WO2017193729A1 - 一种将镜头图像展开为全景图像的方法及装置 - Google Patents

一种将镜头图像展开为全景图像的方法及装置 Download PDF

Info

Publication number
WO2017193729A1
WO2017193729A1 PCT/CN2017/079093 CN2017079093W WO2017193729A1 WO 2017193729 A1 WO2017193729 A1 WO 2017193729A1 CN 2017079093 W CN2017079093 W CN 2017079093W WO 2017193729 A1 WO2017193729 A1 WO 2017193729A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
mesh model
lens image
new
panoramic
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.)
Ceased
Application number
PCT/CN2017/079093
Other languages
English (en)
French (fr)
Inventor
王士博
沈靖程
王超
熊磊
邱汉文
谭海富
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Pisofttech Technology Co Ltd
Original Assignee
Shenzhen Pisofttech Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen Pisofttech Technology Co Ltd filed Critical Shenzhen Pisofttech Technology Co Ltd
Priority to US16/309,806 priority Critical patent/US10798300B2/en
Priority to KR1020187035978A priority patent/KR102206899B1/ko
Priority to EP17795347.8A priority patent/EP3457681A4/en
Publication of WO2017193729A1 publication Critical patent/WO2017193729A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00Geometric image transformations in the plane of the image
    • G06T3/10Selection of transformation methods according to the characteristics of the input images
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00Geometric image transformations in the plane of the image
    • G06T3/04Context-preserving transformations, e.g. by using an importance map
    • G06T3/047Fisheye or wide-angle transformations
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/80Camera processing pipelines; Components thereof
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T15/00Three-dimensional [3D] image rendering
    • G06T15/10Geometric effects
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00Image enhancement or restoration
    • G06T5/80Geometric correction
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60Control of cameras or camera modules
    • H04N23/698Control of cameras or camera modules for achieving an enlarged field of view, e.g. panoramic image capture
    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2200/00Indexing scheme for image data processing or generation, in general
    • G06T2200/28Indexing scheme for image data processing or generation, in general involving image processing hardware

Definitions

  • the present invention belongs to the field of image processing technologies, and in particular, to a method and apparatus for expanding a lens image into a panoramic image.
  • Panoramic images and panoramic video are one of the important components of VR content, and current panoramic images. It is usually shot by a large angle fisheye lens camera, or by several wide-angle or fisheye cameras and stitched together. Since the images taken by wide-angle and fisheye cameras are distorted, they need to pass. The algorithm spreads the image taken by the wide-angle or fisheye camera into a panoramic image. Most of the current panoramic image display algorithms run on the CPU (central processing unit), and a few panoramic image display algorithms run on the GPU (Graphic Processing Unit, GPU), because the panoramic image display algorithm is computationally intensive. The time required to display a frame of image is very long. Generally, high-performance CPUs with high cost, high energy consumption and large volume are used. However, it is difficult for the CPU to achieve high-definition or real-time image display. Realize panoramic live broadcast.
  • the present invention provides a method and apparatus for expanding a lens image into a panoramic image, so as to quickly expand the lens image into a panoramic image and reduce the panoramic device. Cost, energy consumption and reduced volume of panoramic equipment.
  • the present invention provides a method of expanding a lens image into a panoramic image, comprising the following steps:
  • Step A preparing an original mesh model, calculating a new texture coordinate value, performing coordinate value conversion on the original texture coordinate value of the original mesh model and the calculated new texture coordinate value, and generating a new network Grid model and save;
  • Step B preparing a lens image to be expanded, assigning the lens image to be expanded to the new mesh model in the step A, and performing the new mesh model through the GPU Render, get a panoramic image and save it.
  • step A includes the following steps:
  • Step Al preparing an original mesh model
  • Step A2 preparing a lens image for panoramic view for reference
  • Step A3 assigning the lens image of the panoramic view for reference in the step A2 to the original mesh model in the step A1;
  • Step A4 calculating, according to the orientation image, the field of view angle, the distortion parameter, the image eccentricity parameter, and the displacement parameter of the captured lens, according to the lens image of the panoramic view for reference in step A3, for reference
  • Step A5 taking out the lens image for panoramic view in the step A3, and saving the new mesh model in the step A4.
  • step B includes the following steps:
  • Step B l preparing a lens image to be expanded
  • Step B2 the lens image to be expanded in the step B1 is assigned to the new mesh model in the step A5;
  • Step B3 the new mesh model in the step B2 is rendered by the GPU, and a panoramic image corresponding to the to-be-expanded lens image is obtained;
  • Step B4 storing the panoramic image in the step B3.
  • the shape of the original mesh model may be a rectangle, a sphere, a 360° ring view, a ring, a hemisphere or other planar geometric or curved solid geometry.
  • the lens image may be a fisheye lens image, a wide-angle lens image or other forms of lens image.
  • the present invention provides an apparatus for expanding a lens image into a panoramic image, including:
  • a new mesh model generating module configured to store the original mesh model and calculate new map coordinate values, for the original The original texture coordinate value of the initial mesh model and the coordinate value of the new texture coordinate value obtained by the calculation are used to generate a new mesh model and save;
  • a panoramic image acquisition module configured to store a lens image to be displayed, and assign the lens image to be displayed to a new mesh model in the new mesh model generation module, and The new mesh model is rendered, the panoramic image is acquired and saved, and it is connected to the new mesh model generating module.
  • the new mesh model generation module includes:
  • an original mesh model storage unit configured to store the original mesh model
  • the lens image storage unit for panoramic view is used for storing a lens image for panoramic view for reference, which is connected to the original mesh model storage unit;
  • a lens image first giving unit configured to assign, to the original mesh model, a lens image for viewing in the lens image storage unit for reference In the original mesh model in the unit, it is connected to the lens image storage unit of the panoramic view for reference;
  • a mesh model map coordinate processing unit configured to capture a heading parameter, an angle of view, a distortion parameter, and an image of the lens according to the lens image for the reference in the first image giving unit of the lens image
  • the eccentricity parameter and the displacement parameter calculate a new texture coordinate value of each vertex in the original mesh model in which the panoramic lens image is displayed for reference, and the original of each vertex in the original mesh model
  • the texture coordinate values are all changed to the new map coordinate values corresponding thereto, and a new mesh model is generated, which is connected to the first image giving unit of the lens image;
  • a new mesh model storage unit configured to take out a lens image for panoramic view in the first image giving unit of the lens image and save the new network generated in the mesh model mapping coordinate processing unit a grid model that is coupled to the grid model map coordinate processing unit.
  • the panoramic image acquisition module includes:
  • a lens image second giving unit configured to assign the lens image to be expanded in the lens image storage unit to be displayed to a new mesh model in the new mesh model storage unit, Connected to the lens image storage unit to be displayed;
  • a new mesh model rendering unit configured to render a new mesh model in the second image giving unit of the lens image by using a GPU, to obtain a panoramic image corresponding to the lens image to be expanded, and a The second image giving unit of the lens image is connected;
  • a panoramic image storage unit configured to store the panoramic image in the new mesh model rendering unit, which is connected to the new mesh model rendering unit.
  • the shape of the original mesh model may be a rectangle, a sphere, a 360° ring view, a ring, a hemisphere or other planar geometric or curved solid geometry.
  • the lens image may be a fisheye lens image, a wide-angle lens image, or other form of lens image.
  • the present invention assigns a lens image to be expanded as a texture map to a new mesh model in which a new map coordinate value is pre-computed and changed, and a new mesh model is performed by the GPU.
  • the lens image can be quickly expanded into a panoramic image, which greatly reduces the cost and energy consumption of the panoramic device, and greatly reduces the volume.
  • FIG. 1 is a schematic flow chart of a method for expanding a lens image into a panoramic image according to an embodiment of the present invention
  • FIG. 2 is a schematic flow chart of a method for generating a new mesh model in FIG. 1;
  • FIG. 3 is a schematic flow chart of a method for acquiring a panoramic image in FIG. 1;
  • FIG. 5 is a fisheye image of a panoramic view for reference according to an embodiment of the present invention.
  • FIG. 6 is a fisheye image to be displayed by a fisheye lens camera according to an embodiment of the present invention
  • FIG. 7 is a schematic diagram of a new rectangular mesh model provided by an embodiment of the present invention. The obtained panoramic image
  • FIG. 8 is a schematic structural diagram of an apparatus for expanding a lens image into a panoramic image according to an embodiment of the present invention.
  • FIG. 1 is a schematic flowchart of a method for expanding a lens image into a panoramic image according to an embodiment of the present invention.
  • a method for expanding a lens image into a panoramic image includes: The following steps:
  • Step S20 preparing an original mesh model, calculating a new texture coordinate value, performing coordinate value conversion on the original texture coordinate value of the original mesh model and the calculated new texture coordinate value, and generating a new network.
  • Step S21 preparing a lens image to be expanded, assigning the lens image to be expanded to the new mesh model in step S20, and performing the new mesh model through the GPU. Render, get a panoramic image and save it.
  • FIG. 2 is a schematic flowchart of a method for generating a new mesh model in FIG. 1, as shown in FIG.
  • the method of generating a new mesh model includes the following steps:
  • Step S201 preparing an original mesh model
  • step S202 preparing a lens image for panoramic view for reference
  • Step S203 the lens image for zooming in the step S202 is given to the original mesh model in the step S201;
  • Step S204 calculating, according to the orientation image, the angle of view, the distortion parameter, the image eccentricity parameter and the displacement parameter of the captured lens, according to the lens image of the panoramic view for reference in step S203, for reference
  • Step S205 the lens image for the reference in the step S203 is taken out, and the new mesh model in the step S204 is saved.
  • FIG. 3 is a schematic flowchart of a method for acquiring a panoramic image in FIG. 1. As shown in FIG. 3, the method for obtaining a panoramic image includes the following steps:
  • Step S211 preparing a lens image to be expanded
  • Step S212 the lens image to be expanded in step S211 is assigned to the new mesh model in step S205;
  • Step S213 the new mesh model in the step S212 is rendered by the GPU, and a panoramic image corresponding to the to-be-expanded lens image is obtained;
  • Step S214 storing the panoramic image in the step S213.
  • a rectangular original mesh model with an aspect ratio of 2:1 is fabricated in 3d S Ma X or other modeling tools.
  • the rectangular original mesh model is composed of multiple square meshes, so the rectangular original mesh The ratio of the length to the width of the model is also 2:1. Under normal circumstances, the denser the mesh in the model, the higher the quality of the panoramic image, the better the effect, as shown in Figure 4, a length of 50 squares.
  • Grid a rectangular original mesh model consisting of 25 square grids;
  • FIG. 5 is a fisheye image of a panoramic view for reference
  • each mesh in the rectangular original mesh model is It consists of four vertices.
  • the vertices of each mesh contain the position coordinates of the mesh ( ⁇ , ⁇ , ⁇ ) and the original texture coordinates (U, V). We need to use the reference for this reference.
  • the new map coordinate value of each vertex in the rectangular original mesh model by the orientation parameter, the angle of view, the distortion parameter, the image eccentricity parameter and the displacement parameter of the panoramic fisheye image, since the new map coordinate value determines the
  • the vertex uses which pixels of the fisheye image are attached to the grid composed of the vertex, so the original texture coordinate value of each vertex in the rectangular original mesh model is changed to the corresponding new map coordinate value.
  • the original texture coordinate value of the vertex is (0, 0).
  • the new texture coordinate value corresponding to the point is (0.1, 0.3), so We put the top
  • the original texture coordinate value (0, 0) is changed to the new texture coordinate value (0.1, 0.3), and the new texture coordinate value of each vertex in the above rectangular original mesh model is set according to this method, generating a new Rectangular mesh model and save;
  • FIG. 6 is a fisheye image to be displayed by a fisheye lens camera, and the fish to be displayed The eye image is assigned to the new rectangular mesh model;
  • the new rectangular mesh model is rendered by the GPU, and a panoramic image corresponding to the fisheye image is obtained and saved. As shown in FIG. 7, the panoramic image obtained by rendering the new rectangular mesh model is obtained.
  • the new mesh model for calculating and changing the coordinate values of the new map can be saved, and can be directly used next time.
  • FIG. 8 is a schematic structural diagram of an apparatus for expanding a lens image into a panoramic image according to an embodiment of the present invention.
  • the apparatus for expanding a lens image into a panoramic image includes :
  • a new mesh model generating module 30 configured to store the original mesh model and calculate a new map coordinate value, and perform coordinates on the original texture coordinate value of the original mesh model and the calculated new texture coordinate value Value transformation, generate a new mesh model and save it;
  • the panoramic image acquisition module 31 is configured to store a lens image to be displayed, and assign the lens image to be expanded to a new mesh model in the new mesh model generation module 30, and pass the GPU.
  • the new mesh model is rendered, a panoramic image is acquired and saved, which is coupled to the new mesh model generation module 30.
  • the new mesh model generation module 30 includes:
  • the original mesh model storage unit 301 is configured to store the original mesh model
  • the lens image storage unit 302 for panoramic view is used for storing a lens image for panoramic view for reference, which is connected to the original mesh model storage unit 301;
  • the lens image first giving unit 303 is configured to assign the lens image for panoramic view in the lens image storage unit 302 for reference to the panoramic view to the original mesh In the original mesh model in the model storage unit 301, it is connected to the lens image storage unit 302 which is a panoramic view for reference;
  • the mesh model map coordinate processing unit 304 is configured to capture the orientation parameter, the angle of view, and the distortion parameter of the lens according to the lens image of the lens for the reference in the lens image first giving unit 303. And an image eccentricity parameter and a displacement parameter calculate a new texture coordinate value of each vertex in the original mesh model in which the panoramic lens image is displayed for reference, and each vertex in the original mesh model The original map coordinate values are all changed to their corresponding new map coordinate values, and a new mesh model is generated, which is connected to the lens image first assigning unit 303; [0076]
  • the new mesh model storage unit 305 is configured to take out the lens image of the lens image first giving unit 303 for reference and to save it in the mesh model map coordinate processing unit 304. a new mesh model that is coupled to the mesh model map coordinate processing unit 304.
  • the panoramic image acquisition module 31 includes:
  • the lens image storage unit 311 is configured to store a lens image to be displayed, which is connected to the new mesh model storage unit 305.
  • the lens image second giving unit 312 is configured to assign the lens image to be expanded in the lens image storage unit 311 to be expanded to a new mesh model in the new mesh model storage unit 305. Inside, it is connected to the lens image storage unit 311 to be expanded;
  • a new mesh model rendering unit 313, configured to render a new mesh model in the lens image second assigning unit 312 by using a GPU, to obtain a panoramic image corresponding to the lens image to be expanded, Connected to the lens image second giving unit 312;
  • the panoramic image storage unit 314 is configured to store the panoramic image in the new mesh model rendering unit 313, which is connected to the new mesh model rendering unit 313.
  • the shape of the original mesh model and its corresponding new mesh model may be a rectangle, a sphere, a 360° ring view, a ring, a hemisphere or other planar geometric or curved solid geometry;
  • the image can be a fisheye lens image, a wide-angle lens image, or other form of lens image.
  • the present invention can quickly realize that a lens image to be displayed is given as a texture map to a new mesh model in which a new map coordinate value is pre-calculated and changed, and the new mesh model is rendered by the GPU.
  • the lens image is displayed as a panoramic image, which greatly reduces the cost and energy consumption of the panoramic device, and greatly reduces the volume.
  • we can also realize the production of high-definition and real-time panoramic video.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Geometry (AREA)
  • Computer Graphics (AREA)
  • Studio Devices (AREA)
  • Image Processing (AREA)

Abstract

一种将镜头图像展开为全景图像的方法及装置,包括:步骤A,准备一个原始网格模型,计算新贴图坐标值,对原始网格模型的原始贴图坐标值和计算得到的新贴图坐标值进行坐标值变换,生成新网格模型;步骤B,准备一张待展开的镜头图像,将待展开的镜头图像赋予到步骤A中的新网格模型中,通过GPU对新网格模型进行渲染,获取全景图像。通过将待展开的镜头图像赋予到预先计算和更改好新贴图坐标值的新网格模型中,并通过GPU来对新网格模型进行渲染,可实现快速地将镜头图像展开为全景图像,使全景设备的成本、能耗大大降低,体积大大减小,同时,通过这种方法我们还可以实现高清、实时全景视频的制作。

Description

一种将镜头图像展幵为全景图像的方法及装置 技术领域
[0001] 本发明属于图像处理技术领域, 特别是涉及一种将镜头图像展幵为全景图像的 方法及装置。
背景技术
[0002] 随着 VR技术 (虚拟现实技术)的快速发展, 人们对 VR内容 (虚拟现实内容)的需 求也愈加强烈, 全景图像以及全景视频是 VR内容的重要组成部分之一, 目前的 全景图像一般是由一个大角度鱼眼镜头相机拍摄, 或是由几个广角或鱼眼相机 拍摄并将拍摄后的图像拼接而成, 由于广角和鱼眼相机拍摄的图像都会带有畸 变, 所以需要通过算法将广角或鱼眼相机拍摄的图像展幵为全景图像。 目前的 全景图像展幵算法大多运行在 CPU (中央处理器), 少数全景图像展幵算法运行在 GPU(Graphic Processing Unit, 图形处理器, 以下简称 GPU), 由于全景图像展幵 算法计算量很大, 展幵一帧图像需要的吋间很长, 一般都采用成本、 能耗比较 高、 体积比较大的高性能 CPU, 然而 CPU却很难做到高清的或是实吋的图像展 幵, 不能实现全景直播。
技术问题
[0003] 综上所述, 为解决上述技术问题, 本发明提供了一种将镜头图像展幵为全景图 像的方法和装置, 以实现快速地将镜头图像展幵为全景图像, 降低全景设备的 成本、 能耗和减小全景设备的体积。
问题的解决方案
技术解决方案
[0004] 第一方面, 本发明提供了一种将镜头图像展幵为全景图像的方法, 包括下述步 骤:
[0005] 步骤 A, 准备一个原始网格模型, 计算新贴图坐标值, 对所述原始网格模型的 原始贴图坐标值和通过计算得到的所述新贴图坐标值进行坐标值变换, 生成新 网格模型并保存; [0006] 步骤 B, 准备一张待展幵的镜头图像,将所述待展幵的镜头图像赋予到所述步骤 A中的新网格模型中, 并通过 GPU对所述新网格模型进行渲染, 获取全景图像 并保存。
[0007] 进一步地, 所述步骤 A包括下述步骤:
[0008] 步骤 Al, 准备一个原始网格模型;
[0009] 步骤 A2, 准备一张用于参考的展幵为全景的镜头图像;
[0010] 步骤 A3, 将所述步骤 A2中的用于参考的展幵为全景的镜头图像赋予到所述步 骤 A1中的原始网格模型中;
[0011] 步骤 A4, 根据所述步骤 A3中用于参考的展幵为全景的镜头图像在拍摄吋镜头 的朝向参数、 视场角、 畸变参数、 图像偏心参数和位移参数计算所述用于参考 的展幵为全景的镜头图像所在的原始网格模型中的每个顶点的新贴图坐标值, 并将所述原始网格模型中每个顶点的原始贴图坐标值都变更为其所对应的新贴 图坐标值, 生成新网格模型;
[0012] 步骤 A5, 取出所述步骤 A3中的用于参考的展幵为全景的镜头图像, 并保存所 述步骤 A4中的新网格模型。
[0013] 进一步地, 所述步骤 B包括下述步骤:
[0014] 步骤 B l, 准备一张待展幵的镜头图像;
[0015] 步骤 B2, 将所述步骤 B1中的待展幵的镜头图像赋予到所述步骤 A5的新网格 模型中;
[0016] 步骤 B3, 通过 GPU对所述步骤 B2中的新网格模型进行渲染, 得到与所述待 展幵的镜头图像对应的全景图像;
[0017] 步骤 B4, 对所述步骤 B3中的全景图像进行存储。
[0018] 更进一步地, 所述原始网格模型的形状可以为矩形、 球体、 360°环景、 圆环、 半球或其他平面几何图形或曲面立体几何图形。
[0019] 再进一步地, 其特征在于, 所述镜头图像可以为鱼眼镜头图像、 广角镜头图像 或其他形式的镜头图像。
[0020] 第二方面, 本发明提供了一种将镜头图像展幵为全景图像的装置, 包括:
[0021] 新网格模型生成模块, 用于存放原始网格模型和计算新贴图坐标值, 对所述原 始网格模型的原始贴图坐标值和通过计算得到的所述新贴图坐标值进行坐标值 变换, 生成新网格模型并保存;
[0022] 全景图像获取模块, 用于存放待展幵的镜头图像, 将所述待展幵的镜头图像赋 予到所述新网格模型生成模块中的新网格模型中, 并通过 GPU对所述新网格模 型进行渲染, 获取全景图像并保存, 其与所述新网格模型生成模块相连接。
[0023] 进一步地, 所述新网格模型生成模块包括:
[0024] 原始网格模型存放单元, 用于存放原始网格模型;
[0025] 用于参考的展幵为全景的镜头图像存放单元, 用于存放用于参考的展幵为全景 的镜头图像, 其与所述原始网格模型存放单元相连接;
[0026] 镜头图像第一赋予单元, 用于将所述用于参考的展幵为全景的镜头图像存放单 元中的用于参考的展幵为全景的镜头图像赋予到所述原始网格模型存放单元中 的原始网格模型中, 其与所述用于参考的展幵为全景的镜头图像存放单元相连 接;
[0027] 网格模型贴图坐标处理单元, 用于根据所述镜头图像第一赋予单元中用于参考 的展幵为全景的镜头图像在拍摄吋镜头的朝向参数、 视场角、 畸变参数、 图像 偏心参数和位移参数计算所述用于参考的展幵为全景的镜头图像所在的原始网 格模型中的每个顶点的新贴图坐标值, 并将所述原始网格模型中每个顶点的原 始贴图坐标值都变更为其所对应的新贴图坐标值, 生成新网格模型, 其与所述 镜头图像第一赋予单元相连接;
[0028] 新网格模型存放单元, 用于取出所述镜头图像第一赋予单元中的用于参考的展 幵为全景的镜头图像并保存所述网格模型贴图坐标处理单元中生成的新网格模 型, 其与所述网格模型贴图坐标处理单元相连接。
[0029] 进一步地, 所述全景图像获取模块包括:
[0030] 待展幵的镜头图像存放单元, 用于存放待展幵的镜头图像, 其与所述新网格模 型存放单元相连接;
[0031] 镜头图像第二赋予单元, 用于将所述待展幵的镜头图像存放单元中的待展幵的 镜头图像赋予到所述新网格模型存放单元中的新网格模型内, 其与所述待展幵 的镜头图像存放单元相连接; [0032] 新网格模型渲染单元, 用于通过 GPU对所述镜头图像第二赋予单元中的新网 格模型进行渲染, 得到与所述待展幵的镜头图像对应的全景图像, 其与所述镜 头图像第二赋予单元相连接;
[0033] 全景图像存放单元, 用于将所述新网格模型渲染单元中的所述全景图像进行存 储, 其与所述新网格模型渲染单元相连接。
[0034] 更进一步地, 所述原始网格模型的形状可以为矩形、 球体、 360°环景、 圆环、 半球或其他平面几何图形或曲面立体几何图形。
[0035] 再进一步地, 所述镜头图像可以为鱼眼镜头图像、 广角镜头图像或其他形式的 镜头图像。
发明的有益效果
有益效果
[0036] 与现有技术相比, 本发明通过将待展幵的镜头图像作为材质贴图赋予到预先计 算和更改好新贴图坐标值的新网格模型中, 并通过 GPU来对新网格模型进行渲 染, 可实现快速地将镜头图像展幵为全景图像, 使全景设备的成本、 能耗大大 降低, 体积大大减小, 同吋, 通过这种方法我们还可以实现高清、 实吋全景视 频的制作。
对附图的简要说明
附图说明
[0037] 图 1为本发明实施例提供的一种将镜头图像展幵为全景图像的方法的流程示意 图;
[0038] 图 2为图 1中生成新网格模型的方法的流程示意图;
[0039] 图 3为图 1中获取全景图像的方法的流程示意图;
[0040] 图 4为本发明实施例提供的一个矩形原始网格模型;
[0041] 图 5为本发明实施例提供的一张用于参考的展幵为全景的鱼眼图像;
[0042] 图 6为本发明实施例提供的由鱼眼镜头相机拍摄的一张待展幵的鱼眼图像; [0043] 图 7为本发明实施例提供的新的矩形网格模型通过渲染后得到的全景图像; [0044] 图 8为本发明实施例提供的一种将镜头图像展幵为全景图像的装置的结构示意 图。 本发明的实施方式
[0045] 为了使本发明的目的、 技术方案及优点更加清楚明白, 以下结合附图及实施例 , 对本发明进行进一步详细说明。 应当理解, 此处所描述的具体实施例仅仅用 以解释本发明, 并不用于限定本发明。
[0046] 以下结合具体实施例对本发明的实现进行详细的描述。
[0047] 请参阅图 1, 图 1是本发明实施例提供的一种将镜头图像展幵为全景图像的方 法的流程示意图, 如图 1所示, 将镜头图像展幵为全景图像的方法包括以下步 骤:
[0048] 步骤 S20, 准备一个原始网格模型, 计算新贴图坐标值, 对所述原始网格模型 的原始贴图坐标值和通过计算得到的所述新贴图坐标值进行坐标值变换, 生成 新网格模型并保存;
[0049] 步骤 S21, 准备一张待展幵的镜头图像, 将所述待展幵的镜头图像赋予到所述 步骤 S20中的新网格模型中, 并通过 GPU对所述新网格模型进行渲染, 获取全 景图像并保存。
[0050] 请参阅图 2, 图 2为图 1中生成新网格模型的方法的流程示意图, 如图 2所示
, 生成新网格模型的方法包括下述步骤:
[0051] 步骤 S201 , 准备一个原始网格模型;
[0052] 步骤 S202, 准备一张用于参考的展幵为全景的镜头图像;
[0053] 步骤 S203 , 将所述步骤 S202中的用于参考的展幵为全景的镜头图像赋予到所 述步骤 S201中的原始网格模型中;
[0054] 步骤 S204, 根据所述步骤 S203中用于参考的展幵为全景的镜头图像在拍摄吋 镜头的朝向参数、 视场角、 畸变参数、 图像偏心参数和位移参数计算所述用于 参考的展幵为全景的镜头图像所在的原始网格模型中的每个顶点的新贴图坐标 值, 并将所述原始网格模型中每个顶点的原始贴图坐标值都变更为其所对应的 新贴图坐标值, 生成新网格模型;
[0055] 步骤 S205 , 取出所述步骤 S203中的用于参考的展幵为全景的镜头图像, 并保 存所述步骤 S204中的新网格模型。 [0056] 请参阅图 3, 图 3为图 1中获取全景图像的方法的流程示意图, 如图 3所示, 获取全景图像的方法包括下述步骤:
[0057] 步骤 S211, 准备一张待展幵的镜头图像;
[0058] 步骤 S212, 将所述步骤 S211中的待展幵的镜头图像赋予到所述步骤 S205的 新网格模型中;
[0059] 步骤 S213 , 通过 GPU对所述步骤 S212中的新网格模型进行渲染, 得到与所 述待展幵的镜头图像对应的全景图像;
[0060] 步骤 S214, 对所述步骤 S213中的全景图像进行存储。
[0061] 以一个矩形原始网格模型和鱼眼镜头相机拍摄的一张鱼眼图像为例:
[0062] 首先在 3dSMaX或其他建模工具中制作一个长宽比为 2:1的矩形原始网格模型 , 该矩形原始网格模型由多个正方形网格构成, 所以矩形原始网格模型的长宽 分段的比也是 2:1, 正常情况下, 模型中的网格越密集, 展幵的全景图像质量越 高, 效果越好, 如图 4所示为一个长由 50个正方形网格, 宽由 25个正方形网 格构成的矩形原始网格模型;
[0063] 如图 5所示为一张用于参考的展幵为全景的鱼眼图像;
[0064] 将图 5所示的用于参考的展幵为全景的鱼眼图像作为贴图赋予到图 4所示的矩 形原始网格模型中, 该矩形原始网格模型中的每个网格都是由四个顶点所组成 的, 每个网格的顶点内含有网格的位置坐标 (Χ, Υ, Ζ)和原始贴图坐标 (U, V) , 我们需要根据该用于参考的展幵为全景的鱼眼图像的朝向参数、 视场角、 畸变 参数、 图像偏心参数和位移参数等来计算该矩形原始网格模型中的每个顶点的 新贴图坐标值, 由于新贴图坐标值决定了该顶点使用鱼眼图像的哪些像素来贴 到由该顶点组成的网格上, 所以将该矩形原始网格模型中的每个顶点的原始贴 图坐标值都更改为其所对应的新贴图坐标值, 以位置坐标为 (15, 9)的顶点为例 , 该顶点的原始贴图坐标值为 (0, 0), 通过计算, 可以得出该点对应的新贴图坐 标值为 (0.1, 0.3), 于是我们把该顶点的原始贴图坐标值 (0, 0)更改为新贴图坐标 值 (0.1, 0.3) , 并将上述矩形原始网格模型中的每个顶点的新贴图坐标值都按 照这种方法设置好, 生成新的矩形网格模型并保存;
[0065] 如图 6所示为由鱼眼镜头相机拍摄的一张待展幵的鱼眼图像, 将该待展幵的鱼 眼图像赋予到新的矩形网格模型中;
[0066] 通过 GPU对新的矩形网格模型进行渲染, 得到与该鱼眼图像对应的全景图像 并保存, 如图 7所示为新的矩形网格模型通过渲染后得到的全景图像。
[0067] 本发明实施例中, 计算和更改好新贴图坐标值的新网格模型可以保存, 下次可 以直接使用。
[0068] 请参阅图 8, 图 8是本发明实施例提供的一种将镜头图像展幵为全景图像的装 置的结构示意图, 如图 8所示, 将镜头图像展幵为全景图像的装置包括:
[0069] 新网格模型生成模块 30, 用于存放原始网格模型和计算新贴图坐标值, 对所述 原始网格模型的原始贴图坐标值和通过计算得到的所述新贴图坐标值进行坐标 值变换, 生成新网格模型并保存;
[0070] 全景图像获取模块 31, 用于存放待展幵的镜头图像, 将所述待展幵的镜头图像 赋予到所述新网格模型生成模块 30中的新网格模型中, 并通过 GPU对所述新 网格模型进行渲染, 获取全景图像并保存, 其与所述新网格模型生成模块 30相 连接。
[0071] 本发明实施例中, 新网格模型生成模块 30包括:
[0072] 原始网格模型存放单元 301, 用于存放原始网格模型;
[0073] 用于参考的展幵为全景的镜头图像存放单元 302, 用于存放用于参考的展幵为 全景的镜头图像, 其与原始网格模型存放单元 301相连接;
[0074] 镜头图像第一赋予单元 303, 用于将所述用于参考的展幵为全景的镜头图像存 放单元 302中的用于参考的展幵为全景的镜头图像赋予到所述原始网格模型存 放单元 301中的原始网格模型中, 其与所述用于参考的展幵为全景的镜头图像存 放单元 302相连接;
[0075] 网格模型贴图坐标处理单元 304, 用于根据所述镜头图像第一赋予单元 303中 用于参考的展幵为全景的镜头图像在拍摄吋镜头的朝向参数、 视场角、 畸变参 数、 图像偏心参数和位移参数计算所述用于参考的展幵为全景的镜头图像所在 的原始网格模型中的每个顶点的新贴图坐标值, 并将所述原始网格模型中每个 顶点的原始贴图坐标值都变更为其所对应的新贴图坐标值, 生成新网格模型, 其与所述镜头图像第一赋予单元 303相连接; [0076] 新网格模型存放单元 305, 用于取出所述镜头图像第一赋予单元 303中的用于 参考的展幵为全景的镜头图像并保存所述网格模型贴图坐标处理单元 304中生 成的新网格模型, 其与所述网格模型贴图坐标处理单元 304相连接。
[0077] 全景图像获取模块 31包括:
[0078] 待展幵的镜头图像存放单元 311, 用于存放待展幵的镜头图像, 其与所述新网 格模型存放单元 305相连接;
[0079] 镜头图像第二赋予单元 312, 用于将所述待展幵的镜头图像存放单元 311中的 待展幵的镜头图像赋予到所述新网格模型存放单元 305中的新网格模型内, 其 与所述待展幵的镜头图像存放单元 311相连接;
[0080] 新网格模型渲染单元 313, 用于通过 GPU对所述镜头图像第二赋予单元 312中 的新网格模型进行渲染, 得到与所述待展幵的镜头图像对应的全景图像, 其与 所述镜头图像第二赋予单元 312相连接;
[0081] 全景图像存放单元 314, 用于将所述新网格模型渲染单元 313中的所述全景图 像进行存储, 其与所述新网格模型渲染单元 313相连接。
[0082] 本发明实施例中, 原始网格模型和其对应的新网格模型的形状可以为矩形、 球 体、 360°环景、 圆环、 半球或其他平面几何图形或曲面立体几何图形; 镜头图 像可以为鱼眼镜头图像、 广角镜头图像或其他形式的镜头图像。
[0083] 本发明通过将待展幵的镜头图像作为材质贴图赋予到预先计算和更改好新贴图 坐标值的新网格模型中, 并通过 GPU来对新网格模型进行渲染, 可实现快速地 将镜头图像展幵为全景图像, 使全景设备的成本、 能耗大大降低, 体积大大减 小, 同吋, 通过这种方法我们还可以实现高清、 实吋全景视频的制作。
[0084] 以上所述仅为本发明的较佳实施例而已, 并不用以限制本发明, 凡在本发明的 精神和原则之内所作的任何修改、 等同替换和改进等, 均应包含在本发明的保 护范围之内。

Claims

权利要求书
[权利要求 1] 一种将镜头图像展幵为全景图像的方法, 其特征在于, 包括下述步骤 步骤 A, 准备一个原始网格模型, 计算新贴图坐标值, 对所述原始网 格模型的原始贴图坐标值和通过计算得到的所述新贴图坐标值进行 坐标值变换, 生成新网格模型并保存;
步骤 B, 准备一张待展幵的镜头图像, 将所述待展幵的镜头图像赋予 到所述步骤 A中的新网格模型中, 并通过 GPU对所述新网格模型进 行渲染, 获取全景图像并保存。
[权利要求 2] 如权利要求 1所述的将镜头图像展幵为全景图像的方法, 其特征在于
, 所述步骤 A包括下述步骤:
步骤 Al, 准备一个原始网格模型;
步骤 A2, 准备一张用于参考的展幵为全景的镜头图像;
步骤 A3, 将所述步骤 A2中的用于参考的展幵为全景的镜头图像赋 予到所述步骤 A1中的原始网格模型中;
步骤 A4, 根据所述步骤 A3中用于参考的展幵为全景的镜头图像在 拍摄吋镜头的朝向参数、 视场角、 畸变参数、 图像偏心参数和位移 参数计算所述用于参考的展幵为全景的镜头图像所在的原始网格模型 中的每个顶点的新贴图坐标值, 并将所述原始网格模型中每个顶点的 原始贴图坐标值都变更为其所对应的新贴图坐标值, 生成新网格模型 步骤 A5, 取出所述步骤 A3中的用于参考的展幵为全景的镜头图像 , 并保存所述步骤 A4中的新网格模型。
[权利要求 3] 如权利要求 2所述的将镜头图像展幵为全景图像的方法, 其特征在于
, 所述步骤 B包括下述步骤:
步骤 B l, 准备一张待展幵的镜头图像;
步骤 B2, 将所述步骤 B1中的待展幵的镜头图像赋予到所述步骤 A5 的新网格模型中; 步骤 B3, 通过 GPU对所述步骤 B2中的新网格模型进行渲染, 得到 与所述待展幵的镜头图像对应的全景图像;
步骤 B4, 对所述步骤 B3中的全景图像进行存储。
[权利要求 4] 如权利要求 1-2任一项所述的将镜头图像展幵为全景图像的方法, 其 特征在于, 所述原始网格模型的形状可以为矩形、 球体、 360°环景、 圆环、 半球或其他平面几何图形或曲面立体几何图形。
[权利要求 5] 如权利要求 1-4任一项所述的将镜头图像展幵为全景图像的方法, 其 特征在于, 所述镜头图像可以为鱼眼镜头图像、 广角镜头图像或其他 形式的镜头图像。
[权利要求 6] —种将镜头图像展幵为全景图像的装置, 其特征在于, 包括:
新网格模型生成模块, 用于存放原始网格模型和计算新贴图坐标值, 对所述原始网格模型的原始贴图坐标值和通过计算得到的所述新贴图 坐标值进行坐标值变换, 生成新网格模型并保存; 全景图像获取模块, 用于存放待展幵的镜头图像, 将所述待展幵的镜 头图像赋予到所述新网格模型生成模块中的新网格模型中, 并通过 GPU对所述新网格模型进行渲染, 获取全景图像并保存, 其与所述 新网格模型生成模块相连接。
[权利要求 7] 如权利要求 6所述的将镜头图像展幵为全景图像的装置, 其特征在于
, 所述新网格模型生成模块包括:
原始网格模型存放单元, 用于存放原始网格模型; 用于参考的展幵为全景的镜头图像存放单元, 用于存放用于参考的展 幵为全景的镜头图像, 其与所述原始网格模型存放单元相连接; 镜头图像第一赋予单元, 用于将所述用于参考的展幵为全景的镜头图 像存放单元中的用于参考的展幵为全景的镜头图像赋予到所述原始网 格模型存放单元中的原始网格模型中, 其与所述用于参考的展幵为全 景的镜头图像存放单元相连接;
网格模型贴图坐标处理单元, 用于根据所述镜头图像第一赋予单元中 用于参考的展幵为全景的镜头图像在拍摄吋镜头的朝向参数、 视场角 、 畸变参数、 图像偏心参数和位移参数计算所述用于参考的展幵为全 景的镜头图像所在的原始网格模型中的每个顶点的新贴图坐标值, 并 将所述原始网格模型中每个顶点的原始贴图坐标值都变更为其所对应 的新贴图坐标值, 生成新网格模型, 其与所述镜头图像第一赋予单元 相连接;
新网格模型存放单元, 用于取出所述镜头图像第一赋予单元中的用于 参考的展幵为全景的镜头图像并保存所述网格模型贴图坐标处理单元 中生成的新网格模型, 其与所述网格模型贴图坐标处理单元相连接。
[权利要求 8] 如权利要求 6所述的将镜头图像展幵为全景图像的装置, 其特征在于
, 所述全景图像获取模块包括:
待展幵的镜头图像存放单元, 用于存放待展幵的镜头图像, 其与所述 新网格模型存放单元相连接;
镜头图像第二赋予单元, 用于将所述待展幵的镜头图像存放单元中的 待展幵的镜头图像赋予到所述新网格模型存放单元中的新网格模型内 , 其与所述待展幵的镜头图像存放单元相连接; 新网格模型渲染单元, 用于通过 GPU对所述镜头图像第二赋予单元 中的新网格模型进行渲染, 得到与所述待展幵的镜头图像对应的全景 图像, 其与所述镜头图像第二赋予单元相连接; 全景图像存放单元, 用于将所述新网格模型渲染单元中的所述全景图 像进行存储, 其与所述新网格模型渲染单元相连接。
[权利要求 9] 如权利要求 6-7任一项所述的将镜头图像展幵为全景图像的装置, 其 特征在于, 所述原始网格模型的形状可以为矩形、 球体、 360°环景、 圆环、 半球或其他平面几何图形或曲面立体几何图形。
[权利要求 10] 如权利要求 6-9任一项所述的将镜头图像展幵为全景图像的装置, 其 特征在于, 所述镜头图像可以为鱼眼镜头图像、 广角镜头图像或其他 形式的镜头图像。
PCT/CN2017/079093 2016-05-11 2017-03-31 一种将镜头图像展开为全景图像的方法及装置 Ceased WO2017193729A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
US16/309,806 US10798300B2 (en) 2016-05-11 2017-03-31 Method and device for unfolding lens image into panoramic image
KR1020187035978A KR102206899B1 (ko) 2016-05-11 2017-03-31 렌즈 이미지를 파노라마 이미지로 전개하는 방법 및 장치
EP17795347.8A EP3457681A4 (en) 2016-05-11 2017-03-31 METHOD AND DEVICE FOR DEVELOPING A LENS PICTURE INTO A PANORAMIC PICTURE

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610309937.6A CN106993126B (zh) 2016-05-11 2016-05-11 一种将镜头图像展开为全景图像的方法及装置
CN201610309937.6 2016-05-11

Publications (1)

Publication Number Publication Date
WO2017193729A1 true WO2017193729A1 (zh) 2017-11-16

Family

ID=59414053

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/079093 Ceased WO2017193729A1 (zh) 2016-05-11 2017-03-31 一种将镜头图像展开为全景图像的方法及装置

Country Status (5)

Country Link
US (1) US10798300B2 (zh)
EP (1) EP3457681A4 (zh)
KR (1) KR102206899B1 (zh)
CN (1) CN106993126B (zh)
WO (1) WO2017193729A1 (zh)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106254792B (zh) * 2016-07-29 2019-03-12 暴风集团股份有限公司 基于Stage3D播放全景数据的方法及系统
CN110728619B (zh) * 2018-07-17 2024-03-22 中科创达软件股份有限公司 一种全景图像拼接渲染方法及装置
CN108965718B (zh) * 2018-08-03 2021-03-23 北京微播视界科技有限公司 图像生成方法和装置
CN110838163B (zh) * 2018-08-15 2024-02-02 浙江宇视科技有限公司 贴图处理方法及装置
CN112967173B (zh) * 2019-12-12 2023-05-26 杭州海康威视数字技术股份有限公司 一种图像生成方法、装置及系统
CN114784199B (zh) * 2022-04-22 2025-03-07 吉林大学 一种具有广角探测成像能力的窄带光探测器及其制备方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009068942A1 (en) * 2007-11-30 2009-06-04 Dooworks Fz Co Method and system for processing of images
CN101777193A (zh) * 2010-02-03 2010-07-14 中国人民解放军国防科学技术大学 基于双向映射精确插值的折反射全向图柱面全景展开方法
CN103839227A (zh) * 2012-11-23 2014-06-04 富士通株式会社 鱼眼图像校正方法和装置
CN104240236A (zh) * 2014-08-26 2014-12-24 中山大学 一种鱼眼镜头标定后鱼眼图像校正的方法

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002203254A (ja) * 2000-08-30 2002-07-19 Usc Corp 曲面像変換方法及びこの曲面像変換方法を記録した記録媒体
US7336299B2 (en) * 2003-07-03 2008-02-26 Physical Optics Corporation Panoramic video system with real-time distortion-free imaging
CN101072288A (zh) * 2007-06-15 2007-11-14 林进灯 取得鱼眼影像校正关系与鱼眼校正的方法
JP2011217057A (ja) * 2010-03-31 2011-10-27 Asahi Kasei Corp レンズ歪み除去装置、方法およびプログラム
CN102780834A (zh) * 2011-05-11 2012-11-14 张茂军 环眼镜头图像半柱面全景展开方法
CN103827917B (zh) * 2011-07-25 2017-06-09 科英布拉大学 用于使用棋盘图案的一幅或多幅图像的自动相机校准的方法和装置
KR20140100656A (ko) * 2013-02-06 2014-08-18 한국전자통신연구원 전방향 영상 및 3차원 데이터를 이용한 시점 영상 제공 장치 및 방법
CN103247030A (zh) * 2013-04-15 2013-08-14 丹阳科美汽车部件有限公司 基于球面投影模型和逆变换模型的车辆全景显示系统鱼眼图像校正方法
US10397543B2 (en) 2014-09-03 2019-08-27 Nextvr Inc. Methods and apparatus for capturing, streaming and/or playing back content

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009068942A1 (en) * 2007-11-30 2009-06-04 Dooworks Fz Co Method and system for processing of images
CN101777193A (zh) * 2010-02-03 2010-07-14 中国人民解放军国防科学技术大学 基于双向映射精确插值的折反射全向图柱面全景展开方法
CN103839227A (zh) * 2012-11-23 2014-06-04 富士通株式会社 鱼眼图像校正方法和装置
CN104240236A (zh) * 2014-08-26 2014-12-24 中山大学 一种鱼眼镜头标定后鱼眼图像校正的方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3457681A4 *

Also Published As

Publication number Publication date
CN106993126A (zh) 2017-07-28
US20190349524A1 (en) 2019-11-14
KR102206899B1 (ko) 2021-01-22
EP3457681A1 (en) 2019-03-20
CN106993126B (zh) 2023-04-07
EP3457681A4 (en) 2019-12-25
KR20190008306A (ko) 2019-01-23
US10798300B2 (en) 2020-10-06

Similar Documents

Publication Publication Date Title
WO2017193729A1 (zh) 一种将镜头图像展开为全景图像的方法及装置
CN103942754B (zh) 全景图像补全方法及装置
JP6199313B2 (ja) モーションセンサーデータに基づく3dモデルを計算するように構成されたモバイルデバイス
TWI619091B (zh) 全景圖壓縮方法及裝置
CN106934772B (zh) 一种全景图像或视频的水平校准方法、系统及便携式终端
CN102982516B (zh) 一种基于半球环形全景镜头实现全景图像的方法
US20180018807A1 (en) Method and apparatus for generating panoramic image with texture mapping
CN105488766B (zh) 鱼眼镜头图像校正方法及装置
CN106920210B (zh) 一种基于球面展开模型的鱼眼视频实时全景校正方法
WO2019001255A1 (zh) 一种全景图像的拼接方法、装置及便携式终端
CN106780317B (zh) 全景投影方法及装置
CN106780389A (zh) 一种基于坐标转换的鱼眼图像校正方法及装置
CN107333051A (zh) 一种室内全景视频生成方法及装置
TWI635461B (zh) 投影圖構建方法及裝置
JP6451797B2 (ja) 画像処理方法及び画像処理装置
CN114581986A (zh) 图像处理方法、装置、电子设备及存储介质
CN105979241B (zh) 一种柱面立体全景视频快速反变换方法
CN108629828B (zh) 三维大场景的移动过程中的场景渲染过渡方法
CN108205822B (zh) 贴图方法及装置
CN111770326A (zh) 一种全景视频投射的室内三维监控方法
CN108986183A (zh) 一种全景地图的制作方法
JP6682984B2 (ja) 自由視点映像表示装置
CN109961395A (zh) 深度图像的生成及显示方法、装置、系统、可读介质
CN107346530B (zh) 一种修正鱼眼图像的投影方法及系统
CN109816765B (zh) 面向动态场景的纹理实时确定方法、装置、设备和介质

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17795347

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 20187035978

Country of ref document: KR

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 2017795347

Country of ref document: EP

Effective date: 20181211