CN108257170A - Real-time volume matching failure predication and solution based on orthogonal stereo device - Google Patents
Real-time volume matching failure predication and solution based on orthogonal stereo device Download PDFInfo
- Publication number
- CN108257170A CN108257170A CN201810045950.4A CN201810045950A CN108257170A CN 108257170 A CN108257170 A CN 108257170A CN 201810045950 A CN201810045950 A CN 201810045950A CN 108257170 A CN108257170 A CN 108257170A
- Authority
- CN
- China
- Prior art keywords
- matching
- orthogonal
- failure
- stereo
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
- G06T7/55—Depth or shape recovery from multiple images
- G06T7/593—Depth or shape recovery from multiple images from stereo images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/35—Determination of transform parameters for the alignment of images, i.e. image registration using statistical methods
-
- 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
- G06T2207/10012—Stereo images
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Image Processing (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The present invention proposes a kind of real-time volume matching failure predication and solution based on orthogonal stereo device, includes the following steps：Orthogonal photographic device, field programmable gate array, half global registration algorithm, Gaussian Blur core.The weak texture of linear structure can cause Stereo matching to break down, real-time volume matching failure predication and solution based on orthogonal stereo device, its process is to be connected to single field programmable logic sequence by the low-voltage differential signal interface of four imaging sensors of orthogonal photographic device, FPGA resynchronizes four independent data flows and carries out biased operation to image, with half global registration algorithm to the depth map of image sensor scan line and Continuous plus new line into row buffering, it optimizes in a plurality of directions, finally the failure of weak texture region is avoided to occur using the red indicator of texture and Gaussian Blur core.The present invention can effectively solve the problem that the erroneous matching problem of weak texture region, improve Stereo matching precision and accuracy.
Description
Technical field
The invention belongs to Stereo Matching Technology fields, and in particular to a kind of real-time volume matching based on orthogonal stereo device
Failure predication and solution.
Background technology
Stereo Matching Technology is always one of research hotspot of three-dimensional scene structure acquisition of information and computer vision neck
An of great value hot issue in domain, in industry, medical equipment, monitoring system, vision guided navigation, human-computer interaction and unmanned plane
Etc. be widely used, basic principle is to match phase from two same scenery of viewing point to obtain stereogram
Picture point is answered, so as to calculate parallax and obtain three-dimensional information, Stereo matching generally includes four steps：
(1) calculating of Matching power flow；
(2) polymerization of Matching power flow；
(3) calculating and optimization of parallax
(4) the refinement refinement of parallax.
Stereo Matching Algorithm can be mainly divided into three classes, local algorithm, Global Algorithm and half Global Algorithm.Local algorithm base
It polymerize Matching power flow in certain window, arithmetic speed is fast, but low precision.The appearance of Global Algorithm improves the standard of Stereo matching
True property, however slower real-time limits its application in actual scene.The it is proposed of half Global Algorithm effectively balances
The speed of Stereo matching and the relationship of precision, half Global Algorithm have strong robustness and the advantage insensitive to illumination effect, but
There are still weak texture region and repeat the problem of texture region characteristic point error hiding rate is high.
Therefore, applicant proposes a kind of real-time volume matching failure predication and solution based on orthogonal stereo device,
While raising precision avoids failure, run time can also increase substantially for it.
Invention content
In view of the deficiencies of the prior art, the present invention proposes that a kind of real-time volume matching failure based on orthogonal stereo device is pre-
Survey and solution can effectively solve the problem that the error hiding problem of weak texture region, improve Stereo matching precision.
In order to solve the failure that weak texture region occurs in Stereo matching, the present invention proposes a kind of based on orthogonal stereo device
Real-time volume matching failure predication and solution, include the following steps：
S1, orthogonal photographic device；
S2, field programmable gate array；
S3, half global registration algorithm；
S4, Gaussian Blur core；
Being connected to single scene by the low-voltage differential signal interface of four imaging sensors of orthogonal photographic device can compile
Journey logic sequence, FPGA resynchronize four independent data flows and carry out biased operation to image, with half global registration algorithm to figure
As the depth map of sensor scan line and Continuous plus new line is into row buffering, optimizes, finally use in a plurality of directions
The red indicator of texture and Stereo Matching Algorithm avoid the failure of weak texture region from occurring.
Further, the orthogonal photographic device, as shown in figure 3, being sensor camera shooting of the band there are two orthogonal polar curve
Head obtains two width digital pictures simultaneously by different angle, in order to obtain the three-dimensional coordinate that certain in three dimensions is put, needs in left and right
All there are the response point of the point in two video camera image planes, so being necessarily mounted on a stabilised platform, half is global three-dimensional
Matching algorithm is calculated and is run, and pass through USB3.0 synchronism outputs directly on two hardware.
All four video cameras are all corrected using Inline Function, using being obtained in two-dimensional marker initial calibration operational process
Extrinsic data, using half global Stereo Matching Algorithm within hardware calculated level pair and vertically to Stereo matching, count
Two depth maps and two confidence level figures calculated are transferred to by USB3.0 in the computer of connection.
Multiple-camera stereoscopic device merges depth map with the weighted average of each three-dimensional pair of matching score, this is real-time
The problem of previous work of depth map fusion, this method is very high therefore matched along the matching score of straight line
Score provides a unsuitable weighting for depth map fusion.
Above-mentioned adaptation function scores according to failure in normal operating to be selected a source images j is normal for each pixel i
Running, if however, depth map all in the range of mutual 10%, uses its matching score mi,jThe weighting for calculating them is put down
Mean value merges these pixels, this provides a firm cross-check for estimation of Depth, also allows to receive not having by SGM
The pixel of the balloon score of distribution, the step for significantly improve recall rate.
The present invention uses additional video camera, the accuracy and recall rate of single Baseline Stereo Matching system is improved, similary
Hardware on perform all calculating, only increase by two models of cost and weight, the inexpensive image on the three-dimensional magnetic head of standard
Sensor, can successfully avoid the area that mistake measures and three-dimensional associated confidence in the picture is low increases recall rate,
Obtain reliable measurement result.
Further, the field programmable gate array, in order to understand the matched basic constraint of real-time volume on hardware,
The present invention introduces the operation field programmable gate array of Stereo matching on hardware, and field programmable gate array (FPGA) is being handled
The whole frame of video camera is not stored before, it only buffers seldom image sensor scan line and Continuous plus new line enters
The depth map of system, this allows to rapidly process up to 10 video cameras, but its access and storage to image with a chip
Mode has added some to limit.
Not all Stereo Matching Algorithm is suitable for such operation, but can be calculated with half global registration
Method, by the depth map optimized in a plurality of directions, it improves Block- matching, wherein a paths are parallel to polar curve side
To.
Wherein, it is described to be constrained in substantially this refers to texturing element, imaging mode and field in stereo visual system
Geometrical relationship in scape can form the constraints of Stereo matching, they are improving matching just to reducing the calculation amount searched for
True property has very big help, and polar curve condition is the primary condition that geometry imaging should follow, can be the search of Stereo matching
Range dropped to from two dimension it is one-dimensional, if there are one texture along to polar curve be repeat, it meet texture and matching score criteria,
But matching score can be very high, show that this essential attribute is inevitable, texture scene be in city it is very common, so
And it has enough characteristics to lead to Stereo matching dependent failure, this problem has obtained grinding well in the structure of motion field
Study carefully, and often implicitly solved in photo or so consistency check, however, because focus on the application of the present invention is that machine regards
Feel field, therefore the present invention proposes a kind of new, more effective way using the attribute of real-time volume algorithm, unmanned plane and its
He possess single Stereo matching pair autonomous driving vehicle or system all epipolar geom etry is constrained by Stereo matching.
System calculates a prognostic chart picture to each depth map, creates ultimate depth figure and final matching fraction figure, often
The final matching fraction figure c of a location of pixelsi,jIt is the matching score m of the image a or b of half matching Global Algorithmi,jWeighting put down
Mean value, from relatively low failure scoring fi,jImage in selected, which is a united value letter
Number：
ci,j=mi,j+s·fi,j (1)
S=2 provides a stable tradeoff, with the value from the matching score and failure predication of half matching Global Algorithm
All experiments that fixed parameter performs, the region that obtained image becomes clear represent the three-dimensional relevant area for being likely to occur failure
Domain.
Further, half global registration algorithm, what the present invention determined epipolar geom etry using the matching score that SGM is provided can
Energy failure, when it encounters the line parallel with polar curve, matching score can reach peak value：Matching point between alternative route of the present invention
Number, to determine that estimation is likely to occur the region of failure.
SGM algorithms carry out operation to the gray value of image, do not propagate parallax value in the big gradient of image.However, along
The false disparity correspondence in EP point direction does not include any vertical gradient, and SGM algorithms propagate mistake along the direction of EP point
Match, failure is detected using this specific properties of its smoothing step.
Further, Gaussian Blur core, qualitative evaluation show that above-mentioned SGM matchings score is all best in all scenes
Predictive factor, when inputting depth map by medium filtering, the present invention using 3 δ (Pauta criterion) 13 × 13 Gausses
Fuzzy core matches the scale space of score image to adjust, and most probable abort situation, utilization are previously discussed in next step for prediction
It half Global Algorithm attribute rather than is optimized in strong gradient, result is also feasible in image score, by finding this
The gradient magnitude of a little score differences, can reliably predict the cable architecture for being parallel to polar curve.In order to reach this purpose, the present invention will
Integral image is rolled up with one-dimensional Sobel cores in vertical direction and the EP point with 7 pixel regions and the Gauss of 0.3 δ
Product, the significant changes of matching score along EP point are the reliable predictions measured mistake.
The beneficial effects of the invention are as follows：The present invention can effectively solve the problem that the error hiding problem of weak texture region, improve three-dimensional
Matching precision.
Description of the drawings
Fig. 1 is a kind of system of real-time volume matching failure predication and solution based on orthogonal stereo device of the present invention
Frame diagram.
Fig. 2 is a kind of solid of real-time volume matching failure predication and solution based on orthogonal stereo device of the present invention
Matching constraint situation schematic diagram.
Fig. 3 is a kind of algorithm of real-time volume matching failure predication and solution based on orthogonal stereo device of the present invention
Flow chart.
Specific embodiment
The present invention is described in further detail with specific implementation below in conjunction with the accompanying drawings.
Fig. 1 is a kind of system of real-time volume matching failure predication and solution based on orthogonal stereo device of the present invention
Frame diagram, main contents include orthogonal photographic device, field programmable gate array, half global registration algorithm, Gaussian Blur core.
Wherein, the orthogonal photographic device, as shown in figure 3, be sensor camera of the band there are two orthogonal polar curve, by
Different angle obtains two width digital pictures simultaneously, in order to obtain the three-dimensional coordinate that certain in three dimensions is put, needs controlling two
All there are the response point of the point in video camera image planes, so being necessarily mounted on a stabilised platform, half global Stereo matching
Algorithm is calculated and is run, and pass through USB3.0 synchronism outputs directly on two hardware.All four video cameras are all using inline
Function is corrected, and using the extrinsic data obtained in two-dimensional marker initial calibration operational process, uses half global solid
With algorithm within hardware calculated level pair and vertically to Stereo matching, two depth maps of calculating and two confidence level figures lead to
USB3.0 is crossed to be transferred in the computer of connection.The weighted average of each three-dimensional pair of matching score of multiple-camera stereoscopic device
Depth map is merged, the problem of this is the previous work of real-time deep map fusion, this method is along straight line
It is very high with score, therefore matched score provides a unsuitable weighting for depth map fusion.Above-mentioned adaptation function according to
Failure scoring selects a source images in normal operating, and for each pixel i, j is in normal operation, if however, depth map all exists
In the range of mutual 10%, invention uses their matching score mi,jTheir weighted average is calculated to merge these pixels,
This provides a firm cross-check for estimation of Depth, also allows the pixel for receiving not having by the SGM balloon scores distributed,
The step for significantly improve recall rate, in addition the present invention uses additional video camera, improves single Baseline Stereo Matching system
Accuracy and recall rate perform all calculating on same hardware, only increase by two models of cost and weight, and standard is stood
Inexpensive imaging sensor on body magnetic head can successfully avoid mistake measurement and three-dimensional associated confidence in the picture
Low area increases recall rate, otherwise can not obtain reliable measurement result.
Further, the field programmable gate array, in order to understand the matched basic constraint of real-time volume on hardware,
The present invention introduces the operation field programmable gate array of Stereo matching on hardware, and FPGA does not store video camera before treatment
Whole frame, it only buffers seldom image sensor scan line and Continuous plus new line enters the depth map of system, this permits
A chip allowable rapidly processes up to 10 video cameras, but it has added the access of image and storage mode some to limit.
Not all Stereo Matching Algorithm is suitable for such operation, but can use half global registration algorithm, by
The depth map optimized in multiple directions, it improves Block- matching, wherein a paths are parallel to polar curve direction.
Wherein, it is described to be constrained in substantially this refers to texturing element, imaging mode and field in stereo visual system
Geometrical relationship in scape can form the constraints of Stereo matching, they improve matching rate to reducing the calculation amount searched for
There is very big help, polar curve condition is the primary condition that geometry imaging should follow, can be the search range of Stereo matching
Dropped to from two dimension it is one-dimensional, if there are one texture along to polar curve be repeat, it meet texture and matching score criteria, still
Matching score can be very high, shows that this essential attribute is inevitable, and texture scene is very common in city, however,
It has enough characteristics to lead to Stereo matching dependent failure, this problem has obtained good research in the structure of motion field,
And it is often implicitly solved in photo or so consistency check, however, because focus on the application of the present invention is machine vision
Field, therefore the present invention proposes a kind of new, more effective way using the attribute of real-time volume algorithm, unmanned plane and other
The autonomous driving vehicle or system for possessing single Stereo matching pair all constrain epipolar geom etry by Stereo matching.
Further, the half global registration algorithm, the present invention determine that polar curve is several using the matching score that SGM is provided
What possibility mistake, when it encounters the line parallel with polar curve, matching score can reach peak value：Between alternative route of the present invention
Score is matched, to determine that estimation is likely to occur the region of failure.SGM algorithms carry out operation to the gray value of image, not in image
Big gradient on propagate parallax value.However, do not include any vertical gradient along the false disparity correspondence in EP point direction, and
SGM algorithms propagate error hiding along the direction of EP point, and failure is detected using this specific properties of its smoothing step.
Further, the Gaussian Blur core, qualitative evaluation show above-mentioned SGM matchings score in all scenes all
It is best predictive factor, when inputting depth map by medium filtering, the present invention uses 13 × 13 Gaussian Blur cores of 3 δ
Adjust the scale space of matching score image, prediction most probable abort situation in next step utilizes previously discussed half global
It algorithm attribute rather than is optimized in strong gradient, result is also feasible in image score, by finding these scores
The gradient magnitude of difference, can reliably predict the cable architecture for being parallel to polar curve.
In order to reach this purpose, the present invention is by integral image and one-dimensional Sobel cores in vertical direction with having 7
The EP point of the Gauss of pixel region and 0.3 δ do convolution, the significant changes of matching score along EP point be to mistake measure can
By prediction.
Fig. 2 is a kind of solid of real-time volume matching failure predication and solution based on orthogonal stereo device of the present invention
Match the close-up shot to break down.When line appears in the position in figure, which just will appear, but the area
The matching score in domain is very high, there is very high confidence level.
The present invention uses additional video camera, the accuracy and recall rate of single Baseline Stereo Matching system is improved, similary
Hardware on perform all calculating, only increase by two models of cost and weight, the inexpensive image on the three-dimensional magnetic head of standard
Sensor can successfully avoid the area that mistake measures and three-dimensional associated confidence in the picture is low and increase to recall in this way
Rate.
Fig. 3 is a kind of algorithm of real-time volume matching failure predication and solution based on orthogonal stereo device of the present invention
Flow chart, when matching in two directions, the overlapping region between both direction is the dot region in a figure, to this
Region carries out left and right consistency check.Using half global Stereo Matching Algorithm within hardware calculated level pair and vertically to solid
Matching, two depth maps of calculating and two confidence level figures are transferred to by USB3.0 in computer, multiple-camera solid dress
It puts and merges depth map with the weighted average of each three-dimensional pair of matching score, which uses three-dimensional depth and confidence
Degree generates a depth map output as input.
It finally obtained polar curve failure and assume image, the present invention calculates such prognostic chart picture to each depth map,
Create ultimate depth figure and final matching fraction figure, the final matching fraction figure c of each location of pixelsi,jIt is that the half matching overall situation is calculated
The matching score m of the image a or b of methodi,jWeighted average, score f from relatively low failurei,jImage in selected
It selects, which is a united cost function：
ci,j=mi,j+s·fi,j (1)
Empirical value s=2 provides a stable tradeoff from the matching score and failure predication of half matching Global Algorithm, with this
It is worth all experiments that fixed parameter performs, the region that obtained image becomes clear represents the three-dimensional relevant area for being likely to occur failure
Domain, darker regions have no geometrical defect and texture very high confidence level.
Claims (6)
1. a kind of real-time volume matching failure predication and solution based on orthogonal stereo device, which is characterized in that including with
Lower step：
S1, orthogonal photographic device；
S2, field programmable gate array；
S3, half global registration algorithm；
S4, Gaussian Blur core；
Single field-programmable is connected to by the low-voltage differential signal interface of four imaging sensors of orthogonal photographic device to patrol
Sequence is collected, FPGA resynchronizes four independent data flows and carries out biased operation to image, and image is passed with half global registration algorithm
The depth map of sensor scan line and Continuous plus new line optimizes, into row buffering finally using texture in a plurality of directions
Red indicator and Stereo Matching Algorithm avoid the failure of weak texture region from occurring.
2. real-time volume matching failure predication and solution based on orthogonal stereo device as described in claim 1, special
Sign is that S1, orthogonal photographic device include the following steps：
S11, sensor camera of the band there are two orthogonal polar curve is fixed on stabilised platform, is obtained simultaneously by different angle
Two width digital pictures, all there are the response points of the point in the two video camera image planes in left and right, can obtain what certain in three dimensions was put
Three-dimensional coordinate；
S12, half global Stereo Matching Algorithm is directly calculated and run on two hardware, and exported by interface synchronization, four
Video camera is all corrected using Inline Function；
S13, using the extrinsic data obtained in two-dimensional marker initial calibration operational process, multiple-camera stereoscopic device is with each
The weighted average of three-dimensional pair of matching score merges depth map；
Adaptation function scores according to failure and in normal operating selects a source images as each pixel i, and j is in normal operation, so
And if depth map all in the range of mutual 10%, uses its matching score mi,jTheir weighted average is calculated to melt
These pixels are closed, this provides a firm cross-check for estimation of Depth, also allows to receive do not have the height distributed by SGM
The pixel of score；
S14, two other video cameras, the accuracy and recall rate of the single Baseline Stereo Matching system of raising, same are used
All calculating is performed on hardware, only increases by two models of cost and weight, a low cost is added on the three-dimensional magnetic head of standard
Imaging sensor, avoiding the area that mistake measures and three-dimensional associated confidence in the picture is low increases recall rate, obtains
Reliable measurement result.
3. real-time volume matching failure predication and solution based on orthogonal stereo device as described in claim 1, special
Sign is that S2, field programmable gate array include the following steps：
With half global registration algorithm in S21, field programmable gate array, by the depth map optimized in a plurality of directions,
Depth map is merged using united cost function, improves Block- matching, wherein a paths will be parallel to polar curve direction；
If S22, meeting basic constraint, i.e., if there are one texture along being to repeat to polar curve, it meets texture and matching
Score criteria has enough characteristics that Stereo matching dependent failure is caused to be handled in the structure of motion field it is necessary to use, and is shining
It is implicitly solved in piece or so consistency check；
S23, system calculate a prognostic chart picture to each depth map, create ultimate depth figure and final matching fraction figure, often
The final matching fraction figure c of a location of pixelsi,jIt is the matching score m of the image a or b of half matching Global Algorithmi,jWeighting put down
Mean value, from relatively low failure scoring fi,jImage in selected, which is a united value letter
Number：
ci,j=mi,j+s·fi,j (1)
Its matching score and failure predication from half matching Global Algorithm provides a stable tradeoff, with the fixed parameter of the value
All experiments performed, the region that obtained image becomes clear represent the three-dimensional relevant region for being likely to occur failure.
4. real-time volume matching failure predication and solution based on orthogonal stereo device as claimed in claim 3, special
Sign is, in S23, s=2.
5. real-time volume matching failure predication and solution based on orthogonal stereo device as described in claim 1, special
Sign is that S3, half global registration algorithm include the following steps：
S31, the possibility mistake of epipolar geom etry is determined using the matching score of SGM offers, when it encounters the line parallel with polar curve
When, matching score can reach peak value, the matching score between alternative route, to determine that estimation is likely to occur the region of failure；
S32, operation is carried out to the gray value of image using SGM algorithms, does not propagate parallax value in the big gradient of image, utilize it
This specific properties of smoothing step detect failure.
6. real-time volume matching failure predication and solution based on orthogonal stereo device as described in claim 1, special
Sign is that S4, Gaussian Blur core include the following steps：
S41, SGM matches score as predictive factor using in S3；
S42, when input depth map by medium filtering when, adjust matching score chart using 13 × 13 Gaussian Blur cores of 3 δ
The scale space of picture, the most probable abort situation of prediction next step；
S43, using half Global Algorithm attribute previously discussed, by finding the gradient magnitude of these score differences, prediction is parallel to
The cable architecture of polar curve；
S44, by integral image and one-dimensional Sobel cores in vertical direction with the Gauss's with 7 pixel regions and 0.3 δ
EP point does convolution, and the change of matching score along EP point is turned to the reliable prediction value measured mistake.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810045950.4A CN108257170A (en) | 2018-01-17 | 2018-01-17 | Real-time volume matching failure predication and solution based on orthogonal stereo device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810045950.4A CN108257170A (en) | 2018-01-17 | 2018-01-17 | Real-time volume matching failure predication and solution based on orthogonal stereo device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108257170A true CN108257170A (en) | 2018-07-06 |
Family
ID=62740909
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810045950.4A Pending CN108257170A (en) | 2018-01-17 | 2018-01-17 | Real-time volume matching failure predication and solution based on orthogonal stereo device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108257170A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115641382A (en) * | 2022-10-21 | 2023-01-24 | 哈尔滨工业大学 | External parameter calibration method for orthogonal stereoscopic vision structure |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008245004A (en) * | 2007-03-28 | 2008-10-09 | Meidensha Corp | Monitoring control system |
| CN104639933A (en) * | 2015-01-07 | 2015-05-20 | 前海艾道隆科技（深圳）有限公司 | Real-time acquisition method and real-time acquisition system for depth maps of three-dimensional views |
-
2018
- 2018-01-17 CN CN201810045950.4A patent/CN108257170A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008245004A (en) * | 2007-03-28 | 2008-10-09 | Meidensha Corp | Monitoring control system |
| CN104639933A (en) * | 2015-01-07 | 2015-05-20 | 前海艾道隆科技（深圳）有限公司 | Real-time acquisition method and real-time acquisition system for depth maps of three-dimensional views |
Non-Patent Citations (1)
| Title |
|---|
| LORENZ MEIER 等: "Real-time Stereo Matching Failure Prediction and Resolution using Orthogonal Stereo Setups", 《IEEE》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115641382A (en) * | 2022-10-21 | 2023-01-24 | 哈尔滨工业大学 | External parameter calibration method for orthogonal stereoscopic vision structure |
| CN115641382B (en) * | 2022-10-21 | 2023-09-08 | 哈尔滨工业大学 | External parameter calibration method for orthogonal stereoscopic vision structure |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110322702B (en) | Intelligent vehicle speed measuring method based on binocular stereo vision system | |
| EP3510561B1 (en) | Predicting depth from image data using a statistical model | |
| CN104700414B (en) | A kind of road ahead pedestrian's fast ranging method based on vehicle-mounted binocular camera | |
| CN106767399B (en) | The non-contact measurement method of logistics goods volume based on binocular stereo vision and dot laser ranging | |
| CN107038723B (en) | Method and system for estimating rod-shaped pixels | |
| EP2597597B1 (en) | Apparatus and method for calculating three dimensional (3D) positions of feature points | |
| US8831336B2 (en) | Method, system and computer program product for detecting an object in response to depth information | |
| CN102510506B (en) | Virtual and real occlusion handling method based on binocular image and range information | |
| CN105913410A (en) | Long-distance moving object height measurement apparatus and method based on machine vision | |
| CN102831601A (en) | Three-dimensional matching method based on union similarity measure and self-adaptive support weighting | |
| US20220051425A1 (en) | Scale-aware monocular localization and mapping | |
| CN109263637A (en) | A kind of method and device of prediction of collision | |
| WO2024114119A1 (en) | Sensor fusion method based on binocular camera guidance | |
| CN107710091A (en) | For the system and method for the operator scheme for selecting mobile platform | |
| CN111681275A (en) | Double-feature-fused semi-global stereo matching method | |
| CN104471436B (en) | The method and apparatus of the variation of imaging scale for computing object | |
| Wu et al. | Geometry-aware instance segmentation with disparity maps | |
| CN110345924A (en) | A kind of method and apparatus that distance obtains | |
| KR20100066952A (en) | Apparatus for tracking obstacle using stereo vision and method thereof | |
| CN205726180U (en) | Terminal guidance video image three dimensional data collection system | |
| Omar et al. | Detection and localization of traffic lights using YOLOv3 and Stereo Vision | |
| CN108257170A (en) | Real-time volume matching failure predication and solution based on orthogonal stereo device | |
| CN105138979A (en) | Method for detecting the head of moving human body based on stereo visual sense | |
| CN112561979A (en) | Self-supervision monocular depth estimation method based on deep learning | |
| CN108090930A (en) | Barrier vision detection system and method based on binocular solid camera |
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: 20180706 |
|
| RJ01 | Rejection of invention patent application after publication |