CN106952304B - A kind of depth image calculation method using video sequence interframe correlation - Google Patents
A kind of depth image calculation method using video sequence interframe correlation Download PDFInfo
- Publication number
- CN106952304B CN106952304B CN201710172103.XA CN201710172103A CN106952304B CN 106952304 B CN106952304 B CN 106952304B CN 201710172103 A CN201710172103 A CN 201710172103A CN 106952304 B CN106952304 B CN 106952304B
- Authority
- CN
- China
- Prior art keywords
- point
- parallax
- frame
- value
- search range
- 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
Classifications
-
- 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
- G06T2207/10016—Video; Image sequence
Landscapes
- Measurement Of Optical Distance (AREA)
Abstract
The present invention discloses a kind of depth image calculation method using video sequence interframe correlation, comprising: step 1 is corrected the picture of left and right view camera shooting；Step 2 carries out parallax optimization, parallax refinement, obtains initial parallax value d1 for the point p of the left and right picture of first frame in maximum disparity search range calculating matching cost；The gradient of disparity of first frame is g1；Step 3, for next frame p point calculating, when p point is not abnormal point or marginal point, disparity search range is set on the basis of the parallax value of previous frame p point, otherwise matching cost is calculated with maximum disparity search range, carry out parallax optimization, parallax refinement, and the parallax value and gradient of disparity of present frame are obtained, realize real-time Stereo matching.The present invention substantially reduces the calculation amount of present frame algorithm using the parallax of previous frame in video sequence, can reduce time-consuming, efficiency of the raising in practical application scene under conditions of guaranteeing original algorithmic match precision.
Description
Technical field
The present invention relates to computer stereo vision fields, more particularly to a kind of utilize video sequence interframe correlation
Depth image calculation method.
Background technique
Stereoscopic vision is one of widest research topic in computer vision field in recent years.It is from Same Scene not
With the technology for obtaining depth information of scene in the picture of viewpoint shooting.The picture shot from two cameras of different points of view is obtained,
The horizontal displacement of the corresponding points o'clock on two width pictures in scene is calculated for this two width picture, this process is thus referred to as double
Mesh Stereo matching, horizontal displacement are thus referred to as parallax.Stereoscopic vision is widely used in many fields, such as obstacle quality testing
It surveys, drives auxiliary, three-dimensional reconstruction and motion detection etc..
With the development of consumer electronics, the scene of stereoscopic vision application is more and more extensive.As shown in Figure 1, in general, greatly
Most Stereo Matching Algorithms may be summarized to be four steps:
(1) matching cost calculates
(2) cost polymerize
(3) parallax optimization/calculating
(4) parallax refinement (post-processing)
And the implementation detail of specific steps depends on algorithm itself.According to steps 1 and 2,3 difference, most of algorithms are divided into
Two kinds of main Types, local algorithm and Global Algorithm.Local algorithm is based primarily upon window and carries out cost polymerization (step 2), for
One given point, parallax optimize the energy value being calculated in (step 3) and are dependent only on owning in a restriction window
The pixel value of point.Local algorithm can usually be divided into steps 1 and 2,3 with being apparent.Algorithm (the sum- for example, traditional difference of two squares is summed
Of-squared-differences), steps are as follows for its calculating:
(1) matching cost be in the case where giving some parallax, calculate left side viewpoint figure and the right viewpoint figure in it is corresponding
The difference of two squares of two points.
(2) cost polymerization is exactly the matching cost summation to all the points in window are limited.
(3) parallax optimization is then that the polymerization the smallest parallax value of cost is picked out to each point.
In addition to this, there are also absolute value summation algorithm (sum-of-absolute- for classical local algorithm
Differences): matching cost is the absolute value for calculating corresponding points；Adaptive weighting algorithm (adaptive-support-
Window): according to the different weights of point and the color of central point in window, distance impartings, then general i.e. in cost polymerization process
Its cost is added；Normalized crosscorrelation algorithm (Normalized Cross Correlation): the cross-correlation between image is calculated
Value.Census transformation: the relative value of pixel grey scale size in window is converted to bit number, then compares the ratio of left and right corresponding points
The Hamming distance of special number.
Local algorithm has high speed and low energy consumption, but the matching precision that local algorithm obtains is limited, it is more difficult to meet multiple
The demand of miscellaneous scene.Compared with local algorithm, Global Algorithm has better matching precision.Global Algorithm makes smooth item
It is assumed that and stereo matching problem is become an energy-optimised problem.Most of Global Algorithm has skipped cost polymerization (step
2), consider matching cost and smooth item, propose energy function for global point, then minimize energy function, depending on
Difference.And the optimization algorithm minimized to energy function mainly has figure to cut (graph cuts), confidence spread (Belief
Propagation) etc..But compared to local algorithm, the calculation amount of Global Algorithm is bigger, and energy consumption is higher.
In practical applications, for especially for mobile device, the speed and precision of Stereo Matching Algorithm how is balanced
It is exactly one of research emphasis, that is, the higher Real Time Matching Algorithm of precision how to be realized in limited resource.Numerous
Among algorithm, half global registration (Semi-Global Match) be exactly after balancing matching precision and operation time performance compared with
A kind of good algorithm, is widely used in a variety of applications.Half global registration algorithm proposes a global energy function, calculates with the overall situation
Method optimizes difference to global point, and the energy function of each point is divided into the path in 16 directions 8or by it, it is only necessary to right
Then the value addition in all paths is exactly the energy of the point by each path evaluation, and the evaluation of single-pathway can be adopted
It is solved with Dynamic Programming.With the development of hardware, more and more hardware platforms (GPU, FPGA etc.) are utilized to realize three-dimensional
Matching, especially half global registration, to achieve the purpose that real time execution.However, current algorithm is mostly just for static list
Frame image is handled, it is not intended that the Stereo matching under video sequence.There is no relevant work sutdies at present, and realize essence
It is still expensive that true real-time volume matches the cost for mobile platform.
Summary of the invention
For above-mentioned the problems of the prior art, the invention discloses a kind of depth using video sequence interframe correlation
Image calculation method, amount when reducing the calculating of Stereo matching solve one or more above problems or disadvantage in whole or in part.
The present invention is achieved through the following technical solutions, the present invention the following steps are included:
Step 1, left and right view camera shoot Same Scene, are corrected to the picture of shooting, after being corrected
Left and right different points of view identical size video sequence frame.
Step 2, for first frame left and right picture point p, calculate matching cost in maximum disparity search range, according to
The disparity search range and cost value of p point carry out parallax optimization, according to obtained in the disparity search range of p point and parallax optimization
The energy value of p point carries out parallax refinement；Obtain initial parallax value d1；The gradient of disparity of first frame is g1.
Step 3, the calculating for carrying out next frame, for p point: when p point is not abnormal point or marginal point, with previous frame p point
Disparity search range is set on the basis of parallax value, calculates matching cost；Otherwise, in matching generation, is calculated with maximum disparity search range
Valence；Parallax optimization is carried out according to the disparity search range of p point and cost value, is optimized according to the disparity search range of p point and parallax
Obtained in p point energy value carry out parallax refinement；Obtain the parallax value and gradient of disparity of present frame；
Step 4 obtains video sequence depth image, realizes real-time Stereo matching.
Further, the search range for retaining the energy value of each point in calculating process, and recording the point is subsequent
The optimization of step parallax and parallax refinement are prepared.
Step 2 further comprises following technical characteristic: enabling disparity search range is (0, D), and wherein D is that disparity search is maximum
Value；For the left and right picture of first frame, half global registration algorithm is taken to calculate initial parallax value, and retained each in calculating process
The energy value Energy [0] of a point~Energy [D-1]；Wherein using census transformation calculations matching cost cost [0]~
Cost [D-1], carries out energy function accumulative optimization in 8 directions, and the original disparity map obtained for optimization is regarded
Difference refinement.
Further, parallax refinement (post-processing) process includes: sub-pix refinement, and singular values standard form rationalizes parallax etc.
Means.
Step 3 further comprises following technical characteristic: the calculating of next frame i+1 frame is carried out, when the cost value for calculating point p
When, find the parallax value di of previous frame the i-th frame p point, the disparity search range of the point p of i+1 frame on the basis of di, from (0,
D it) is reduced to (di-range, di+range), range is the search range of setting.
Given threshold t1, t2 seek the gradient map Gi of previous frame the i-th frame parallax, judge two conditions respectively:
(1) in cost calculating process, cost value minimum value min (cost) for calculating in (di-range, di+range) >
t1；
(2) gradient gi > t2 of the point p in Gi；
The point for meeting condition (1) is abnormal point, and the point for meeting condition (2) is that marginal point expands as long as meeting one of them
Big current search range is (0, D), after the search range that return continues to calculate uncalculated cost value, and record the point is
Continuous step parallax optimization and parallax refinement are prepared.
Further, it when carrying out the calculating of i+1 frame, for the point within the scope of disparity search, calculates to correspond to and be somebody's turn to do
Energy value in point range；Energy value for the point except disparity search range, directly outside the i-th frame of succession point range.
Further, in parallax thinning process, the thin of parallax only is carried out to the disparity search range determined in step 3
Change, determines optimal parallax；It is found within the scope of point p disparity search and enables the smallest parallax value of energy function, which is point p's
Parallax.
The invention has the following advantages: being retained by the obtained parallax of previous frame image to working as in 1, application in real time
Previous frame.By retrieving the information of previous frame parallax, the search range of present frame parallax is reduced, is in particular in matching cost meter
It calculates, parallax optimization and refinement aspect.The algorithm improves entire algorithm and exists under conditions of guaranteeing original algorithmic match precision
Efficiency in practical application scene.2, the marginal point to error increase, abnormal point, corresponding expansion disparity search range are reduced
The error that may cause.3, retain the energy value of each point in calculating process, the view of each point in calculation method of the invention
Poor search range is that the optimization of subsequent step parallax and parallax refinement are prepared, and is reduced largely using video sequence interframe correlation
Calculation amount.Therefore the calculating time of Stereo matching can be greatly decreased in whole process of the invention, while having ensured the essence of algorithm
Degree has very big application prospect on real-time platform.
Detailed description of the invention
Fig. 1 is the matched structure of depth image based on single-frame images.
Fig. 2 is the structure of the depth image calculation method of the embodiment of the present invention.
Fig. 3 is the algorithm flow of the start frame of the embodiment of the present invention.
Fig. 4 is the algorithm flow of the subsequent frame of the embodiment of the present invention.
Specific embodiment
For the ease of the understanding of those skilled in the art, the present invention is made further below with reference to embodiment and attached drawing
It is bright.
When realizing Stereo matching in real time, acquisition is all a series of continuous picture frames, and is had perhaps between frame and frame
More correlations, these correlations can be used to reduce the calculation amount of entire algorithm.As shown in Fig. 2, the processing of the present embodiment
Step includes:
1, the camera of left and right viewpoint shoots Same Scene, the picture of shooting is corrected, after being corrected
Left and right different points of view identical size picture.
2, might as well enable disparity search range is (0, D), and wherein D is disparity search maximum value.For the left and right figure of first frame
Piece takes unmodified half global registration algorithm to calculate initial parallax value, and retains the energy value of each point in calculating process
Energy [0]~Energy [D-1].Census transformation calculations matching cost cost [0]~cost [D-1] is wherein used, for
Energy function carries out accumulative optimization in 8 directions, optimizes to obtain energy value Energy [0]~Energy using energy function
[D-1], the original disparity map obtained for optimization carry out parallax refinement.Parallax refinement (post-processing) process specifically includes that sub- picture
Plain refinement, singular values standard form rationalize the means such as parallax.For modified hydrothermal process, parallax thinning process is constant.Details please see Figure
3。
3, the calculating for carrying out next frame finds the parallax value d1 of previous frame p point when calculating the cost value of point p, then
The disparity search range of the point p of present frame is reduced to (d1-range, d1+range), range from (0, D) then with d1 benchmark
For the disparity search range of setting, 2~3 are generally taken.Dimension of picture is bigger, and the value of range is bigger.Matching cost calculates
(referred to herein as census transformation) calculates the cost value cost [d1-range] obtained in (d1-range, d1+range) range
~cost [d1+range].
4, reducing matching range reduces the matched time, but for the parallax value of edge, error be will increase.Mainly exist
In:
(1), algorithm itself has error, and parallax of previous frame itself has error, and the disparity search range of present frame is base
It searches on the basis of the parallax value of previous frame, then part parallax can scan on the basis of the above frame error point, causes
As a result it is not inconsistent with truth.And the calculating of present frame can also generate new error.
(2), for the edge of object, the parallax value difference on edge both sides is larger.So for previous frame position in p point
For marginal point, it is moved at p+n in present frame marginal point, and the parallax of the still above frame border point point p of the point p of present frame
On the basis of search for parallax value, will result in error.
For error, given threshold t1, t2 seek the gradient map G of previous frame parallax, judge two conditions respectively
(1) in cost calculating process, cost value minimum value min (cost) for calculating in (d1-range, d1+range) >
t1。
(2) gradient g1 > t2 of the point p in G.
Meet condition (1) thinks that the point is abnormal point, and meet condition (2) thinks that the point is marginal point, as long as meeting
One of them, then expanding current search range is (0, D), and return continues to calculate uncalculated cost value.
If p point meets (1) or (2), the cost value not calculated is calculated according to widened disparity search range (0, D)
[d1-range] &cost [d1+range]~cost [D-1] thus obtains p point in disparity search range to cost [0]~cost
The cost value cost [0] of (0, D)~cost [D-1], and the disparity search range of the point is updated for the optimization of subsequent step parallax
It prepares with parallax refinement.If being unsatisfactory for decision condition (1) and (2), it is considered as qualification, is directly entered next step, parallax
Optimization.
5, it is exactly parallax optimization that cost calculates later.If the search range of fruit dot p is (d1-range, d1+range), then
Calculate and update energy value Energy [d1-range-1]~Energy [d1+range-1], as fruit dot p search range be (0,
D), calculate and update energy value Energy [0]~Energy [D-1].Value not in range remains unchanged.Then it is searched in point p
It is found within the scope of rope and enables the smallest parallax value of energy function, which is the parallax of point p.Details see Fig. 4.
6, in parallax refinement last handling process, the refinement of parallax is only carried out to the search range that abovementioned steps determine, really
Fixed optimal parallax.
In order to illustrate the improvement of whole service time.Assuming that sharing N number of point, disparity search range is (0, D), then in cost
Accumulative, parallax optimizes, the former algorithm of parallax refinement aspect needs 3DN related operation altogether, and the algorithm after optimizing, and does not consider to calculate
If the gradient calculating for measuring very little, it is assumed that the ratio for needing to retain search range D is n, and the range after diminution is 2range+1, no
Consider the judgement in calculating process, then needs 3* (2range+1) * (1-n) N+3*n*N*D related operation of *.For 640*
The case where 480, D=64, range=2, n=0.1, then the calculation amount of the latter accounts for the former 44%, it is contemplated that gradient calculates, meter
Calculation amount is 50% or so.
So the calculating time of Stereo matching can be greatly decreased in whole process, while the precision of algorithm is ensured, real-time
There is very big application prospect on platform.
Above embodiment is merely illustrative of the invention's technical idea, and this does not limit the scope of protection of the present invention, all
It is any changes made on the basis of the technical scheme according to the technical idea provided by the invention, each falls within present invention protection model
Within enclosing.The technology that the present invention is not directed to can be realized by existing technology.
Claims (7)
1. a kind of depth image calculation method using video sequence interframe correlation, which comprises the following steps:
Step 1, left and right view camera shoot Same Scene, are corrected to the picture of shooting, the left side after being corrected
The identical size video sequence frame of right different points of view；
Step 2 calculates matching cost, according to p point in maximum disparity search range for the point p of the left and right picture of first frame
Disparity search range and cost value carry out parallax optimization, according to p point obtained in the disparity search range of p point and parallax optimization
Energy value carries out parallax refinement；Obtain initial parallax value d1；The gradient of disparity of first frame is g1；
Step 3, the calculating for carrying out next frame, for p point: when p point is not abnormal point or marginal point, with the parallax of previous frame p point
Disparity search range is set on the basis of value, calculates matching cost；Otherwise, matching cost is calculated with maximum disparity search range；Root
Parallax optimization is carried out according to the disparity search range and cost value of p point, is obtained according in the disparity search range of p point and parallax optimization
P point energy value carry out parallax refinement, obtain the parallax value and gradient of disparity of present frame；
Step 4 obtains video sequence depth image.
2. the depth image calculation method according to claim 1 using video sequence interframe correlation, it is characterised in that:
Retain the energy value of each point in calculating process；The disparity search range for recording each point is that subsequent step parallax optimizes and regards
Difference refinement is prepared.
3. the depth image calculation method according to claim 1 using video sequence interframe correlation, which is characterized in that
Step 2 further comprises following technical characteristic:
Enabling disparity search range is (0, D), and wherein D is disparity search maximum value；For the left and right picture of first frame, use
Census transformation calculations matching cost cost [0] ~ cost [D-1], parallax optimization in energy function is carried out in 8 directions
Accumulative optimization, obtains energy value Energy [0] ~ Energy [D-1], and the original disparity map progress parallax obtained for optimization is thin
Change.
4. the depth image calculation method according to claim 1 using video sequence interframe correlation, which is characterized in that
The calculating matching cost of step 3 further comprises following technical characteristic: the calculating of next frame i+1 frame is carried out, when calculating point p's
When cost value, find the parallax value di of previous frame the i-th frame p point, the disparity search range of the point p of i+1 frame on the basis of di,
It is reduced to (di-range, di+range) from (0, D), range is the search value of setting；
Given threshold t1, t2 seek the gradient map Gi of previous frame the i-th frame parallax, judge two conditions respectively:
(1) cost value minimum value min (cost) > t1 in matching cost calculating process, in (di-range, di+range)；
(2) gradient gi > t2 of the point p in Gi；
The point for meeting condition (1) is abnormal point, and the point for meeting condition (2) is marginal point, as long as meeting one of them, expansion is worked as
Preceding search range is (0, D), and return continues to calculate uncalculated cost value.
5. the depth image calculation method according to claim 4 using video sequence interframe correlation, it is characterised in that:
When carrying out the calculating of i+1 frame, for the point within the scope of disparity search, the energy value corresponded in the point range is calculated；
Energy value for the point except disparity search range, directly outside the i-th frame of succession point range.
6. the depth image calculation method according to claim 4 using video sequence interframe correlation, it is characterised in that:
In parallax thinning process, the refinement of parallax is only carried out to the disparity search range determined in step 3, determines optimal parallax；?
It is found within the scope of point p disparity search and enables the smallest parallax value of energy function, which is the parallax of point p.
7. according to claim 1 to the depth image calculation method using video sequence interframe correlation any in 6,
Be characterized in that: parallax thinning process includes sub-pix refinement, and singular values standard form rationalizes parallax.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710172103.XA CN106952304B (en) | 2017-03-22 | 2017-03-22 | A kind of depth image calculation method using video sequence interframe correlation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710172103.XA CN106952304B (en) | 2017-03-22 | 2017-03-22 | A kind of depth image calculation method using video sequence interframe correlation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106952304A CN106952304A (en) | 2017-07-14 |
| CN106952304B true CN106952304B (en) | 2019-09-20 |
Family
ID=59472179
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710172103.XA Active CN106952304B (en) | 2017-03-22 | 2017-03-22 | A kind of depth image calculation method using video sequence interframe correlation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106952304B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108876835A (en) * | 2018-03-28 | 2018-11-23 | 北京旷视科技有限公司 | Depth information detection method, device and system and storage medium |
| CN110033483A (en) * | 2019-04-03 | 2019-07-19 | 北京清微智能科技有限公司 | Based on DCNN depth drawing generating method and system |
| CN110148168B (en) * | 2019-05-23 | 2023-03-24 | 南京大学 | Three-eye camera depth image processing method based on size double baselines |
| CN110246169B (en) * | 2019-05-30 | 2021-03-26 | 华中科技大学 | Gradient-based window adaptive stereo matching method and system |
| CN110191330A (en) * | 2019-06-13 | 2019-08-30 | 内蒙古大学 | Depth map FPGA implementation method and system based on binocular vision green crop video flowing |
| CN111553296B (en) * | 2020-04-30 | 2021-08-03 | 中山大学 | Two-value neural network stereo vision matching method based on FPGA |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102026013A (en) * | 2010-12-18 | 2011-04-20 | 浙江大学 | Stereo video matching method based on affine transformation |
| CN103136750A (en) * | 2013-01-30 | 2013-06-05 | 广西工学院 | Stereo matching optimization method of binocular visual system |
| CN103810690A (en) * | 2012-11-07 | 2014-05-21 | 富士通株式会社 | Stereo matching method and device thereof |
| CN104065954A (en) * | 2014-07-03 | 2014-09-24 | 中国传媒大学 | Method for quickly detecting parallax scope of high-definition stereoscopic video |
| CN106228605A (en) * | 2016-07-29 | 2016-12-14 | 东南大学 | A kind of Stereo matching three-dimensional rebuilding method based on dynamic programming |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9544613B2 (en) * | 2013-04-24 | 2017-01-10 | Sony Corporation | Local detection model (LDM) for recursive motion estimation |
-
2017
- 2017-03-22 CN CN201710172103.XA patent/CN106952304B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102026013A (en) * | 2010-12-18 | 2011-04-20 | 浙江大学 | Stereo video matching method based on affine transformation |
| CN103810690A (en) * | 2012-11-07 | 2014-05-21 | 富士通株式会社 | Stereo matching method and device thereof |
| CN103136750A (en) * | 2013-01-30 | 2013-06-05 | 广西工学院 | Stereo matching optimization method of binocular visual system |
| CN104065954A (en) * | 2014-07-03 | 2014-09-24 | 中国传媒大学 | Method for quickly detecting parallax scope of high-definition stereoscopic video |
| CN106228605A (en) * | 2016-07-29 | 2016-12-14 | 东南大学 | A kind of Stereo matching three-dimensional rebuilding method based on dynamic programming |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106952304A (en) | 2017-07-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106952304B (en) | A kind of depth image calculation method using video sequence interframe correlation | |
| CN114782691B (en) | Robot target identification and motion detection method based on deep learning, storage medium and equipment | |
| CN111462207A (en) | RGB-D simultaneous positioning and map creation method integrating direct method and feature method | |
| US11367195B2 (en) | Image segmentation method, image segmentation apparatus, image segmentation device | |
| CN106651897B (en) | Parallax correction method based on super-pixel segmentation | |
| CN110688905A (en) | Three-dimensional object detection and tracking method based on key frame | |
| CN109389156B (en) | Training method and device of image positioning model and image positioning method | |
| US20160117573A1 (en) | Method and apparatus for extracting feature correspondences from multiple images | |
| KR101869605B1 (en) | Three-Dimensional Space Modeling and Data Lightening Method using the Plane Information | |
| US7602966B2 (en) | Image processing method, image processing apparatus, program and recording medium | |
| CN112418288A (en) | GMS and motion detection-based dynamic vision SLAM method | |
| CN109859249B (en) | Scene flow estimation method based on automatic layering in RGBD sequence | |
| US11985421B2 (en) | Device and method for predicted autofocus on an object | |
| Pinies et al. | Dense mono reconstruction: Living with the pain of the plain plane | |
| CN116091574A (en) | 3D target detection method and system based on plane constraint and position constraint | |
| CN113034681B (en) | Three-dimensional reconstruction method and device for spatial plane relation constraint | |
| Zhu et al. | PairCon-SLAM: Distributed, online, and real-time RGBD-SLAM in large scenarios | |
| CN112270748B (en) | Three-dimensional reconstruction method and device based on image | |
| CN111402429B (en) | Scale reduction and three-dimensional reconstruction method, system, storage medium and equipment | |
| CN111882613B (en) | Visual odometer method, device, storage medium and equipment based on edge semantics | |
| CN112233149A (en) | Scene flow determination method and device, storage medium and electronic device | |
| CN111179327A (en) | Depth map calculation method | |
| KR101178015B1 (en) | Generating method for disparity map | |
| KR20170037804A (en) | Robust visual odometry system and method to irregular illumination changes | |
| CN115908485A (en) | Real-time pose tracking method and system for non-cooperative target in space |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |