CN111351449A - Stereo matching method based on cost aggregation - Google Patents
Stereo matching method based on cost aggregation Download PDFInfo
- Publication number
- CN111351449A CN111351449A CN202010093245.9A CN202010093245A CN111351449A CN 111351449 A CN111351449 A CN 111351449A CN 202010093245 A CN202010093245 A CN 202010093245A CN 111351449 A CN111351449 A CN 111351449A
- Authority
- CN
- China
- Prior art keywords
- image
- guide
- stereoscopic vision
- vision module
- cost
- 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
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
- G01B11/25—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures by projecting a pattern, e.g. one or more lines, moiré fringes on the object
- G01B11/254—Projection of a pattern, viewing through a pattern, e.g. moiré
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
- G06T7/85—Stereo camera calibration
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention discloses a cost aggregation-based stereo matching method, which comprises the steps of establishing a measuring system, wherein the measuring system comprises a mounting plate, a stereo vision module and a digital speckle projection module, wherein the stereo vision module and the digital speckle projection module are arranged on the mounting plate in a sliding manner; calibrating and correcting images of two industrial cameras in the stereoscopic vision module; forming digital speckles through a digital speckle projection module and projecting the digital speckles to the surface of a measured object; and shooting an image with digital speckle textures through a stereoscopic vision module, preprocessing the image, and performing stereoscopic matching calculation to obtain three-dimensional data information of the measured object. The invention can adapt to the measured objects with different sizes, improves the measurement precision of the measured object without texture or with weak texture, and is not easily influenced by the ambient illumination, so that the appearance measurement is carried out in the measurement system, and the measurement system has good stability and better adaptability.
Description
Technical Field
The invention relates to the technical field of image data processing, in particular to a stereo matching method based on cost aggregation.
Background
The existing non-contact three-dimensional vision measurement is mainly divided into stereoscopic vision and structured light three-dimensional vision. The linear structured light measuring method is based on the principle of optical triangulation, and mainly comprises an optical projector, a camera and a computer system. The optical projector projects the structured light with a certain mode on the surface of the object to form a light bar three-dimensional image modulated by the surface shape of the object to be measured on the surface. The stereoscopic vision is based on the parallax principle, and three-dimensional information is acquired by the trigonometry principle. The method is characterized in that a plurality of cameras are used for shooting the same object, images of the object to be detected under a plurality of visual angles are obtained, and finally a three-dimensional model of the images is restored through parallax calculation. In the existing stereoscopic vision method, binocular vision and multi-view vision can be divided according to the number of cameras, and the multi-view vision mainly comprises monocular multi-pose stereoscopic vision and multi-view stereoscopic vision.
The traditional binocular stereo vision excessively depends on texture characteristics of a measured object, so that the adaptability to the measured object is poor, and the measurement precision of weak texture or no texture is poor. The device for rapidly measuring the appearance of the binocular stereo vision and the speckle projection has better precision, but the binocular camera and the speckle projector are fixed in position due to small design size of the whole machine, and the device is only suitable for small workpieces. The traditional active structured light has the defects of difficult phase expansion, error diffusion, easy influence of illumination and unsuitability for measurement in environments with different brightness.
Disclosure of Invention
The invention provides a cost polymerization-based stereo matching method, which aims to solve the problems of poor adaptability and low accuracy of the existing stereo matching method due to the difficulty in adapting to measured objects with different sizes, excessive dependence on texture characteristics of the measured objects and easiness in influence of illumination.
In order to achieve the above purpose, the technical means adopted is as follows:
a stereo matching method based on cost aggregation comprises the following steps:
s1, establishing a measuring system, wherein the measuring system comprises a mounting plate, a stereoscopic vision module and a digital speckle projection module, and the stereoscopic vision module and the digital speckle projection module are arranged on the mounting plate in a sliding manner; wherein the stereoscopic vision module comprises two industrial cameras that are relatively parallel in physical location;
s2, calibrating and correcting images of two industrial cameras in the stereoscopic vision module;
s3, forming digital speckles through a digital speckle projection module and projecting the digital speckles to the surface of a measured object;
and S4, shooting the image with the digital speckle texture through a stereoscopic vision module, preprocessing the image, and performing stereoscopic matching calculation to obtain three-dimensional data information of the measured object.
In the scheme, the three-dimensional information is acquired by the measuring system combining the stereoscopic vision module and the digital speckle projection module, and the measuring system can adjust the measuring pose in a certain measuring range so as to adapt to measured objects with different sizes. The method has the advantages that the appropriate industrial camera and relevant parameters of the digital speckles are selected according to different measured objects, so that the measurement accuracy of the measured objects without textures or weak textures is improved, and meanwhile, the measured objects are not easily influenced by ambient light, so that the measurement system is good in shape measurement, good in stability and more adaptive.
Preferably, the step S1 further includes the steps of: and the stereoscopic vision module and the digital speckle projection module are subjected to model selection and position adjustment according to the measured object and different measurement precision requirements, and then the positions of the stereoscopic vision module and the digital speckle projection module are relatively fixed. In this preferred scheme, in the lectotype process, can select different cameras to different testees and different measurement accuracy requirements, stereo vision module and digital speckle projection module can slide on the mounting panel simultaneously to have better adaptability to different testees. After the positions of the stereo vision module and the digital speckle projection module are relatively fixed, three-dimensional measurement is performed.
Preferably, the preprocessing of step S4 includes: and correcting the left and right images with the digital speckle textures shot by the stereoscopic vision module according to the calibration parameters obtained by the calibration of the industrial camera.
Preferably, the performing stereo matching calculation in step S4 specifically includes the following steps:
s41, performing matching cost calculation on the left image and the right image preprocessed by the stereoscopic vision module through an ADcensus stereoscopic matching algorithm;
s42, performing cost aggregation calculation on the matching cost obtained by the matching cost calculation by using secondary guide filtering based on LOG operator weighting;
s43, extracting initial parallax for the matching cost after cost aggregation calculation by using an algorithm with a winner as a king;
and S44, carrying out optimization processing on the initial parallax by adopting rapid weighted median filtering to obtain three-dimensional data information of the measured object.
In the preferred scheme, noise suppression is performed on noise introduced during digital speckle projection through a cost aggregation algorithm based on secondary guided filtering weighted by an LOG operator, so that matching precision is higher, and mismatching rate is reduced.
Preferably, the specific step of step S42 includes:
s421, constructing a first guide filter and a second guide filter, wherein the first guide filter and the second guide filter are two same guide filters weighted by an LOG operator;
s422, filtering the input image p by using the first guide filter, wherein the guide image of the first guide filter adopts a guide image I or an input image p preset according to specific application to obtain an output image q with a maintained edge; the input image p is a matching cost obtained by calculating the matching cost of the left image and the right image after the pretreatment of the stereoscopic vision module through an ADcensus stereoscopic matching algorithm;
and S423, taking the output image q as a guide image of the second guide filter, filtering the input image p by using the second guide filter, and outputting an image q1 subjected to second guide filtering.
Preferably, the guiding filter weighted by the LOG operator in step S421 specifically includes:
constructing a guide filtering model:
in the formula wkA square window with a radius r and a pixel k as a center;akand bkIs the invariant coefficient of the linear function when the window center is located at time k; i is a guide image, and q is an output image; i is a domain point with the point k as a central window;k being arbitrary and all of wk;
Constructing a function model of a guide filtering algorithm:
in the formula: e is the function model energy; epsilon is a penalty parameter, preventing akThe value is too large; p is a radical of%I is the pixel value of the input image p; wherein a iskAnd bkThe value of (d) is found using least squares:
in the formula: mu.skAndpartial windows w, respectively, for the guide image IkMean and variance in (a); is window wkThe number of pixels in (1);for the image p to be filtered in the window wkThe average value of (1);
the weighted filtering of the LOG operator is used, so that the penalty parameter is adaptively adjusted, different textures of a measured object are considered, wherein the expression of the LOG operator is as follows:
in the formula: x and y are respectively the row and column coordinates of the pixel points; delta is the standard deviation; the expression of LOG operator weight term t(s) is:
in the formula: s is a central pixel in the local window; sg all pixels in the local window; | LOG (S) | represents the magnitude of the Gaussian Laplacian operator; n is the total number of pixels in the local window; delta is one tenth of the maximum value of the amplitude of the LOG operator in the window; the magnitude of LOG operator is always larger than that of low texture region at the image edge, so T (S) is always larger than 1 at the edge region and smaller than 1 at the low texture region;
the following steps are taken in a simultaneous manner to obtain a guide filter weighted by an LOG operator:
in the formula: gamma is a parameter less than 1.
Compared with the prior art, the technical scheme of the invention has the beneficial effects that:
the stereo matching method based on cost aggregation provided by the invention uses the digital speckles to increase the texture information of the measured object, and meanwhile, the position between the binocular stereo vision module and the digital speckle projection module in the measuring system can be adjusted in a relatively movable manner, so that the stereo matching method based on cost aggregation has good adaptability to the measured objects with different sizes; meanwhile, for the measured object with weak texture and no texture, the measured object is not easily influenced by the illumination environment, so that the measurement precision is improved. In addition, for noise which can be introduced when speckles are projected, the invention provides a cost aggregation algorithm of secondary guided filtering based on LOG operator weighting, and compared with the traditional guided filtering cost aggregation algorithm, the cost aggregation algorithm introduces LOG operator weighting to enable the penalty parameter epsilon to be self-adaptive, so that the algorithm gives consideration to different textures; meanwhile, secondary guide filtering is introduced, so that the influence of noise can be inhibited, the stereo matching precision is improved, and the mismatching rate is reduced.
The invention solves the problems of poor adaptability and low precision of the existing stereo matching method due to the difficulty in adapting to measured objects with different sizes, excessive dependence on the texture characteristics of the measured objects and easiness in being influenced by illumination.
Drawings
Fig. 1 is a front view of the measurement system in this embodiment.
Fig. 2 is a bottom view of the measuring system of the present embodiment.
Fig. 3 is a general flowchart of the cost aggregation-based stereo matching method in this embodiment.
Fig. 4 is a flowchart of cost aggregation calculation performed by quadratic guided filtering based on LOG operator weighting in this embodiment.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for the purpose of better illustrating the embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;
it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
The embodiment provides a stereo matching method based on cost aggregation, as shown in fig. 3, including the following steps:
s1, establishing a measuring system, wherein the measuring system comprises a mounting plate 1, a stereoscopic vision module 3 and a digital speckle projection module 4, wherein the stereoscopic vision module 3 and the digital speckle projection module 4 are arranged on the mounting plate 1 in a sliding manner; the stereoscopic vision module 3 and the digital speckle projection module 4 are subjected to model selection and position adjustment according to a measured object and different measurement precision requirements, and then the positions of the stereoscopic vision module 3 and the digital speckle projection module are relatively fixed;
in this embodiment, be equipped with spout 5 on the mounting panel 1, stereoscopic vision module 3 and digital speckle projection module 4 are respectively through a slider 2 and 5 sliding connection of spout, and stereoscopic vision module 3 includes two relative parallel industry cameras in the physical position, and digital speckle projection module 4 includes digital speckle projecting apparatus, and digital speckle projecting apparatus is located between two industry cameras. The digital speckle projector and the industrial camera can be selected according to different measured objects and different measurement precisions, and the positions of the digital speckle projector and the industrial camera are adjusted on the slide rail, so that the digital speckle projector and the industrial camera are suitable for different measured objects. After the adjustment is finished, the positions of the stereoscopic vision module 3 and the digital speckle projection module 4 are relatively fixed, and then the subsequent three-dimensional measurement step is carried out. It should be noted that in this embodiment, the stereoscopic vision module 3 and the digital speckle projection module 4 can slide on the mounting plate 1 by using the matching of the sliding block 2 and the sliding groove 5, and in practical application, other connection methods can be adopted to adjust the positions of the two; in addition, the stereoscopic vision module 3 and the digital speckle projection module 4 can be directly obtained in the market, and therefore detailed description of the specific structure or working principle is omitted in this embodiment.
S2, calibrating and correcting images of two industrial cameras in the stereoscopic vision module 3;
and calibrating the two industrial cameras to obtain relevant parameters of the two industrial cameras, and then correcting the left and right imaging views of the two industrial cameras according to the calibration parameters of the industrial cameras to eliminate distortion and line alignment.
S3, forming digital speckles through the digital speckle projection module 4 and projecting the digital speckles onto the surface of a measured object; the digital speckle is used for increasing the texture information of the measured object, so that the measured object with weak texture and no texture is not easily influenced by the illumination environment;
and S4, shooting the image with the digital speckle texture through the stereoscopic vision module 3, preprocessing the image, and performing stereoscopic matching calculation to obtain three-dimensional data information of the measured object.
In this step, the stereoscopic vision module 3 first performs left and right view shooting through two industrial cameras at the same time, then corrects the left and right images with digital speckle textures according to calibration parameters obtained by calibrating the industrial cameras, and then performs stereo matching calculation:
s41, performing matching cost calculation on the left image and the right image preprocessed by the stereoscopic vision module 3 through an ADcensus stereoscopic matching algorithm;
s42, performing cost aggregation calculation on the matching cost obtained by the matching cost calculation by using secondary guide filtering based on LOG operator weighting;
the purpose of the cost aggregation is to enable the matching cost value obtained in step S41 to accurately reflect the correlation between pixels. The matching cost calculation only considers local information and is easily influenced by noise, the cost aggregation is similar to a parallax transmission step, the matching effect of a region with a high signal-to-noise ratio is good, the initial cost can well reflect the correlation, the optimal parallax value can be obtained more accurately, and the cost aggregation is transmitted to a region with a low signal-to-noise ratio and a poor matching effect, so that the real correlation can be accurately reflected by the cost values of all images.
In the embodiment, cost aggregation calculation is performed by using secondary guided filtering based on LOG operator weighting;
firstly, a guide filter weighted by a LOG operator is adopted:
the guiding filter weighted by the LOG operator is specifically:
the guide filtering is an edge-preserving algorithm of a local linear model, and a guide filtering model is constructed:
in the formula wkA square window with a radius r and a pixel k as a center; a iskAnd bkIs the invariant coefficient of the linear function when the window center is located at time k; i is a guide image, and q is an output image; i is a domain point with the point k as a central window;is any kBelong to wk;
Constructing a function model of a guide filtering algorithm:
in the formula: e is the function model energy; epsilon is a penalty parameter, preventing akThe value is too large; p is a radical of%I is the pixel value of the input image p; wherein a iskAnd bkThe value of (d) is found using least squares:
in the formula: mu.skAndpartial windows w, respectively, for the guide image IkMean and variance in (a); is window wkThe number of pixels in (1);for the image p to be filtered in the window wkThe average value of (1);
since in conventional filtering, to prevent akToo large, a fixed penalty parameter epsilon less than 1 is used. The ideal method is to use a large penalty parameter in the low texture area and a small penalty parameter in the edge area. Based on this, this embodiment uses the weighted filtering of the LOG operator to make punishment parameter self-adaptation adjustment, compromise the different textures of the measured object, wherein the expression of the LOG operator is:
in the formula: x and y are respectively the row and column coordinates of the pixel points; delta is the standard deviation; the expression of LOG operator weight term t(s) is:
in the formula: s is a central pixel in the local window; sg all pixels in the local window; | LOG (S) | represents the magnitude of the Gaussian Laplacian operator; n is the total number of pixels in the local window; delta is one tenth of the maximum value of the amplitude of the LOG operator in the window; the magnitude of the LOG operator is always greater at the image edge than in the low texture region, so t(s) is always greater than 1 in the edge (high texture) region and less than 1 in the low texture region;
the following steps are taken in a simultaneous manner to obtain a guide filter weighted by an LOG operator:
in the formula: gamma is a parameter less than 1.
The guiding filtering of the guiding filter weighted by the LOG operator effectively retains the edge information of the image, but cannot effectively estimate the noise information of the image, especially the high-frequency noise information. After one pilot filtering, there are still noise residuals in different bands. In order to overcome the defect of the primary guided filtering, the embodiment further provides a secondary guided filtering model based on LOG operator weighting, and the model is used for performing cost aggregation calculation to further suppress the influence of noise. As shown in fig. 4, the details are as follows:
s421, constructing a first guide filter and a second guide filter, wherein the first guide filter and the second guide filter are two same guide filters which are weighted by the LOG operator; are not described in detail herein;
s422, filtering the input image p by using the first guide filter, wherein the guide image of the first guide filter adopts a preset guide image I to obtain an output image q with maintained edges; the input image p is a matching cost obtained by calculating the matching cost of the left image and the right image after the pretreatment of the stereoscopic vision module through an ADcensus stereoscopic matching algorithm;
and S423, taking the output image q as a guide image of the second guide filter, filtering the input image p by using the second guide filter, and outputting an image q1 subjected to second guide filtering. Wherein the window of the first guiding filter is larger and the window of the second guiding filter is smaller in order to preserve image details. The quadratic pilot filter output image q1 can be considered as a pilot image q in the window wkLinear change in g.
S43, extracting initial parallax for the matching cost after cost aggregation calculation by using an algorithm with a Winner being a king (namely Winner-Take-All);
and S44, carrying out optimization processing on the initial parallax by adopting rapid weighted median filtering to obtain three-dimensional data information of the measured object.
The terms describing positional relationships in the drawings are for illustrative purposes only and are not to be construed as limiting the patent;
it should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (6)
1. A stereo matching method based on cost aggregation is characterized by comprising the following steps:
s1, establishing a measuring system, wherein the measuring system comprises a mounting plate, a stereoscopic vision module and a digital speckle projection module, and the stereoscopic vision module and the digital speckle projection module are arranged on the mounting plate in a sliding manner; wherein the stereoscopic vision module comprises two industrial cameras that are relatively parallel in physical location;
s2, calibrating and correcting images of two industrial cameras in the stereoscopic vision module;
s3, forming digital speckles through a digital speckle projection module and projecting the digital speckles to the surface of a measured object;
and S4, shooting the image with the digital speckle texture through a stereoscopic vision module, preprocessing the image, and performing stereoscopic matching calculation to obtain three-dimensional data information of the measured object.
2. The stereo matching method based on cost aggregation according to claim 1, wherein the step S1 further includes the steps of: and the stereoscopic vision module and the digital speckle projection module are subjected to model selection and position adjustment according to the measured object and different measurement precision requirements, and then the positions of the stereoscopic vision module and the digital speckle projection module are relatively fixed.
3. The stereo matching method based on cost aggregation according to claim 1, wherein the preprocessing of step S4 includes: and correcting the left and right images with the digital speckle textures shot by the stereoscopic vision module according to the calibration parameters obtained by the calibration of the industrial camera.
4. The stereo matching method based on cost aggregation according to claim 3, wherein the performing stereo matching calculation in step S4 specifically includes the following steps:
s41, performing matching cost calculation on the left image and the right image preprocessed by the stereoscopic vision module through an ADcensus stereoscopic matching algorithm;
s42, performing cost aggregation calculation on the matching cost obtained by the matching cost calculation by using secondary guide filtering based on LOG operator weighting;
s43, extracting initial parallax for the matching cost after cost aggregation calculation by using an algorithm with a winner as a king;
and S44, carrying out optimization processing on the initial parallax by adopting rapid weighted median filtering to obtain three-dimensional data information of the measured object.
5. The stereo matching method based on cost aggregation according to claim 4, wherein the specific step of step S42 includes:
s421, constructing a first guide filter and a second guide filter, wherein the first guide filter and the second guide filter are two same guide filters weighted by an LOG operator;
s422, filtering the input image p by using the first guide filter, wherein the guide image of the first guide filter adopts a preset guide image I or the input image p to obtain an output image q with maintained edges; the input image p is a matching cost obtained by calculating the matching cost of the left image and the right image after the pretreatment of the stereoscopic vision module through an ADcensus stereoscopic matching algorithm;
and S423, taking the output image q as a guide image of the second guide filter, filtering the input image p by using the second guide filter, and outputting an image q1 subjected to second guide filtering.
6. The cost aggregation-based stereo matching method according to claim 5, wherein the guided filter weighted by the LOG operator in step S421 is specifically:
constructing a guide filtering model:
in the formula wkA square window with a radius r and a pixel k as a center; a iskAnd bkIs the invariant coefficient of the linear function when the window center is located at time k; i is a guide image, and q is an output image; i is a domain point with the point k as a central window;k being arbitrary and all of wk;
Constructing a function model of a guide filtering algorithm:
in the formula: e is the function model energy; epsilon is a penalty parameter, preventing akThe value is too large; p is a radical of%I is the pixel value of the input image p; wherein a iskAnd bkThe value of (d) is found using least squares:
in the formula: mu.skAndpartial windows w, respectively, for the guide image IkMean and variance in (a); is window wkThe number of pixels in (1);for the image p to be filtered in the window wkThe average value of (1);
the weighted filtering of the LOG operator is used, so that the penalty parameter is adaptively adjusted, different textures of a measured object are considered, wherein the expression of the LOG operator is as follows:
in the formula: x and y are respectively the row and column coordinates of the pixel points; delta is the standard deviation; the expression of LOG operator weight term t(s) is:
in the formula: s is a central pixel in the local window; s' all pixels within a local window; | LOG (S) | represents the magnitude of the Gaussian Laplacian operator; n is the total number of pixels in the local window; delta is one tenth of the maximum value of the amplitude of the LOG operator in the window; the magnitude of LOG operator is always larger than that of low texture region at the image edge, so T (S) is always larger than 1 at the edge region and smaller than 1 at the low texture region;
the following steps are taken in a simultaneous manner to obtain a guide filter weighted by an LOG operator:
in the formula: gamma is a parameter less than 1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010093245.9A CN111351449A (en) | 2020-02-14 | 2020-02-14 | Stereo matching method based on cost aggregation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010093245.9A CN111351449A (en) | 2020-02-14 | 2020-02-14 | Stereo matching method based on cost aggregation |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111351449A true CN111351449A (en) | 2020-06-30 |
Family
ID=71194021
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010093245.9A Pending CN111351449A (en) | 2020-02-14 | 2020-02-14 | Stereo matching method based on cost aggregation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111351449A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112212806A (en) * | 2020-09-18 | 2021-01-12 | 南京理工大学 | Three-dimensional phase unfolding method based on phase information guidance |
CN112258591A (en) * | 2020-12-08 | 2021-01-22 | 之江实验室 | Method for obtaining high-precision depth map by combining multiple depth cameras |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040179204A1 (en) * | 2003-03-10 | 2004-09-16 | Fuji Photo Optical Co., Ltd. | Speckle interferometer apparatus |
CN204405025U (en) * | 2015-03-02 | 2015-06-17 | 武汉克诺德智能科技有限公司 | A kind of face battle array measurement mechanism based on binocular vision |
CN106340036A (en) * | 2016-08-08 | 2017-01-18 | 东南大学 | Binocular stereoscopic vision-based stereo matching method |
TW201731254A (en) * | 2015-10-22 | 2017-09-01 | 趨勢科技股份有限公司 | Phishing detection by login page census |
CN207382406U (en) * | 2017-10-17 | 2018-05-18 | 深圳市木白生物科技有限公司 | A kind of stent of three dimensional face imager |
CN109658443A (en) * | 2018-11-01 | 2019-04-19 | 北京华捷艾米科技有限公司 | Stereo vision matching method and system |
CN109816605A (en) * | 2019-01-16 | 2019-05-28 | 大连海事大学 | A kind of MSRCR image defogging method based on multichannel convolutive |
CN110084841A (en) * | 2019-04-29 | 2019-08-02 | 优乐圈(武汉)科技有限公司 | A kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator |
CN110148181A (en) * | 2019-04-25 | 2019-08-20 | 青岛康特网络科技有限公司 | A kind of general binocular solid matching process |
-
2020
- 2020-02-14 CN CN202010093245.9A patent/CN111351449A/en active Pending
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040179204A1 (en) * | 2003-03-10 | 2004-09-16 | Fuji Photo Optical Co., Ltd. | Speckle interferometer apparatus |
CN204405025U (en) * | 2015-03-02 | 2015-06-17 | 武汉克诺德智能科技有限公司 | A kind of face battle array measurement mechanism based on binocular vision |
TW201731254A (en) * | 2015-10-22 | 2017-09-01 | 趨勢科技股份有限公司 | Phishing detection by login page census |
CN106340036A (en) * | 2016-08-08 | 2017-01-18 | 东南大学 | Binocular stereoscopic vision-based stereo matching method |
CN207382406U (en) * | 2017-10-17 | 2018-05-18 | 深圳市木白生物科技有限公司 | A kind of stent of three dimensional face imager |
CN109658443A (en) * | 2018-11-01 | 2019-04-19 | 北京华捷艾米科技有限公司 | Stereo vision matching method and system |
CN109816605A (en) * | 2019-01-16 | 2019-05-28 | 大连海事大学 | A kind of MSRCR image defogging method based on multichannel convolutive |
CN110148181A (en) * | 2019-04-25 | 2019-08-20 | 青岛康特网络科技有限公司 | A kind of general binocular solid matching process |
CN110084841A (en) * | 2019-04-29 | 2019-08-02 | 优乐圈(武汉)科技有限公司 | A kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator |
Non-Patent Citations (4)
Title |
---|
NAKEUN CHOI: "Robust Stereo Matching Based on Cost Volume Fusion for Optimal Disparity Estimation", 《2019 IEEE INTERNATIONAL CONFERENCE ON CONSUMER ELECTRONICS (ICCE)》 * |
周博等: "基于高斯拉普拉斯算子的加权引导图滤波立体匹配算法", 《激光与光电子学进展》 * |
王凯: "基于二次引导滤波的局部立体匹配算法", 《激光与光电子学进展》 * |
陈田: "基于自适应权重的立体匹配技术研究与实现", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112212806A (en) * | 2020-09-18 | 2021-01-12 | 南京理工大学 | Three-dimensional phase unfolding method based on phase information guidance |
CN112258591A (en) * | 2020-12-08 | 2021-01-22 | 之江实验室 | Method for obtaining high-precision depth map by combining multiple depth cameras |
CN112258591B (en) * | 2020-12-08 | 2021-03-30 | 之江实验室 | Method for obtaining high-precision depth map by combining multiple depth cameras |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US8243056B2 (en) | Method for reconstructing a three-dimensional surface of an object | |
US8593524B2 (en) | Calibrating a camera system | |
CN108510540B (en) | Stereoscopic vision camera and height acquisition method thereof | |
Fuhl et al. | Fast camera focus estimation for gaze-based focus control | |
CN108181319B (en) | Accumulated dust detection device and method based on stereoscopic vision | |
CN102263985B (en) | Quality evaluation method, device and system of stereographic projection device | |
CN109961417B (en) | Image processing method, image processing apparatus, and mobile apparatus control method | |
CN110555878B (en) | Method and device for determining object space position form, storage medium and robot | |
CN109447908A (en) | A kind of coil of strip recognition positioning method based on stereoscopic vision | |
CN108491810A (en) | Vehicle limit for height method and system based on background modeling and binocular vision | |
GB2464453A (en) | Determining Surface Normals from Three Images | |
CN110830781A (en) | Automatic projected image correction method and system based on binocular vision | |
CN111351449A (en) | Stereo matching method based on cost aggregation | |
CN110209184A (en) | A kind of unmanned plane barrier-avoiding method based on binocular vision system | |
KR20110014067A (en) | Method and system for transformation of stereo content | |
CN103034330A (en) | Eye interaction method and system for video conference | |
CN109410234A (en) | A kind of control method and control system based on binocular vision avoidance | |
JP7378219B2 (en) | Imaging device, image processing device, control method, and program | |
CN111738241B (en) | Pupil detection method and device based on double cameras | |
CN108154536A (en) | The camera calibration method of two dimensional surface iteration | |
CN113240592A (en) | Distortion correction method for calculating virtual image plane based on AR-HUD dynamic eye position | |
CN105683707A (en) | Image photographing device and phase difference detecting method | |
EP4068207A1 (en) | Method of pixel-by-pixel registration of an event camera to a frame camera | |
CN109059868A (en) | A kind of binocular distance measuring method based on Adaptive matching window | |
CN109855845A (en) | A kind of binocular measures vehicle-mounted HUD virtual image distance and modification 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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200630 |