CN111784586B - Self-supervision learning panoramic image horizontal correction method and system - Google Patents
Self-supervision learning panoramic image horizontal correction method and system Download PDFInfo
- Publication number
- CN111784586B CN111784586B CN202010446601.0A CN202010446601A CN111784586B CN 111784586 B CN111784586 B CN 111784586B CN 202010446601 A CN202010446601 A CN 202010446601A CN 111784586 B CN111784586 B CN 111784586B
- Authority
- CN
- China
- Prior art keywords
- image
- training
- angle
- panoramic image
- horizontal
- 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
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000012937 correction Methods 0.000 title claims abstract description 15
- 238000012549 training Methods 0.000 claims abstract description 83
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 29
- 239000011159 matrix material Substances 0.000 claims abstract description 23
- 238000004364 calculation method Methods 0.000 claims abstract description 11
- 230000006870 function Effects 0.000 claims abstract description 9
- 230000002194 synthesizing effect Effects 0.000 claims abstract description 5
- 238000012544 monitoring process Methods 0.000 claims description 5
- 238000009795 derivation Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 5
- 238000007781 pre-processing Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003702 image correction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005067 remediation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/60—Rotation of whole images or parts thereof
- G06T3/608—Rotation of whole images or parts thereof by skew deformation, e.g. two-pass or three-pass rotation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a panoramic image horizontal correction method and a panoramic image horizontal correction system for self-supervision learning, wherein the method comprises the following steps: constructing a training image library; carrying out self-supervision training on the convolutional neural network by using the training images in the training image library until the error function of the convolutional neural network is converged; inputting the image to be corrected in the training image library into a trained convolutional neural network, and deducing a pitch angle and a roller angle of a camera when the current panoramic image is shot; and carrying out Euler angle calculation rotation matrix calculation on the pitch angle and the roller angle to obtain a rotation matrix of the camera pose, correcting the image to be corrected according to the rotation matrix of the camera pose, and synthesizing a horizontal image. The method effectively solves the problem of panoramic image distortion caused by non-vertical camera attitude.
Description
Technical Field
The invention relates to the technical field of image correction, in particular to a panoramic image horizontal correction method and a panoramic image horizontal correction system for self-supervision learning.
Background
The existing panoramic image can be applied to scenes such as virtual reality and three-dimensional reconstruction. When a panoramic image is shot, the attitude of the panoramic camera can cause great influence on the acquired image. Ideally, the attitude of the panoramic camera should be kept perpendicular to the ground, and the panoramic image acquired at this time is horizontal. It can be intuitively understood that the horizon in an image approximates a horizontal straight line. However, in practical situations, the panoramic camera is not necessarily placed vertically, or there are interference situations such as camera shake, which results in that the photographed panoramic image is not horizontal. As shown in fig. 1, the horizon in the image approximates a sinusoidal curve and is accompanied by large distortions and distortions, which can be troublesome for subsequent use of the panoramic image.
In the related art, an additional gyroscope is used to acquire the three-axis pose of a camera, and the acquired image is subjected to rotation correction to generate a horizontal image. However, this approach increases the hardware cost and complexity of the system, and some panoramas do not have matching gyroscope data to acquire.
Therefore, a problem of distortion and distortion of a panoramic image caused by non-vertical posture of a camera is urgently needed to be solved.
Disclosure of Invention
The present invention is directed to solving, at least in part, one of the technical problems in the related art.
Therefore, an object of the present invention is to provide a method for horizontally rectifying a panoramic image through auto-supervised learning, which can solve the problem of distortion and distortion of the panoramic image caused by the vertical camera without additional auxiliary equipment, and has low hardware cost and low complexity.
Another objective of the present invention is to provide a panoramic image level rectification system for self-supervised learning.
In order to achieve the above object, an embodiment of the invention provides a panoramic image horizontal rectification method for self-supervised learning, including the following steps: s1, constructing a training image library; s2, performing self-supervision training on a convolutional neural network by using the training images of the training image library until an error function of the convolutional neural network is converged; s3, inputting the image to be corrected in the training image library into a trained convolutional neural network, and deducing a pitch angle and a roller angle of a camera when the image to be corrected is shot; and S4, carrying out Euler angle calculation rotation matrix calculation on the pitch angle and the roller angle to obtain a rotation matrix of the camera pose, correcting the image to be corrected according to the rotation matrix of the camera pose, and synthesizing a horizontal image.
According to the self-supervised learning panoramic image horizontal correction method, enough horizontal panoramic images are obtained, a great amount of horizontal panoramic images are used for carrying out self-supervised training on the neural network, so that the network has the capability of deducing the attitude angle of the camera, the rotation angle of the camera during panoramic image shooting is estimated, the attitude of the camera is compensated according to the rotation angle, a horizontal panoramic image is obtained, and the problem of distortion and distortion of the panoramic image caused by non-vertical attitude of the camera is effectively solved.
In addition, the panoramic image level rectification method of the self-supervised learning according to the above embodiment of the present invention may further have the following additional technical features:
further, in one embodiment of the present invention, each panoramic image in the training image library is horizontally angled.
Further, in an embodiment of the present invention, the step S2 includes: using a pitch angle and a roller angle in Euler angles as monitoring information; performing iterative training on each panoramic image in the training image library through the supervision information to generate a plurality of rotated non-horizontal images; and when the error between the rotation angle of the non-horizontal image and the rotation angle in the supervision information is minimum, finishing the training of the convolutional neural network.
Further, in one embodiment of the present invention, the panoramic image is mapped onto a sphere before the panoramic image is iteratively trained.
In order to achieve the above object, another embodiment of the present invention provides a panoramic image horizontal rectification system for self-supervised learning, including: the construction module is used for constructing a training image library; the training module is used for carrying out self-supervision training on the convolutional neural network by utilizing the training images of the training image library until the error function of the convolutional neural network is converged; the derivation module is used for inputting the image to be corrected in the training image library into a trained convolutional neural network and deriving a pitch angle and a roller angle of a camera when the image to be corrected is shot; and the correction module is used for calculating Euler angle calculation rotation matrixes of the pitch angle and the roller angle to obtain a rotation matrix of a camera pose, correcting the image to be corrected according to the rotation matrix of the camera pose and synthesizing a horizontal image.
The panoramic image horizontal correction system for the self-supervised learning of the embodiment of the invention acquires a sufficient number of horizontal panoramic images, uses a large number of horizontal panoramic images to carry out self-supervised training on the neural network, so that the network has the capability of reasoning the attitude angle of the camera, estimates the rotation angle of the camera during panoramic image shooting, compensates the camera pose according to the rotation angle to obtain the horizontal panoramic image, and effectively solves the problem of distortion and distortion of the panoramic image caused by non-vertical attitude of the camera.
In addition, the panoramic image level rectification system for the self-supervised learning according to the above embodiment of the present invention may further have the following additional technical features:
further, in one embodiment of the present invention, each panoramic image in the training image library is horizontally angled.
Further, in an embodiment of the present invention, the training module is further configured to: a determination unit for using a pitch angle and a drum angle among euler angles as supervision information; the training unit is used for carrying out iterative training on each panoramic image in the training image library through the supervision information to generate a plurality of rotated non-horizontal images; and the judging unit is used for finishing the training of the convolutional neural network when the error between the rotation angle of the non-horizontal image and the rotation angle in the supervision information is minimum.
Further, in one embodiment of the present invention, the panoramic image is mapped onto a sphere before the panoramic image is iteratively trained.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The foregoing and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
FIG. 1 is a schematic diagram of a non-horizontal panoramic image with a large amount of distortion and deformity;
FIG. 2 is a flowchart of a panoramic image level rectification method for auto-supervised learning according to an embodiment of the present invention;
FIG. 3 is a diagram illustrating a planar to spherical correspondence of a panoramic image, according to one embodiment of the present invention;
FIG. 4 is a simplified flowchart of a method for self-supervised learning panoramic image level rectification according to an embodiment of the present invention;
FIG. 5 is a schematic diagram illustrating image comparison before and after rectification according to one embodiment of the present invention;
fig. 6 is a schematic structural diagram of a panoramic image level rectification system for auto-supervised learning according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
The method and system for correcting the level of a panoramic image by self-supervised learning according to an embodiment of the present invention will be described below with reference to the accompanying drawings, and first, the method for correcting the level of a panoramic image by self-supervised learning according to an embodiment of the present invention will be described with reference to the accompanying drawings.
Fig. 1 is a flowchart of a panoramic image horizontal rectification method of the self-supervised learning according to an embodiment of the present invention.
As shown in fig. 1, the method for leveling a panoramic image through self-supervised learning includes the following steps:
in step S1, a training image library is constructed, wherein each panoramic image in the training image library is a horizontal angle.
Specifically, a large number of horizontally shot panoramic images are collected, the panorama can be collected on a tripod or a trolley, a public panoramic image database can be used, or the panorama can be crawled from a street view map website. Regardless of the source of the images, it is necessary to ensure that each panoramic image in the image library is horizontal. Some pre-processing of the image may be appropriate, such as cropping the area in which the capture device or support device appears in the image, etc. Step S1 is performed by acquiring a sufficient number of horizontal panoramic images for subsequent model training.
In step S2, the training images in the training image library are used to perform an auto-supervised training on the convolutional neural network until the error function of the convolutional neural network converges.
Further, in one embodiment of the present invention, step S2 includes: the pitch angle and the roller angle in the Euler angles are used as monitoring information; performing iterative training on each panoramic image in the training image library through the monitoring information to generate a plurality of rotated non-horizontal images; and when the error between the rotation angle of the non-horizontal image and the rotation angle in the supervision information is minimum, finishing the training of the convolutional neural network.
That is, step S2 is to apply a random three-dimensional rotation to each training image and record the angle of the rotation. The three-dimensional rotation does not mean simple plane rotation, but is rotation of the posture of the camera in a three-dimensional space when the image is captured. The rotation requires mapping the panoramic image onto a spherical surface of the device, multiplying the matrix and the coordinates corresponding to the rotation, and then mapping back to a plane, which is described in detail in the following steps.
In which three-dimensional rotation can be represented by euler angles, i.e. three angles, and one of the angles is rotation in the horizontal direction, which is meaningless for horizontal correction, so that only two of the angles are selected: pitch angle pitch and roll angle roll. To this end, each horizontal image may generate several rotated non-horizontal images, and the two rotation angles pitch and roll it experiences are known. And using the two known rotation angles as the training supervision information to minimize the errors of the rotation angle output by the network and the real rotation angle when the corresponding non-horizontal image is input as much as possible. And repeatedly carrying out iterative training until the error function of the convolutional neural network is converged, and storing the model parameters at the moment.
In step S3, the image to be corrected in the training image library is input to the trained convolutional neural network, and the pitch angle and the drum angle of the camera when the image to be corrected is captured are derived.
That is, the model parameters trained in the previous step are loaded for the network model. And (4) carrying out preprocessing such as scaling and cutting on the panoramic image, and then sending the panoramic image into a convolutional neural network. And the neural network reasoning obtains the pitch angle pitch and the roller angle roll of the camera when the panoramic image is shot.
In step S4, euler angle calculation rotation matrix calculation is carried out on the pitch angle and the roller angle to obtain a rotation matrix of the camera pose, the image to be corrected is corrected according to the rotation matrix of the camera pose, and a horizontal image is synthesized.
That is, after the pitch angle pitch and the roll angle roll of the camera at the time of input panoramic image capturing are estimated, the rotation matrix R representing the camera posture is calculated with reference to the correlation data estimated from the euler angle. As shown in fig. 3, a panoramic image can be mapped onto a panoramic spherical surface, and the horizontal direction can be regarded as the longitudinal direction, and the vertical direction can be regarded as the latitudinal direction. Generally, a certain pixel (U, v) on the panoramic image and the coordinates (x, y, z) on the sphere are mapped as follows, where U is c ，V c And f is respectively the abscissa, the ordinate and the pixel of the center point of the panoramic image.
After obtaining the point coordinates on the spherical surface corresponding to each pixel on the panoramic image, using the inverse matrix R of the camera pose rotation matrix R -1 Compensating the spherical coordinates to obtain a new spherical seat (x) - ,y _ ,z _ )＝R -1 (x,y,z) T . Further, by using the mapping relation, theThe coordinates (u) of the corresponding pixels in the corrected panoramic image are calculated reversely - ,v _ ). According to the pixel corresponding relation (u, v) (u) in the panoramic image before and after rectification - ,v _ ) A new rectified horizontal panoramic image can be synthesized.
Briefly, as shown in fig. 4, the operation principle of the panoramic image level rectification method of the embodiment of the present invention is as follows: firstly, a training image library is constructed, images in the training image library are utilized to carry out self-supervision type training to obtain a trained neural model, images to be corrected are input into the trained neural model, the rotation angle of a camera is calculated, the images to be corrected are corrected according to the angle, and a new horizontal panoramic image is synthesized.
According to the self-supervised learning panoramic image horizontal correction method provided by the embodiment of the invention, a large-scale training image database is constructed, a network model capable of predicting the three-axis posture of a camera during panoramic image shooting is obtained through training by using the self-supervised learning method, the pitch angle pitch and the roller angle roll of the camera during input panoramic image shooting can be estimated by using the model, and then the panoramic image is subjected to three-dimensional rotation to generate a horizontal panoramic image. As shown in fig. 5, the horizontal panorama has a smaller distortion and distortion degree than an unprocessed panorama, so that subsequent operations such as retrieval, deduplication, feature extraction, feature matching and the like have better effects, and the horizontal panorama can be used as a powerful preprocessing means for a panoramic image and has a greater theoretical and practical value for technical research in the field.
Next, a panoramic image level rectification system for auto-supervised learning proposed according to an embodiment of the present invention is described with reference to the drawings.
Fig. 6 is a schematic structural diagram of a panoramic image level rectification system for self-supervised learning according to an embodiment of the present invention.
As shown in fig. 6, the system 10 includes: a building module 100, a training module 200, a derivation module 300, and a remediation module 400.
The construction module 100 is configured to construct a training image library, where each panoramic image in the training image library is a horizontal angle. The training module 200 is configured to perform an auto-supervised training on the convolutional neural network by using the training images in the training image library until an error function of the convolutional neural network converges. The derivation module 300 is configured to input the image to be corrected in the training image library into the trained convolutional neural network, and derive a pitch angle and a drum angle of the camera when the image to be corrected is captured. The correction module 400 is configured to calculate an euler angle calculation rotation matrix for the pitch angle and the drum angle to obtain a rotation matrix of the camera pose, correct the image to be corrected according to the rotation matrix of the camera pose, and synthesize a horizontal image.
Further, in an embodiment of the present invention, the training module 200 is further configured to:
a determination unit for using a pitch angle and a drum angle among euler angles as supervision information;
the training unit is used for carrying out iterative training on each panoramic image in the training image library through the monitoring information to generate a plurality of rotated non-horizontal images;
and the judging unit is used for finishing the training of the convolutional neural network when the error between the rotation angle of the non-horizontal image and the rotation angle in the supervision information is minimum.
Further, in one embodiment of the present invention, the panoramic image is mapped onto a sphere before being iteratively trained.
According to the self-supervised learning panoramic image horizontal rectification system provided by the embodiment of the invention, a large-scale training image database is constructed, a network model capable of predicting the three-axis attitude of a camera during panoramic image shooting is obtained through training by using a self-supervised learning method, the pitch angle pitch and the roller angle roll of the camera during input panoramic image shooting can be estimated by using the model, and then the panoramic image is subjected to three-dimensional rotation to generate a horizontal panoramic image. Compared with an unprocessed panoramic image, the horizontal panoramic image has smaller distortion and distortion degree, so that the subsequent operations of retrieval, duplicate removal, feature extraction, feature matching and the like have better effects, can be used as a powerful preprocessing means for the panoramic image, and has larger theoretical and practical values for technical research in the field.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or to implicitly indicate the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
In the description of the specification, reference to the description of "one embodiment," "some embodiments," "an example," "a specific example," or "some examples" or the like means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
Claims (4)
1. A panoramic image horizontal rectification method for self-supervision learning is characterized by comprising the following steps:
s1, constructing a training image library;
s2, performing self-supervision training on a convolutional neural network by using the training images in the training image library until an error function of the convolutional neural network is converged;
s3, inputting the image to be corrected in the training image library into a trained convolutional neural network, and deducing a pitch angle and a roller angle of a camera when the image to be corrected is shot; and
s4, carrying out Euler angle calculation rotation matrix calculation on the pitch angle and the roller angle to obtain a rotation matrix of a camera pose, correcting the image to be corrected according to the rotation matrix of the camera pose, and synthesizing a horizontal image;
each panoramic image in the training image library is a horizontal angle;
wherein the step S2 includes:
the pitch angle and the roller angle in the Euler angles are used as monitoring information;
performing iterative training on each panoramic image in the training image library through the supervision information to generate a plurality of rotated non-horizontal images;
and when the error between the rotation angle of the non-horizontal image and the rotation angle in the supervision information is minimum, finishing the training of the convolutional neural network.
2. The method of claim 1, wherein the panoramic image is mapped onto a spherical surface before iterative training of the panoramic image.
3. A panoramic image level correction system for self-supervised learning, comprising:
the construction module is used for constructing a training image library;
the training module is used for carrying out self-supervision training on the convolutional neural network by utilizing the training images of the training image library until the error function of the convolutional neural network is converged;
the derivation module is used for inputting the image to be corrected in the training image library into a trained convolutional neural network and deriving a pitch angle and a roller angle of a camera when the image to be corrected is shot; and
the correction module is used for calculating Euler angle calculation rotation matrixes of the pitch angle and the roller angle to obtain a rotation matrix of a camera pose, correcting the image to be corrected according to the rotation matrix of the camera pose and synthesizing a horizontal image;
each panoramic image in the training image library is a horizontal angle;
wherein the training module is further to:
a determination unit for using a pitch angle and a drum angle among euler angles as supervision information;
the training unit is used for carrying out iterative training on each panoramic image in the training image library through the supervision information to generate a plurality of rotated non-horizontal images;
and the judging unit is used for finishing the training of the convolutional neural network when the error between the rotation angle of the non-horizontal image and the rotation angle in the supervision information is minimum.
4. The system of claim 3, wherein the panoramic image is mapped onto a sphere before being iteratively trained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010446601.0A CN111784586B (en) | 2020-05-25 | 2020-05-25 | Self-supervision learning panoramic image horizontal correction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010446601.0A CN111784586B (en) | 2020-05-25 | 2020-05-25 | Self-supervision learning panoramic image horizontal correction method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111784586A CN111784586A (en) | 2020-10-16 |
| CN111784586B true CN111784586B (en) | 2022-10-18 |
Family
ID=72753202
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010446601.0A Active CN111784586B (en) | 2020-05-25 | 2020-05-25 | Self-supervision learning panoramic image horizontal correction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111784586B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109472739A (en) * | 2018-10-29 | 2019-03-15 | 深圳云行智能科技有限公司 | A kind of fish eye images antidote and device for wide-angle camera |
| CN110248079A (en) * | 2019-02-14 | 2019-09-17 | 南京泓众电子科技有限公司 | A kind of full view image generating system and panorama image generation method |
| CN110675453A (en) * | 2019-10-16 | 2020-01-10 | 北京天睿空间科技股份有限公司 | Self-positioning method for moving target in known scene |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102227583B1 (en) * | 2018-08-03 | 2021-03-15 | 한국과학기술원 | Method and apparatus for camera calibration based on deep learning |
-
2020
- 2020-05-25 CN CN202010446601.0A patent/CN111784586B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109472739A (en) * | 2018-10-29 | 2019-03-15 | 深圳云行智能科技有限公司 | A kind of fish eye images antidote and device for wide-angle camera |
| CN110248079A (en) * | 2019-02-14 | 2019-09-17 | 南京泓众电子科技有限公司 | A kind of full view image generating system and panorama image generation method |
| CN110675453A (en) * | 2019-10-16 | 2020-01-10 | 北京天睿空间科技股份有限公司 | Self-positioning method for moving target in known scene |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111784586A (en) | 2020-10-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110211043B (en) | Registration method based on grid optimization for panoramic image stitching | |
| US5706416A (en) | Method and apparatus for relating and combining multiple images of the same scene or object(s) | |
| CN108564617A (en) | Three-dimensional rebuilding method, device, VR cameras and the panorama camera of more mesh cameras | |
| US20070008312A1 (en) | Method for determining camera position from two-dimensional images that form a panorama | |
| CN111355884B (en) | Monitoring method, device, system, electronic equipment and storage medium | |
| CN102289803A (en) | Image Processing Apparatus, Image Processing Method, and Program | |
| CN110111388A (en) | Three-dimension object pose parameter estimation method and visual apparatus | |
| CN111123962A (en) | Rotor unmanned aerial vehicle repositioning photographing method for power tower inspection | |
| CN116883251B (en) | Image orientation splicing and three-dimensional modeling method based on unmanned aerial vehicle video | |
| CN114372992A (en) | Edge corner point detection four-eye vision algorithm based on moving platform | |
| CN113962853A (en) | Automatic precise resolving method for rotary linear array scanning image pose | |
| CN109934765B (en) | High-speed camera panoramic image splicing method | |
| CN112733773B (en) | Object detection method, device, computer equipment and storage medium | |
| CN111784586B (en) | Self-supervision learning panoramic image horizontal correction method and system | |
| CN111008932B (en) | Panoramic image splicing method based on image screening | |
| CN116402693B (en) | Municipal engineering image processing method and device based on remote sensing technology | |
| CN113436269A (en) | Image dense stereo matching method and device and computer equipment | |
| CN113436267A (en) | Visual inertial navigation calibration method and device, computer equipment and storage medium | |
| CN113361400B (en) | Head posture estimation method, device and storage medium | |
| CN108681985B (en) | Stripe splicing method of video satellite images | |
| CN116132636A (en) | Video splicing method and device for fully-mechanized coal mining face | |
| CN113205591B (en) | Method and device for acquiring three-dimensional reconstruction training data and electronic equipment | |
| US20230326126A1 (en) | Generation of a 3d point cloud of a scene | |
| CN114187344A (en) | Map construction method, device and equipment | |
| EP4281936A1 (en) | Systems and methods for roof area and slope estimation using a point set |
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 |