CN113179396B - Double-viewpoint stereo video fusion method based on K-means model - Google Patents
Double-viewpoint stereo video fusion method based on K-means model Download PDFInfo
- Publication number
- CN113179396B CN113179396B CN202110295931.9A CN202110295931A CN113179396B CN 113179396 B CN113179396 B CN 113179396B CN 202110295931 A CN202110295931 A CN 202110295931A CN 113179396 B CN113179396 B CN 113179396B
- Authority
- CN
- China
- Prior art keywords
- image
- viewpoint
- depth
- foreground
- depth image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000007500 overflow downdraw method Methods 0.000 title claims abstract description 7
- 238000000034 method Methods 0.000 claims abstract description 43
- 238000009877 rendering Methods 0.000 claims abstract description 37
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 230000004927 fusion Effects 0.000 claims abstract description 5
- 239000013598 vector Substances 0.000 claims description 10
- 239000000203 mixture Substances 0.000 claims description 9
- 238000003706 image smoothing Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 abstract description 4
- 230000000694 effects Effects 0.000 abstract description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/20—Image signal generators
- H04N13/275—Image signal generators from 3D object models, e.g. computer-generated stereoscopic image signals
- H04N13/279—Image signal generators from 3D object models, e.g. computer-generated stereoscopic image signals the virtual viewpoint locations being selected by the viewers or determined by tracking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
- H04N13/106—Processing image signals
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
- H04N13/106—Processing image signals
- H04N13/15—Processing image signals for colour aspects of image signals
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
- H04N13/106—Processing image signals
- H04N13/156—Mixing image signals
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention discloses a double-viewpoint three-dimensional video fusion method based on a K-means model. The method comprises the steps of firstly preprocessing a left viewpoint depth image and a right viewpoint depth image to obtain the left viewpoint depth image and the right viewpoint depth image; then, the depth images of the left viewpoint and the right viewpoint are respectively segmented by using a K-means method, and the three-dimensional projection operation is carried out on the segmented depth images of the foreground and background areas to obtain foreground and background drawing images of the left viewpoint and the right viewpoint; filling a vacant area of the foreground rendering image with the background rendering image when the foreground rendering image is a bluebook, and carrying out image fusion on the filled left and right viewpoint rendering images to obtain a virtual viewpoint rendering image; and finally, performing weighted filling on the hole area of the virtual viewpoint rendering image according to pixel information around the hole to obtain a final output image. The method of the invention adopts the operation of pixel level to accurately process the hollow area, so that the drawing effect is better and more harmonious in visual effect.
Description
Technical Field
The invention belongs to the technical field of stereo video coding and decoding, relates to a double-viewpoint stereo video fusion method based on a K-means model, and aims to improve the double-viewpoint image fusion process.
Background
Currently, depth Image Based Rendering (DIBR) is a main method for Rendering observation images from different viewing angles. And drawing the image of another visual angle according to the existing image so as to obtain the image observed under different visual angles. The most critical part of the mapping method is to perform the 3D-WARPING process. The process is that an image is restored to a three-dimensional model, and then the three-dimensional model is re-projected to a target plane near another target viewpoint to obtain an image at the virtual viewpoint. In the process of restoring the three-dimensional model and projecting onto a plane, the depth information of the image is very critical. The depth information of each pixel is very critical, and the multiple between the depth information of two views directly influences the result after rendering.
The most critical part of the DIBR technique is 3D Image transformation (3D Image warping), which is an operation of changing pixels of an Image. And mapping the pixel points of the reference image into the target view through three-dimensional transformation, thereby forming an original target view corresponding to the reference image.
The whole virtual viewpoint drawing process can be divided into two parts, and firstly, a depth map corresponding to the virtual viewpoint can be obtained by performing projection operation through the input depth image. In order to obtain the depth image at the virtual viewpoint, the most convenient method is completed by a three-dimensional projection operation process (3 d-forwarding). In the process, an image is reversely projected into a three-dimensional space to form a three-dimensional model, and then the three-dimensional model is re-projected to a target plane at a virtual viewpoint to obtain a virtual viewpoint image.
Disclosure of Invention
The invention aims to provide a double-viewpoint stereo video fusion method based on a K-means model.
The method comprises the following steps:
the method comprises the following steps of (1) preprocessing a left viewpoint depth map and a right viewpoint depth map to obtain a left viewpoint depth image and a right viewpoint depth image;
respectively segmenting the left viewpoint depth image and the right viewpoint depth image by using a K-means method, dividing the left viewpoint depth image and the right viewpoint depth image into a foreground region depth image and a background region depth image, and respectively performing three-dimensional projection operation on the foreground region depth image and the background region depth image to obtain a foreground rendering image and a background rendering image of a left viewpoint and a foreground rendering image and a background rendering image of a right viewpoint;
respectively fusing the foreground rendering image and the background rendering image of the left viewpoint and the right viewpoint: filling the vacant areas of the foreground drawing image with the background drawing image, wherein the foreground drawing image is a bluebook; obtaining a left viewpoint drawing image and a right viewpoint drawing image; carrying out image fusion on the filled left viewpoint drawing image and right viewpoint drawing image to obtain a virtual viewpoint drawing image;
and (4) carrying out weighted filling on the hole area of the virtual viewpoint drawing image according to pixel information around the hole to obtain a final output image.
Further, the preprocessing in the step (1) comprises a noise removing process and an image smoothing process; the noise removal processing is to select a residual error neural network for processing, and the image smoothing processing is to perform opening operation processing on the image after the noise removal processing.
Further, the step (2) is specifically:
carrying out clustering operation on the input left viewpoint depth image by a K-means method and segmenting, wherein the method comprises the following steps:
the pixel value distribution probability of the pixel points of the input depth image is { p } 0 ,p 1 ,…,p 255 }，p i I =0,1, \ 8230;, 255, which is the distribution probability of a pixel point having a pixel value i;
setting k thresholds { τ 1 ,τ 2 ,…,τ k Inputting the data into K-means operation, calculating the minimum Euclidean distance between each pixel value and each threshold value
Taking a threshold value corresponding to the minimum Euclidean distance as a first filling value; outputting new k threshold values { tau 'simultaneously' 1 ,τ′ 2 ,…,τ′ k The obtained value is used as the input of the next iteration operation;
selecting a threshold value tau each time after one iteration j A change occurs; will tau j Adjusted to set C j Statistical mean of all elements within, C j Is at τ j To tau j+1 Set of pixels in between:
x is C j A pixel of (1);
foreground region mask map
Foreground region Depth image FG = Depth × MOD FG ；
Background area mask map
Background region Depth image BG = Depth × MOD BG (ii) a K (i, j) represents a pixel value of the filling result, the corresponding pixel coordinate is (i, j), and Depth represents a Depth image;
respectively carrying out three-dimensional projection operation on the depth image of the foreground area and the depth image of the background area to obtain a drawing depth image of the foreground area and a drawing depth image of the background area;
respectively carrying out three-dimensional projection operation on the foreground area rendering depth image, the background area rendering depth image and the color image of the original viewpoint to obtain a foreground rendering image and a background rendering image of the left viewpoint;
and performing the same operation on the depth image of the right viewpoint to obtain a foreground drawing image and a background drawing image of the right viewpoint.
And further, in the step (3), carrying out image fusion on the filled left viewpoint drawing image and right viewpoint drawing image to obtain a virtual viewpoint drawing image, wherein the pixel value I of a pixel point with coordinates (x, y) in the image blend (x, y) is:
left viewpoint weight
Right viewpoint weight
R L And R R Spatial rotation matrices, T, for the left and right viewpoints, respectively L And T R The spatial translation vectors for the left and right views, respectively.
Further, in the step (4), the hollow area is formed by all I blend (x, y) =0 pixel configuration for each I blend A region in which the coordinates (x, y) of the pixel (x, y) =0 are 5 × 5 around the center is used as a filling region Ω, and the other 24 pixels except for (x, y) in the filling region Ω are divided into four groups of six pixels;
finally outputting the pixel value of the pixel point with the (x, y) image coordinate
Mean of non-hole pixels in mth group
Wherein, H (x, y) represents whether the coordinate (x, y) is a hole region, H (x, y) =0 if the coordinate is the hole region, and H (x, y) =1 if the coordinate is not the hole region;
weighted sum of pixels
Y (x, Y) is the priority of the coordinate (x, Y), W is the priority weight,
t is abbreviated as priority Y (x, Y).
The method of the invention uses a noise removal network and a smooth filtering method to preprocess the depth image, effectively reduces the drawing problem possibly generated by the depth image in the virtual viewpoint drawing process, and has positive influence on the improving process of drawing the color image. The method fuses the two-viewpoint images and uses a weighting filter based on geometric distance to optimize the fused images. The method breaks through the previous method of searching the best matching block, adopts the operation of pixel level to accurately process the hollow area, and leads the drawing effect to be better and more harmonious in visual effect.
Detailed Description
The double-viewpoint stereo video fusion method based on the K-means model specifically comprises the following steps:
the method comprises the following steps of (1) preprocessing a left viewpoint depth map and a right viewpoint depth map to obtain a left viewpoint depth image and a right viewpoint depth image; the preprocessing comprises noise removal processing and image smoothing processing; the noise removing processing is to select a residual error neural network for processing, and the image smoothing processing is to perform opening operation processing on the image after the noise removing processing.
Respectively segmenting the left viewpoint depth image and the right viewpoint depth image by using a K-means method, dividing the left viewpoint depth image and the right viewpoint depth image into a foreground region depth image and a background region depth image, and respectively performing three-dimensional projection operation (3 d-forwarding) on the foreground region depth image and the background region depth image to obtain a foreground drawing image and a background drawing image of a left viewpoint and a foreground drawing image and a background drawing image of a right viewpoint; the method comprises the following steps:
clustering operation is carried out on the input left viewpoint depth image through a K-means method and segmentation is carried out on the left viewpoint depth image, the K-means method is a clustering method based on data statistics, data with high similarity in a group of data are gathered into a class, then the data are divided into K groups of clusters, and the distance between the clusters is maximized as far as possible, and the method is as follows:
the pixel value distribution probability of the pixel points of the input depth image is { p } 0 ,p 1 ,…,p 255 }，p i I =0,1, \ 8230;, 255, which is the distribution probability of a pixel point having a pixel value i;
setting k thresholdsValue { tau 1 ,τ 2 ,…,τ k Inputting the data into K-means operation, calculating the minimum Euclidean distance between each pixel value and each threshold value
Taking a threshold value corresponding to the minimum Euclidean distance as a first filling value; simultaneously outputting new k threshold values { tau } 1 ′,τ 2 ′,…,τ k ' } is used as the input of the next iteration operation;
the selected threshold value tau is obtained after each iteration j A change occurs; will tau j Adjusted to set C j Statistical average of all elements in, C j Is at τ j To tau j+1 Set of pixels in between:
x is C j The pixel of (2);
foreground region mask map
Foreground region Depth image FG = Depth × MOD FG ；
Background region mask map
Background region Depth image BG = Depth × MOD BG ；
K (i, j) represents the fill result pixel value, the corresponding pixel coordinate is (i, j), and Depth represents the Depth image.
Respectively carrying out three-dimensional projection operation on the depth image of the foreground area and the depth image of the background area to obtain a drawing depth image of the foreground area and a drawing depth image of the background area; the three-dimensional projection operation method comprises the following steps:
Taking an original viewpoint coordinate vector M of 4 × 1 as an example, M = [ x, y, C, D (x, y)] T (ii) a T denotes transpose, x is abscissa of M, y is ordinate of M, C denotes constant coefficient, in this embodiment C =1, d (x, y) is depth value of M, the depth value reflects distance of a scene from a camera in an image, and the depth value can be calculated by depth image information:
Respectively carrying out three-dimensional projection operation on the foreground area rendering depth image, the background area rendering depth image and the color image of the original viewpoint to obtain a foreground rendering image and a background rendering image of the left viewpoint;
and performing the same operation on the depth image of the right viewpoint to obtain a foreground rendering image and a background rendering image of the right viewpoint.
And (3) fusing the foreground rendering image and the background rendering image of the left viewpoint and the right viewpoint respectively: filling the vacant areas of the foreground drawing image with the background drawing image, wherein the foreground drawing image is a bluebook; obtaining a left viewpoint drawing image and a right viewpoint drawing image; carrying out image fusion on the filled left viewpoint drawing image and right viewpoint drawing image to obtain a virtual viewpoint drawing image, wherein the pixel value I of a pixel point with coordinates (x, y) in the image blend (x, y) is:
left viewpoint weight
Right viewpoint weight
R L And R R Spatial rotation matrices, T, for the left and right viewpoints, respectively L And T R The spatial translation vectors of the left viewpoint and the right viewpoint, respectively.
And (4) carrying out weighted filling on the hole area of the virtual viewpoint drawing image according to the pixel information around the hole to obtain a final output image. The hollow region is composed of all I blend (x, y) =0 pixel configuration per I blend A region having a size of 5 × 5 with the coordinate (x, y) of the pixel of (x, y) =0 as the center is set as a filled region Ω, and the other 24 pixels in the filled region Ω except for (x, y) are divided into four groups of six pixels each.
Finally outputting the pixel value of the pixel point with the (x, y) image coordinate
Mean of non-hole pixels in mth group
Wherein, H (x, y) represents whether the coordinate (x, y) is a hole region, H (x, y) =0 if the coordinate is the hole region, and H (x, y) =1 if the coordinate is not the hole region; weightingSummation of pixels
Y (x, Y) is the priority of the coordinate (x, Y), W is the priority weight,
t is abbreviated as priority Y (x, Y).
Claims (1)
1. The double-view point stereo video fusion method based on the K-means model is characterized by comprising the following steps:
the method comprises the following steps of (1) preprocessing a left viewpoint depth map and a right viewpoint depth map to obtain a left viewpoint depth image and a right viewpoint depth image; the preprocessing comprises noise removal processing and image smoothing processing; the noise removal processing is to select a residual error neural network for processing, and the image smoothing processing is to perform opening operation processing on the image after the noise removal processing;
respectively segmenting the left viewpoint depth image and the right viewpoint depth image by using a K-means method, dividing the left viewpoint depth image and the right viewpoint depth image into a foreground region depth image and a background region depth image, and respectively performing three-dimensional projection operation on the foreground region depth image and the background region depth image to obtain a foreground rendering image and a background rendering image of a left viewpoint and a foreground rendering image and a background rendering image of a right viewpoint; the method comprises the following steps:
carrying out clustering operation on the input left viewpoint depth image by a K-means method and segmenting, wherein the method comprises the following steps:
the pixel value distribution probability of the pixel points of the input depth image is { p } 0 ,p 1 ,…,p 255 }，p i I =0,1, \ 8230;, 255, which is the distribution probability of a pixel point having a pixel value i;
set k thresholds { tau } 1 ,τ 2 ,…,τ k Inputting the pixel values into a K-means operation, calculating the minimum Euclidean distance between each pixel value and each threshold value
Threshold value corresponding to minimum Euclidean distanceFilling the value for the first time; outputting new k threshold values { tau 'simultaneously' 1 ,τ′ 2 ,…,τ′ k The arithmetic is used as the input of the next iteration operation;
the selected threshold value tau is obtained after each iteration j A change occurs; will tau j Adjusted to set C j Statistical mean of all elements within, C j Is at tau j To tau j+1 Set of pixels in between:
x is C j The pixel of (2);
foreground region mask map
Foreground region Depth image FG = Depth × MOD FG ；
Background area mask map
Background region Depth image BG = Depth × MOD BG (ii) a K (i, j) represents a filling result pixel value, the corresponding pixel coordinate is (i, j), and Depth represents a Depth image;
respectively carrying out three-dimensional projection operation on the depth image of the foreground area and the depth image of the background area to obtain a drawing depth image of the foreground area and a drawing depth image of the background area;
respectively carrying out three-dimensional projection operation on the foreground region rendering depth image, the background region rendering depth image and the color image of the original viewpoint to obtain a foreground rendering image and a background rendering image of the left viewpoint; the three-dimensional projection operation method comprises the following steps:
wherein M is v And M respectively represent coordinate vectors of a virtual viewpoint and an original viewpoint; i is v And I represents the internal parameter matrix of the camera of the virtual viewpoint and the original viewpoint respectively, and is only related to the internal performance of the camera; r v And R represents a virtual viewpoint and a raw viewpoint, respectivelyA partial rotation matrix with the size of 3 multiplied by 3 and representing the rotation angle of the camera in the three-dimensional space; t is v And T represents the translation vector of the camera of the virtual viewpoint and the original viewpoint respectively, the size of the translation vector is 3 multiplied by 1, and the translation vector of the camera in the three-dimensional space is represented;
performing the same operation on the depth image of the right viewpoint to obtain a foreground drawn image and a background drawn image of the right viewpoint;
and (3) fusing the foreground rendering image and the background rendering image of the left viewpoint and the right viewpoint respectively: filling the vacant areas of the foreground drawing image with the background drawing image, wherein the foreground drawing image is a bluebook; obtaining a left viewpoint drawing image and a right viewpoint drawing image; carrying out image fusion on the filled left viewpoint drawing image and right viewpoint drawing image to obtain a virtual viewpoint drawing image, wherein the pixel value I of a pixel point with coordinates (x, y) in the image blend (x, y) is:
wherein l (x, y) ∈ </h and r (x, y) ∈ </h respectively indicate that the left viewpoint and the right viewpoint are hollow regions at coordinates (x, y), l (x, y), and
respectively indicating that the left viewpoint and the right viewpoint are not hole areas at coordinates (x, y); i is L (x, y) and I R (x, y) respectively represent pixel values of the left viewpoint rendered image and the right viewpoint rendered image at coordinates (x, y);
left viewpoint weight
Right viewpoint weight
R L And R R Spatial rotation matrices, T, for the left and right viewpoints, respectively L And T R Spatial translation vectors for the left viewpoint and the right viewpoint, respectively;
step (4), carrying out weighted filling on a cavity area of the virtual viewpoint drawing image according to pixel information around the cavity to obtain a final output image;
the hollow region is composed of all I blend (x, y) =0 pixel configuration for each I blend A region with the size of 5 × 5 with the coordinate (x, y) of the pixel of (x, y) =0 as the center is used as a filling region Ω, and the other 24 pixels except for (x, y) in the filling region Ω are divided into four groups, each group including six pixels;
finally outputting the pixel value of the pixel point with (x, y) image coordinates
Mean of non-hole pixels in mth group
Wherein, H (x, y) represents whether the coordinate (x, y) is a hole region, H (x, y) =0 if the coordinate is the hole region, and H (x, y) =1 if the coordinate is not the hole region;
weighted sum of pixels
Y (x, Y) is the priority of the coordinate (x, Y), W is the priority weight,
t is abbreviated as priority Y (x, Y).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110295931.9A CN113179396B (en) | 2021-03-19 | 2021-03-19 | Double-viewpoint stereo video fusion method based on K-means model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110295931.9A CN113179396B (en) | 2021-03-19 | 2021-03-19 | Double-viewpoint stereo video fusion method based on K-means model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113179396A CN113179396A (en) | 2021-07-27 |
| CN113179396B true CN113179396B (en) | 2022-11-11 |
Family
ID=76922161
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110295931.9A Active CN113179396B (en) | 2021-03-19 | 2021-03-19 | Double-viewpoint stereo video fusion method based on K-means model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113179396B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102055982A (en) * | 2011-01-13 | 2011-05-11 | 浙江大学 | Coding and decoding methods and devices for three-dimensional video |
| CN105141940A (en) * | 2015-08-18 | 2015-12-09 | 太原科技大学 | 3D video coding method based on regional division |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102592275B (en) * | 2011-12-16 | 2013-12-25 | 天津大学 | Virtual viewpoint rendering method |
| CN102903110B (en) * | 2012-09-29 | 2015-11-25 | 宁波大学 | To the dividing method of image with deep image information |
| CN103269438A (en) * | 2013-05-27 | 2013-08-28 | 中山大学 | Method for drawing depth image on the basis of 3D video and free-viewpoint television |
| CN103996174B (en) * | 2014-05-12 | 2017-05-10 | 上海大学 | Method for performing hole repair on Kinect depth images |
| CN109712067B (en) * | 2018-12-03 | 2021-05-28 | 北京航空航天大学 | Virtual viewpoint drawing method based on depth image |
| CN111385554B (en) * | 2020-03-28 | 2022-07-08 | 浙江工业大学 | High-image-quality virtual viewpoint drawing method of free viewpoint video |
-
2021
- 2021-03-19 CN CN202110295931.9A patent/CN113179396B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102055982A (en) * | 2011-01-13 | 2011-05-11 | 浙江大学 | Coding and decoding methods and devices for three-dimensional video |
| CN105141940A (en) * | 2015-08-18 | 2015-12-09 | 太原科技大学 | 3D video coding method based on regional division |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113179396A (en) | 2021-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113192179B (en) | Three-dimensional reconstruction method based on binocular stereo vision | |
| CN111325693B (en) | Large-scale panoramic viewpoint synthesis method based on single viewpoint RGB-D image | |
| CN110853151A (en) | Three-dimensional point set recovery method based on video | |
| KR100560464B1 (en) | Multi-view display system with viewpoint adaptation | |
| CN111462030A (en) | Multi-image fused stereoscopic set vision new angle construction drawing method | |
| CN115298708A (en) | Multi-view neural human body rendering | |
| Tomiyama et al. | Algorithm for dynamic 3D object generation from multi-viewpoint images | |
| Xu et al. | Layout-guided novel view synthesis from a single indoor panorama | |
| CN111951368A (en) | Point cloud, voxel and multi-view fusion deep learning method | |
| Zhu et al. | An improved depth image based virtual view synthesis method for interactive 3D video | |
| Ma et al. | Depth-guided inpainting algorithm for free-viewpoint video | |
| Jantet et al. | Joint projection filling method for occlusion handling in depth-image-based rendering | |
| Zhu et al. | Occlusion-free scene recovery via neural radiance fields | |
| CN117501313A (en) | Hair rendering system based on deep neural network | |
| CN113179396B (en) | Double-viewpoint stereo video fusion method based on K-means model | |
| CN117730530A (en) | Image processing method and device, equipment and storage medium | |
| CN116681839A (en) | Live three-dimensional target reconstruction and singulation method based on improved NeRF | |
| CN113450274B (en) | Self-adaptive viewpoint fusion method and system based on deep learning | |
| CN115409932A (en) | Texture mapping and completion method of three-dimensional human head and face model | |
| CN114742954A (en) | Method for constructing large-scale diversified human face image and model data pairs | |
| CN111178163B (en) | Stereoscopic panoramic image salient region prediction method based on cube projection format | |
| CN113763474A (en) | Scene geometric constraint-based indoor monocular depth estimation method | |
| CN112364711A (en) | 3D face recognition method, device and system | |
| Lee et al. | Hole concealment for depth image using pixel classification in multiview system | |
| Lee et al. | Removing foreground objects by using depth information from multi-view images |
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 |