coordinate Euclidean distance dp(pi,pj), wherein piAnd pjFor different segmentation blocks, that is to say, that according between pixel value
Distance and the distance between location of pixels, compare diversity between points.Wherein the distance between pixel value is pressed with block
Column composition vector, calculates the Euclidean distance d being worth between pixel vectorv(pi,pj), and positional distance, then directly between two points
Coordinate Euclidean distance indicate dp(pi,pj).It is a with the immediate K of current pixel point in the right view finally by choosing
Pixel calculates the conspicuousness of the current pixel point, and circulation is executed up to obtaining the near-infrared Saliency maps, wherein K=
30。
In the above-described embodiments, diversity (different degree) is obtained by following formula:
Wherein, d (pi,pj) it is segmentation block piAnd pjBetween diversity, d (pi,pj) the distance that both shows more greatly of value
Bigger, then also bigger with regard to difference, similitude is smaller.
The calculation formula of the conspicuousness are as follows:
Wherein, SniriTo divide block piConspicuousness.
Step S103 carries out conspicuousness average ratio to the visible light Saliency maps and the near-infrared Saliency maps respectively
Compared with, and the area to be targeted in the left view and the right view is determined according to comparison result.In other words, based on having obtained
The visible light Saliency maps and the near-infrared Saliency maps, the mean value using Saliency maps itself obtained compare, and carry out image point
It cuts, obtains the area to be targeted of left and right view.In an example of the invention, it is contemplated that evacuation safety, it can all
Energy region is first taken by the ranks maximum difference of pixel coordinate rectangular.
Specifically, conspicuousness average ratio is carried out to the visible light Saliency maps and the near-infrared Saliency maps first
Compared with, then according to comparison result respectively to the visible light Saliency maps and the near-infrared Saliency maps carry out image segmentation with
Obtain the area to be targeted in the left view and the right view.
Step S104, the area to be targeted of area to be targeted and the right view to the left view matched with
Target area is determined in the left view and the right view.
Specifically, using area to be targeted obtained in above-described embodiment, respectively with the area to be positioned in the view of left and right
On the basis of domain, matching area is found in another figure.If it find that matching area itself is also area to be targeted in another figure,
Then taking the region is final target area.
Step S105 carries out disparity computation to the target area to determine that (collection point is i.e. apart from collection point for target area
Binocular) distance, and to the area to be targeted carry out edge matching with the shape of the determination target area.
Specifically, it is determined that the left view transformation is first based on RGB color model by the shape of the target area
Left view；By it is described based on the left view transformation of RGB color model to HVS spatially, V component is extracted, and to the V component
Edge feature is matched to obtain the shape of the target area.
More specifically, the final target area of binocular image centering is chosen, disparity computation is carried out, obtains target area
Current distance.Left view is transformed to RGB color image first, then transforms to HVS spatially, V component is extracted, is regarded as
Visible images and infrared image associated picture.At this point, binocular image can be understood as area to be targeted V component image, it is denoted as
Rv and near-infrared area to be targeted, are denoted as Rnir.Because near-infrared image largely reflects the high-frequency information of image,
So being matched using edge feature, target area shape is acquired.
In one embodiment of the invention, after determining the distance and shape in final goal region, to final goal
The distance and shape in region are further analyzed, and are dominant with determining that the width of target area is dominant with height.
As shown in Fig. 2, another aspect of the present invention correspondingly proposes a kind of positioning device based on binocular vision, with reference to figure
2, the positioning device 600 based on binocular vision of the embodiment of the present invention includes image collecting device 610, Saliency maps picture acquisition mould
Block 620, area to be targeted determining module 630, target area determining module 640 and target-region locating module 650.
Image collecting device 610 is used to acquire the left view and right view of scene, wherein the left view is visible light figure
Picture, the right view are near-infrared image.Conspicuousness image collection module 620 is for showing the left view and right view
Write property calculate with obtain and the corresponding visible light Saliency maps of the left view and near-infrared corresponding with the right view it is significant
Property figure.Area to be targeted determining module 630 is for showing the visible light Saliency maps and the near-infrared Saliency maps
Work property mean value compares, and determines the area to be targeted in the left view and the right view respectively according to comparison result.Mesh
Mark area determination module 640 is for matching the area to be targeted of the left view and the area to be targeted of the right view
To determine target area in the left view and the right view.Target-region locating module 650 is used for the target
Region carries out disparity computation to determine distance of the target area apart from collection point, and carries out edge matching to the area to be targeted
With the shape of the determination target area.
In one embodiment of the invention, conspicuousness image collection module 620 to left view carry out conspicuousness calculate with
Obtain visible light Saliency maps, comprising: be smoothed to the left view by difference of Gaussian filtering；After smoothing processing
Left view transformation be the left view based on Lab color model；According to the L * component of the left view based on Lab color model,
The color mean value computation of a component and b component obtains the visible light Saliency maps.
In the above-described embodiments, the color mean value of L * component, a component and b component passes through following formula respectively and obtains:
Wherein, m_L, m_a, m_b are respectively the color mean value of L * component, a component and b component, Iv_L, Iv_a, Iv_b difference
For the L * component, a component and b component of the visible images, N is the line number of the visible images, and M is the visible light figure
The columns of picture；I and j is respectively the position of corresponding pixel.
The visible light Saliency maps are obtained by following formula:
Sv(i, j)=[Iv_L (i, j)-m_L]2+[Iv_a(i,j)-m_a]2+[Iv_b(i,j)-m_b]2
Wherein, Sv(i, j) is the visible light conspicuousness of a pixel, and it is significant that the conspicuousness of all pixels point constitutes visible light
Property figure.
In another embodiment of the invention, conspicuousness image collection module 620 to right view carry out conspicuousness calculate with
Obtain near-infrared Saliency maps, comprising: first by the right view be divided into it is mutually disjoint be made of 7*7 pixel it is multiple
Divide block, and using the pixel value of segmentation block where the pixel value as its of the central pixel point in each segmentation block, then
The distance between the distance between pixel value according to different segmentation blocks location of pixels, the diversity of more different segmentation blocks,
Wherein the distance between pixel value of different segmentation blocks is the Euclidean distance d formed with each segmentation block by column between pixel vectorv
(pi,pj), the distance between described location of pixels is with the Euclidean distance d of the coordinate between different central pixel pointsp(pi,pj),
Wherein pi and pj is different segmentation block, finally by selection and the immediate K pixel of current pixel point in the right view
Point calculates the conspicuousness of the current pixel point, and circulation is executed up to obtaining the near-infrared Saliency maps, wherein K=30.
In the above-described embodiments, diversity (different degree) is obtained by following formula:
Wherein, d (pi,pj) it is segmentation block piAnd pjBetween diversity, d (pi,pj) the distance that both shows more greatly of value
Bigger, then also bigger with regard to difference, similitude is smaller.
The calculation formula of conspicuousness are as follows:
Wherein, SniriTo divide block piConspicuousness.
In one embodiment of the invention, area to be targeted determining module 630 is used for the visible light Saliency maps
Conspicuousness mean value is carried out with the near-infrared Saliency maps to compare；And according to comparison result respectively to the visible light Saliency maps
Image segmentation is carried out with the near-infrared Saliency maps to obtain the area to be targeted in the left view and the right view.
In specific example of the invention, target-region locating module 650 is used to the left view transformation be based on RGB
The left view of color model；And by it is described based on the left view transformation of RGB color model to HVS spatially, extract V component, and
The edge feature of the V component is matched to obtain the shape of the target area.
Embodiment according to the present invention can be accurately detected the position of target area, and only to portion few in the view of left and right
Subregion carries out the matching analysis, and computation complexity is effectively reduced, and detection efficiency is high.
Specifically, the above embodiment of the present invention has the advantages that
The device of the invention structure is simple, versatile, and to good environmental adaptability, because near-infrared is in mist and low illumination
Imaging effect is better than visible light in degree.Above method step of the invention is simple, it is easy to accomplish, various barriers can be effectively treated
Hinder analyte detection problem.In addition, this method only carries out the matching analysis to salient region of image, therefore processing capability in real time is improved,
And method of the invention is only handled visual pattern, independent of other knowledge, therefore the independence of improvement method and logical
It is applied widely with property.
Above-described specific embodiment has carried out further the purpose of the present invention, technical scheme and beneficial effects
It is described in detail, it should be understood that being not intended to limit the present invention the foregoing is merely a specific embodiment of the invention
Protection scope, all within the spirits and principles of the present invention, any modification, equivalent substitution, improvement and etc. done should all include
Within protection scope of the present invention.
Claims (10)
1. the localization method based on binocular vision, which comprises the following steps:
Step S1, the left view and right view of scene are acquired；
Step S2, the left view and right view progress conspicuousness are calculated respectively corresponding with the left view significant to obtain
Property image and Saliency maps picture corresponding with the right view；
Step S3, the corresponding Saliency maps picture of the left view and Saliency maps picture corresponding with the right view are carried out respectively
Conspicuousness mean value compares, and determines the area to be targeted in the left view and right view according to comparison result；
Step S4, the area to be targeted to the left view and the area to be targeted of the right view are matched described in determination
Target area in left view and the right view；
Step S5, disparity computation is carried out to determine the position of target area to the target area in the left view and right view.
2. the localization method according to claim 1 based on binocular vision, which is characterized in that acquired in the step S1
Any view is visible light view in left view and right view, and another view is near-infrared view.
3. the localization method according to claim 2 based on binocular vision, which is characterized in that the step S2 is specifically wrapped
It includes:
Step S2.1 is smoothed the visible light view by difference of Gaussian filtering；It will be visible after smoothing processing
Light view transformation is the visible light view based on Lab color model；According to the visible light view based on Lab color model
The color mean value computation of L * component, a component and b component obtains the visible light Saliency maps；
Step S2.2, by the near-infrared view publishing at the mutually disjoint multiple segmentation block p being made of 7*7 pixel, and
Using the pixel value of segmentation block where the pixel value as its of the central pixel point in each segmentation block；According to different segmentation blocks
The distance between the distance between pixel value location of pixels, the diversities of more different segmentation blocks；By choose with it is described
The immediate K pixel of current pixel point calculates the conspicuousness of the current pixel point in near-infrared view, and circulation executes straight
To obtaining the near-infrared Saliency maps.
4. the localization method according to claim 3 based on binocular vision, which is characterized in that the L * component, a component and b
The color mean value of component passes through following formula respectively and obtains:
Wherein, m_L, m_a, m_b are respectively the color mean value of L * component, a component and b component, and Iv_L, Iv_a, Iv_b are respectively institute
The L * component, a component and b component of visible light view are stated, N is the line number of the visible light view, and M is the visible light view
Columns；
The visible light Saliency maps are obtained by following formula:
Sv(i, j)=[Iv_L (i, j)-m_L]2+[Iv_a(i,j)-m_a]2+[Iv_b(i,j)-m_b]2
Wherein, Sv(i, j) is the visible light conspicuousness of a pixel, and the conspicuousness of all pixels point constitutes visible light Saliency maps.
5. the localization method according to claim 3 based on binocular vision, which is characterized in that the step S3 is specifically wrapped
It includes:
Step S3.1 carries out conspicuousness mean value to the visible light Saliency maps and the near-infrared Saliency maps and is compared；
Step S3.2 carries out the visible light Saliency maps and the near-infrared Saliency maps according to the comparison result root
Image segmentation is to obtain the area to be targeted in the left view and the right view.
6. the localization method according to claim 2 based on binocular vision, which is characterized in that the step S5 is specifically wrapped
It includes:
Step S5.1, the target area chosen in the left view and right view carry out disparity computation, obtain target area distance
The distance of collection point；
The visible light view transformation is the visible light view based on RGB color model by step S5.2；
Step S5.3 spatially to HVS by the visible light view transformation based on RGB color model extracts V component, and right
The edge feature of the V component is matched to obtain the shape of the target area.
7. the localization method according to claim 1 based on binocular vision, which is characterized in that this method further includes step
S7, distance and shape to the target area are analyzed to be dominant and height is dominant with the width of the determination target area.
8. the positioning device based on binocular vision characterized by comprising
Image collecting device, for acquiring the left view and right view of scene；
Conspicuousness image collection module is calculated for carrying out conspicuousness to the left view and right view to obtain and the left view
Scheme corresponding Saliency maps and Saliency maps picture corresponding with the right view；
Area to be targeted determining module, for the corresponding Saliency maps picture of the left view and corresponding with the right view aobvious
Work property image carries out conspicuousness mean value and compares, and the area to be positioned in the left view and right view is determined according to comparison result
Domain；
Target area determining module, the area to be targeted for area to be targeted and the right view to the left view carry out
Matching is with the target area in the determination left view and the right view；
Target-region locating module, for carrying out disparity computation to the target area in the left view and right view to determine mesh
Mark the position in region.
9. the positioning device according to claim 8 based on binocular vision, which is characterized in that the left view and right view
In any view be visible light view, another view be near-infrared view.
10. the positioning device according to claim 9 based on binocular vision, which is characterized in that shown Saliency maps picture obtains
Modulus block is smoothed the visible light view by difference of Gaussian filtering；Visible light view after smoothing processing is become
It is changed to the visible light view based on Lab color model；According to L * component, a of the visible light view based on Lab color model
The color mean value computation of component and b component obtains the visible light Saliency maps；By the near-infrared view publishing at mutually not phase
Hand over the multiple segmentation block p being made of 7*7 pixel, and using it is described it is each segmentation block in central pixel point pixel value as
The pixel value of segmentation block where it；The distance between the distance between pixel value according to different segmentation blocks location of pixels, than
The diversity of more different segmentation blocks；By choosing and the immediate K pixel meter of current pixel point in the near-infrared view
The conspicuousness of the current pixel point is calculated, circulation executes until obtaining the near-infrared Saliency maps.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811259210.7A CN109472826A (en) | 2018-10-26 | 2018-10-26 | Localization method and device based on binocular vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811259210.7A CN109472826A (en) | 2018-10-26 | 2018-10-26 | Localization method and device based on binocular vision |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109472826A true CN109472826A (en) | 2019-03-15 |
Family
ID=65666063
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811259210.7A Pending CN109472826A (en) | 2018-10-26 | 2018-10-26 | Localization method and device based on binocular vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109472826A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111583342A (en) * | 2020-05-14 | 2020-08-25 | 中国科学院空天信息创新研究院 | Target rapid positioning method and device based on binocular vision |
| CN112504472A (en) * | 2020-11-26 | 2021-03-16 | 浙江大华技术股份有限公司 | Thermal imager, thermal imaging method and storage medium |
| CN113834571A (en) * | 2020-06-24 | 2021-12-24 | 杭州海康威视数字技术股份有限公司 | Target temperature measurement method, device and temperature measurement system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102435174A (en) * | 2011-11-01 | 2012-05-02 | 清华大学 | Method and device for detecting barrier based on hybrid binocular vision |
| CN106780476A (en) * | 2016-12-29 | 2017-05-31 | 杭州电子科技大学 | A kind of stereo-picture conspicuousness detection method based on human-eye stereoscopic vision characteristic |
| CN108573221A (en) * | 2018-03-28 | 2018-09-25 | 重庆邮电大学 | A kind of robot target part conspicuousness detection method of view-based access control model |
-
2018
- 2018-10-26 CN CN201811259210.7A patent/CN109472826A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102435174A (en) * | 2011-11-01 | 2012-05-02 | 清华大学 | Method and device for detecting barrier based on hybrid binocular vision |
| CN106780476A (en) * | 2016-12-29 | 2017-05-31 | 杭州电子科技大学 | A kind of stereo-picture conspicuousness detection method based on human-eye stereoscopic vision characteristic |
| CN108573221A (en) * | 2018-03-28 | 2018-09-25 | 重庆邮电大学 | A kind of robot target part conspicuousness detection method of view-based access control model |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111583342A (en) * | 2020-05-14 | 2020-08-25 | 中国科学院空天信息创新研究院 | Target rapid positioning method and device based on binocular vision |
| CN111583342B (en) * | 2020-05-14 | 2024-02-23 | 中国科学院空天信息创新研究院 | Target rapid positioning method and device based on binocular vision |
| CN113834571A (en) * | 2020-06-24 | 2021-12-24 | 杭州海康威视数字技术股份有限公司 | Target temperature measurement method, device and temperature measurement system |
| WO2021259365A1 (en) * | 2020-06-24 | 2021-12-30 | 杭州海康威视数字技术股份有限公司 | Target temperature measurement method and apparatus, and temperature measurement system |
| CN112504472A (en) * | 2020-11-26 | 2021-03-16 | 浙江大华技术股份有限公司 | Thermal imager, thermal imaging method and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107463918B (en) | Lane line extraction method based on fusion of laser point cloud and image data | |
| CN106485275B (en) | A method of realizing that cover-plate glass is bonded with liquid crystal display positioning | |
| CN102435174B (en) | Method and device for detecting barrier based on hybrid binocular vision | |
| CN104848851B (en) | Intelligent Mobile Robot and its method based on Fusion composition | |
| CN107907048A (en) | A kind of binocular stereo vision method for three-dimensional measurement based on line-structured light scanning | |
| CA2867365C (en) | Method, system and computer storage medium for face detection | |
| CN104200461B (en) | The remote sensing image registration method of block and sift features is selected based on mutual information image | |
| CN110647850A (en) | Automatic lane deviation measuring method based on inverse perspective principle | |
| CN107084680B (en) | A kind of target depth measurement method based on machine monocular vision | |
| CN109472826A (en) | Localization method and device based on binocular vision | |
| CN104361314A (en) | Method and device for positioning power transformation equipment on basis of infrared and visible image fusion | |
| CN109961399B (en) | Optimal suture line searching method based on image distance transformation | |
| CN105913013A (en) | Binocular vision face recognition algorithm | |
| CN103927758B (en) | Saliency detection method based on contrast ratio and minimum convex hull of angular point | |
| CN102999916A (en) | Edge extraction method of color image | |
| JPH0877334A (en) | Automatic feature point extracting method for face image | |
| CN104268853A (en) | Infrared image and visible image registering method | |
| CN104764407B (en) | A kind of fine measuring method of thickness of cable sheath | |
| CN111462503A (en) | Vehicle speed measuring method and device and computer readable storage medium | |
| CN106970620A (en) | A kind of robot control method based on monocular vision | |
| CN110119190A (en) | Localization method, device, recognition and tracking system and computer-readable medium | |
| CN117036641A (en) | Road scene three-dimensional reconstruction and defect detection method based on binocular vision | |
| CN109389165A (en) | Oil level gauge for transformer recognition methods based on crusing robot | |
| CN105654479A (en) | Multispectral image registering method and multispectral image registering device | |
| CN113344990B (en) | Hole site representation projection system and self-adaptive fitting hole site alignment method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190315 |
|
| RJ01 | Rejection of invention patent application after publication |