WO2021093803A1 - 立体摄像机的校正品质的检测方法及检测系统 - Google Patents

立体摄像机的校正品质的检测方法及检测系统 Download PDF

Info

Publication number
WO2021093803A1
WO2021093803A1 PCT/CN2020/128352 CN2020128352W WO2021093803A1 WO 2021093803 A1 WO2021093803 A1 WO 2021093803A1 CN 2020128352 W CN2020128352 W CN 2020128352W WO 2021093803 A1 WO2021093803 A1 WO 2021093803A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
stereo camera
eye
square
eye image
Prior art date
Application number
PCT/CN2020/128352
Other languages
English (en)
French (fr)
Inventor
薛乐山
Original Assignee
南京深视光点科技有限公司
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 南京深视光点科技有限公司 filed Critical 南京深视光点科技有限公司
Publication of WO2021093803A1 publication Critical patent/WO2021093803A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N17/00—Diagnosis, testing or measuring for television systems or their details
    • H04N17/002—Diagnosis, testing or measuring for television systems or their details for television cameras
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N17/00—Diagnosis, testing or measuring for television systems or their details

Definitions

  • the present invention relates to the technical field of stereopsis, in particular to a detection method and a detection system that can evaluate the correction quality of a binocular or multi-eye camera after correction.
  • the manufacturer will perform camera calibration during the manufacturing process of the stereo camera, that is, pick out some sample points from the acquired images, and find out the internal and external parameters of the camera. (intrinsic and extrinsic parameters) and distortion parameters to ensure the high optical accuracy required by the stereo camera, and after the stereo camera is shipped and operated for a period of time, its optical accuracy will also be impossible due to the camera's use environment or method of use Maintain, at this time the stereo camera will also need to be calibrated to compensate for the drift of the stereo camera’s mechanism and optical accuracy.
  • the internal parameters are the inherent property parameters of the camera, mainly related to the conversion relationship between camera coordinates and image coordinates, such as imaging The center point, focal length, etc.
  • the external parameters indicate the conversion relationship between the camera's world coordinates and the camera coordinates, that is, the deflection angle and position of the camera.
  • the related technology and derivation principle of camera correction please refer to the OpenCV open source code library The "Camera Calibration" tutorial file of "Camera Calibration" will not be repeated here.
  • the current methods for evaluating the accuracy of the correction mainly include: (A) Let the stereo camera shoot a correction pattern with multiple feature points (such as a checkerboard and other test patterns) to obtain the image curvature of the checkerboard, but this way It can only roughly see whether the corrected distortion parameters are accurate from the degree of curvature, but cannot verify other camera parameters (such as internal and external parameters); or (B) use specific applications (such as 3D Point Cloud, 3D depth map, etc.) To check the quality of camera calibration, but because these applications are built on these internal and external parameters, and applications will have their own optimization and convergence methods, it is difficult to determine whether it is a camera calibration problem or a program optimization algorithm.
  • the current methods for evaluating the correction quality of stereo cameras still have many shortcomings. Therefore, how to propose a stereo camera that can test the accuracy of internal and external parameters and distortion parameters at the same time and make the detection results of the correction quality more accurate There are still problems to be solved in the detection method and detection system of the calibration quality.
  • the present invention provides a method for detecting the correction quality of a stereo camera, which includes:
  • the calibration plate image has a plurality of squares arranged in a matrix.
  • the squares have a plurality of square corners adjacent to each other. The distance between the corners of two adjacent squares is determined by Defined as a reference value;
  • the image coordinate of one of the corner points of the square is converted into a world coordinate
  • the comparison error can be used to detect the accuracy of the internal and external parameters of the stereo camera.
  • the detection method proposed by the present invention can continuously perform the following steps after the above steps are performed:
  • the present invention proposes a calibration quality detection system for a stereo camera, which includes a display, a stereo camera, and a calibration quality detection device; the display can display a calibration plate image, and the calibration plate images have a plurality of matrix images.
  • the stereo camera can be placed in front of the flat screen display, which uses A left-eye and a right-eye image acquisition unit captures the image of the calibration plate to obtain a left-eye and a right-eye image respectively;
  • the calibration quality detection device is respectively coupled to the display and the stereo camera, which have mutual information connections A feature detection unit, a coordinate conversion unit, and a correction evaluation unit;
  • the feature detection unit can detect multiple square corner point images from the images acquired by the stereo camera, and can extract the square corner points and image them on a left side of an image coordinate The imaging coordinates of the eye and a right eye;
  • the coordinate conversion unit can calculate a parallax according to the imaging coordinates of the left eye and the right eye, and can convert the image coordinates of the corner points of the box into a world coordinate according to the parallax and the internal and external parameters of the stereo camera;
  • the correction evaluation unit can calculate the difference between the world
  • the accuracy of the internal and external parameters and distortion parameters of the stereo camera can be tested at the same time, and the detection result of the stereo camera correction quality can be more accurate, and even in the case of limited detection space (such as a production line), the solution of the present invention can still be obtained Perform calibration quality testing.
  • the step of detecting a plurality of square corner point images from the left-eye and right-eye images is completed, before executing the next step, it is determined whether a variation of the square corner point image in a temporal and spatial domain is consistent with a The displacement threshold value, if there is, it is determined that the stereo camera is in a stable state.
  • the reference value when defining the reference value, it is calculated based on a screen size of the display, a screen resolution, and the number of squares of the calibration plate image.
  • the step of detecting a plurality of square corner point images from the left-eye and right-eye images is performed, if it is determined that the stereo camera is in the stable state, it is first calculated in the left-eye image and the right-eye image. A standard deviation of the square corner point image in the image is used to determine the feature point distribution state of each square corner point image. If the standard deviation is lower than a standard deviation threshold, the left-eye image and all the The distribution state of the feature points of the right-eye image is relatively uniform, and the next step is continuously executed.
  • the feature detection unit is also used to determine whether a variation of the square corner image in a temporal and spatial domain meets a displacement threshold, so as to determine whether the stereo camera meets a stable state.
  • the reference value of the calibration plate image of the display is calculated based on a screen size, a screen resolution, and the number of squares of the calibration plate image of the display.
  • the feature detection unit is also used to calculate a standard deviation of the square corner point image in the left-eye image and the right-eye image to determine The feature point distribution state of each of the square corner point images.
  • Figure 1 is a system architecture diagram of the present invention.
  • Figure 2 is a flow chart of the system implementation of the present invention.
  • Fig. 3 is a schematic diagram of disparity calculation of the present invention.
  • Figure 4 is a schematic diagram of the distortion error calculation of the present invention.
  • Fig. 5 is a system implementation flowchart of another embodiment (1) of the present invention.
  • Fig. 6 is a system implementation flowchart of another embodiment (2) of the present invention.
  • S10 displays an image of the calibration board with a known square size on the monitor
  • S301 analyzes the image stability of the corner points of the block in the time and space domain
  • S302 analyzes the standard deviation of the corner point image of the block
  • S40 calculates corner parallax and converts from image coordinates to world coordinates
  • S60 projects multiple reference images to observe distortion
  • FC1 box corner image FC2 box corner image
  • FIG. 1 is a system architecture diagram of the present invention.
  • the present invention proposes a calibration quality detection system 1 of a stereo camera, which includes a display 10, a stereo camera 20, and a calibration quality detection device 30, in which:
  • the display 10 can display a pattern board 101.
  • the pattern board 101 can have a plurality of squares arranged in a matrix.
  • the squares can have a plurality of square corners adjacent to each other, and two adjacent The distance between the corners of the square is defined as a known ground truth;
  • the stereo camera 20 is placed in front of the flat-screen display 10, and has a left-eye image acquisition unit 201 and a right-eye image acquisition unit 202 that are information-linked to each other, and can respectively photograph the calibration plate image 101 to Obtain a left-eye image and a right-eye image respectively.
  • the stereo camera 20 may also be a multi-eye camera and have more than two image acquisition units;
  • the calibration quality detection device 30 is respectively coupled to the display 10 and the stereo camera 20 for communication and data transmission with the display 10 and the stereo camera 20 respectively. It may have a feature detection unit 301 and a coordinate which are connected to each other. A conversion unit 302, a calibration evaluation unit 303, and a processor (not shown in the figure). The processor is used to run the calibration quality detection device 30 and can control the actions of the above-mentioned units, and has logic operations, temporary storage of results, Save functions such as the position of execution instructions.
  • the feature detection unit 301 can detect multiple box images (FG1, FG2%) and multiple box corner point images (FC1, FC2%) from the left-eye and right-eye images acquired by the stereo camera 20, and can extract them
  • One of the corner points of a square is imaged at a left-eye imaging coordinate and a right-eye imaging coordinate of an image coordinate, and the technical means of detecting or rendering the square image and the square corner image is not the main feature of the present invention. This will not be described in detail.
  • Related technologies can refer to functions such as findChessboardCorners() of the OpenCV open source code library, and are not limited to this;
  • the coordinate conversion unit 302 can calculate a parallax based on the left-eye imaging coordinates and the right-eye imaging coordinates, and can further convert the image coordinates of the corner points of the square to a parallax based on the parallax, an internal parameter and an external parameter of the stereo camera 20
  • the world coordinate (world coordinate), where the internal parameters can include a focal length (focal length), a position deviation (cx, cy, the unit is pixel, which means that in the positive x-axis direction and the positive y-axis direction, the left eye And the deviation value of the center of the right-eye image acquisition unit (201, 202) relative to the optical axis of the lens), a distortion coefficients, and the coordinate conversion process needs to be converted to a camera coordinate.
  • the internal parameters can include a focal length (focal length), a position deviation (cx, cy, the unit is pixel, which means that in the positive x-axis direction and the positive y-axis direction, the
  • the image coordinates are introduced to describe the projection transmission relationship of the object from the camera coordinate system to the image coordinate system during the imaging process. It is the coordinate system where the image that we actually read from the stereo camera 20 is located, and the unit is pixel (pixel) ;
  • Camera coordinates are the coordinate system established with the stereo camera 20 as the origin, defined in order to describe the position of the object from the perspective of the stereo camera 20, as an intermediate link between the world coordinate system and the image coordinate system, the unit is m;
  • the world coordinate is the user
  • the defined three-dimensional space coordinate system is imported in order to describe the position of a target in the real world, and the unit is m;
  • the correction evaluation unit 303 may calculate the difference between the world coordinates of the corner points of two adjacent squares, and compare the difference with the reference value to obtain a comparison error, and the comparison error can be used to detect the inside of the stereo camera 20 And the accuracy of external parameters, therefore, it can solve the problem that there is no positive solution (ie reference value) for comparison with conventionally well-known calibration quality detection methods or systems;
  • the correction evaluation unit 303 of this embodiment can also project a plurality of rectangular reference images to the left-eye image and the right-eye image, respectively, and extract them with the left-eye and right-eye image acquisition units (201, 202) of the stereo camera 20.
  • a principal point that is, the intersection of the camera's optical axis and the imaging plane
  • the rectified stereo camera 20 can be used to evaluate the various block images (FG1, FG2%) and the corner images (FC1) acquired by the stereo camera 20.
  • the feature detection unit 301 of this embodiment can also determine whether a variation of the square corner point images (FC1, FC2%) in a temporal and spatial domain meets a displacement threshold, so as to determine whether the stereo camera 20 meets a stable state If it is in a stable state, the feature detection unit 301 can also calculate a standard deviation (STD) of the corner point images (FC1, FC2%) in the left-eye image and the right-eye image to determine the corner point images ( FC1, FC2%) feature point distribution status;
  • STD standard deviation
  • the reference value of the calibration plate image 101 displayed on the display 10 is calculated by a screen size, a screen resolution, and the number of squares of the calibration plate image of the display 10;
  • the calibration board image 101 can be one of a Chessboard pattern, an ArUco pattern, or a ChArUco pattern, and is not limited to this.
  • the present invention provides a method for detecting the calibration quality of a stereo camera. S, can include the following steps:
  • a calibration quality detection device 30 causes a display 10 to display a calibration plate image 101.
  • the calibration plate image 101 has a plurality of squares arranged in a matrix.
  • the square has a plurality of square corner points adjacent to each other, and the distance between the two adjacent square corner points is defined as a reference value by the correction quality detection device 30. More specifically, the reference value can be determined by the screen size and The current resolution of the screen and the number of squares of the calibration plate image 101 are calculated.
  • the screen size is 128x72 cm
  • the resolution of the calibration plate image 101 is 3840x2160 pixels
  • the width is 16 squares
  • the height is 9 squares
  • each square is 240x24 pixels
  • the number of pixels per centimeter (pixel/cm) is 30, that is, 1 centimeter on the screen displays 30 pixels, and then each square can be calculated
  • the reference value (positive solution) is 8x8 cm;
  • Step S20 A stereo camera 20 placed in front of the display 10 with a flat screen uses a left-eye image acquisition unit 201 and a right-eye image acquisition unit 202 to adjust the correction plate image 101 respectively. Shooting to obtain a left-eye image and a right-eye image respectively;
  • a feature detection unit 301 of the calibration quality detection device 30 detects a plurality of block images (FG1, FG2%) and a plurality of blocks from the left-eye image and the right-eye image, respectively Corner image (FC1, FC2...);
  • a coordinate conversion unit 302 of the correction quality detection device 30 extracts one of the square corner points from the left-eye image and the right-eye image respectively ( For example, after the target point P) is viewed by the left-eye and right-eye image acquisition units (201, 202), it is imaged at a left-eye imaging coordinate XL and a right-eye imaging coordinate XR of an image coordinate, and according to the left-eye imaging coordinates XL and The difference between the right-eye imaging coordinates XR calculates a parallax d, that is, calculates the difference between the horizontal coordinates of the left-eye and right-eye images of the target point P, thus realizing the principle of binocular ranging, as shown in " Figure 3" Shown
  • the coordinate conversion unit 302 calculates the parallax d, it then converts the image coordinates of the corner points of the square into a world coordinate (XW, YW, ZW), for another square corner point adjacent to the extracted square corner point, repeat step S40 to calculate the world coordinate of the other square corner point, wherein, when the world coordinate is obtained in this step S40, the following is obtained
  • b in the following formula is the center distance difference between the left eye and the right eye glasses (OL, OR) in "Figure 3"
  • Figure 3 Represents the point coordinates of the corner point of the block (for example, the target point P) in the image coordinate system
  • d is the parallax
  • XW, YW, ZW respectively represent the point coordinates of the corner point of the block in the world coordinate system.
  • the related derivation principle can also be referred to OpenCV API Reference is the tutorial document published in "Camera Calibration and 3D Reconstruction". However, the following formulas and OpenCV related
  • the internal parameters required to perform step S40 may include parameters such as a focal length, a positional deviation (cx, cy), and a distortion parameter;
  • step S50 Compare the world coordinate distance of the corner points of the two squares extracted by the camera with the reference value to obtain a comparison error (step S50): a correction evaluation unit 303 of the correction quality detection device 30 calculates the adjacent two squares The difference of the world coordinates of the corner points is compared with the reference value to obtain a comparison error.
  • the comparison error can be used to detect the accuracy of the internal and external parameters of the stereo camera 20. Therefore, it can solve the conventional well-known calibration quality.
  • the detection method or system has no positive solution (ie benchmark value) for comparison.
  • the detection method S of this embodiment can also continuously perform the following steps:
  • the correction evaluation unit 303 projects a plurality of rectangular reference images (R1, R2,%) to the left-eye image and the right-eye image, and extracts and three-dimensional
  • the image of the corner point of the square closest to a principal axis point Pp (principle point) of the camera 20 is used as a central reference point Fc, as shown in "FIG. 4", and it can be seen from FIG. 4 that the reference image (R1, R2,...) is preferably presented in a rectangular shape with four sides of equal length, and each reference image (R1, R2,7) is preferably arranged at the same interval;
  • step S70 The correction evaluation unit 303 sequentially calculates the difference between a reference image corner point (R1C1, R2C1%) of each reference image (R1, R2,%) and the world coordinate of the center reference point Fc The value is recorded as a distortion error, and it is judged whether the distortion error meets a distortion threshold. If not, an image correction action is performed again. According to this, the corrected stereo camera 20 can be evaluated for the stereo camera 20.
  • each block (FG1, FG2%) and corner images (FC1, FC2%) acquired by the camera 20 are farther from the main axis point (center), and the more severe the distortion has been corrected (that is, the corner images (FC1) , FC2%) whether the distances between each other are the same), so it can be judged whether the distortion parameters are calibrated accurately, and the distortion can be divided into radial distortion and tangential distortion.
  • Radial distortion can be divided into Pincushion and Barrel distortion, which will not be repeated here.
  • Figure 5" is a system implementation flowchart of another embodiment (1) of the present invention. Please also refer to " Figure 1".
  • the technology of this embodiment is similar to that of " Figure 2", with the main difference being Since the stereo camera 20 can acquire multiple images at the same angle in step S20, when step S30 is executed, and before step S40 is executed, step S301 (analyze the image stability of the corner points of the block in the temporal and spatial domain):
  • the detection unit 301 determines whether a variation of the corner point images (FC1, FC2%) of the block in a temporal and spatial domain meets a displacement threshold, and if so, determines that the stereo camera 20 is in a stable state (stability), and can perform the steps continuously S40; On the contrary, if the displacement threshold value is not met, that is, it is determined that the image coordinates of the corner point images (FC1, FC2...) of the block will be inconsistent due to jitter; therefore, after this embodiment is implemented, the stereo camera 20 can be tested in advance Whether the acquired image
  • Figure 6 is a system implementation flow chart of another embodiment (2) of the present invention, and please refer to " Figure 1".
  • This embodiment has the same technology as " Figure 5", but the main differences are That is, when the analysis result of step S301 is consistent, that is, when it is determined that the stereo camera 20 is in a stable state, step S302 (analysis of the standard deviation of the square corner point image) can be performed continuously: the feature detection unit 301 calculates the left-eye image and the right-eye image The standard deviation of the corner point image in the square is used to determine the feature point distribution status of the corner point images of each block. If the standard deviation is lower than a standard deviation threshold, then the feature points of the left eye image and the right eye image are judged The distribution state is relatively uniform, and step S40 can be executed continuously.
  • the present invention after the present invention is implemented, at least the accuracy of the internal and external parameters and distortion parameters of the stereo camera can be tested at the same time, and the detection result of the stereo camera correction quality can be more accurate.
  • the solution of the present invention can still be used to detect the calibration quality.
  • the image of the calibration plate of the present invention is displayed on a flat screen, the light source is uniform and calibrated. When the board image is controllable, the calibration quality inspection result will be more accurate.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Image Processing (AREA)
  • Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
  • Length Measuring Devices By Optical Means (AREA)

Abstract

本发明提供了一种立体摄像机的校正品质的检测方法及检测系统，主要让一立体摄像机拍摄一显示器显示的一校正板影像，并提取出多个方块角点影像、分析角点影像稳定度及标准差后，可依据视差及摄像机的内外部参数，分别计算出摄像机所获取影像的两相邻方块角点的世界坐标，接着，可将显示器所显示的校正板影像的两相邻方块角点的距离视为基准值(正解)，并与两相邻方块角点的世界坐标的差值计算出一比对误差，以检测摄像机的内部与外部参数的准确性，本发明更提出可进一步确认距离一中心基准点较远的各方块大小是否均一致的作法，以检测摄像机的畸变参数是否准确。

Description

立体摄像机的校正品质的检测方法及检测系统 技术领域
本发明涉及立体视觉(stereopsis)的技术领域，尤指一种可针对校正后的双目或多目摄像机评估其校正品质的检测方法及检测系统。
背景技术
为了要使立体摄像机产生高品质的立体影像，在摄像机出厂前，制造商会在立体摄像机的制造过程中进行摄像机校正，即从被获取影像中挑出一些样本点，从中找出摄像机的内外部参数(intrinsic and extrinsic parameters)及畸变参数，进而确保立体摄像机所需的高光学准度，且在立体摄像机出厂且被操作一段时间后，其光学准度也会因为摄像机的使用环境或使用方法而无法维持，此时立体摄像机也会需要再进行校正以补偿立体摄像机的机构与光学准度漂移，其中，内部参数是摄像机的固有性质参数，主要有关于摄像机坐标与影像坐标间的转换关系，例如成像中心点、焦距等，外部参数则表示摄像机在世界坐标与摄像机坐标间的转换关系，即摄像机的偏转角度与位置，而有关摄像机校正的相关技术及推导原理，可参见OpenCV开放原始码函式库的「Camera Calibration」教程文件，于此不再赘述。
目前评估校正是否准确的方法主要有：(A)让立体摄像机对具有复数个特征点的校正图案(例如棋盘格等测试图案)进行拍摄，以获取出棋盘格的影像弯曲状况，但此种方式仅能从弯曲程度粗略地看出校正后的畸变参数是否准确，而无法验证其它摄像机参数(如内外部参数)；或(B)利用特定的应用程式(例如3D Point Cloud、3D depth map等)来检视摄像机校正的品质，但由于这些应用程式都建构在这些内外部参数之上，且应用程式都会有自身的优化与收敛方式，而难以判 断是摄像机校正的问题，或是程式优化的演算法问题；或(C)将被获取影像的样本点信息，带入由内外部参数与畸变参数所组成的方程式(例如OpenCV开放原始码函式库的「projectPoints()」函式)，以计算一重投影误差(re-projection error)，该数值其若接近于0，则代表影像校正的品质越好，惟该数值取决于被选取样本点在被获取影像中的分布，并不能表示三维空间的任意点都在这个误差之内。
综上可知，目前评估立体摄像机的校正品质的方式，仍有诸多缺点，因此如何提出一种可同时检验内外部参数与畸变参数的准确性、可让校正品质的检测结果更臻准确的立体摄像机的校正品质的检测方法及检测系统，仍有待解决的问题。
发明内容
为达上述目的，本发明提出一种立体摄像机的校正品质的检测方法，包含：
(1)使一显示器显示一校正板影像，校正板影像具有多个以矩阵方式排列的方块，方块具有彼此相邻的多个方块角点，相邻的两方块角点之间的距离，被定义为一基准值；
(2)使置于显示器前的一立体摄像机对校正板影像进行拍摄，以分别获取出一左眼影像与一右眼影像；
(3)分别从左眼及右眼影像检测出多个方块角点影像，并提取其中一方块角点成像于一影像坐标的一左眼成像坐标及一右眼成像坐标，并依据左眼及右眼成像坐标计算一视差；
(4)依据视差、立体摄像机的内部及外部参数，将其中一方块角点的影像坐标转换为一世界坐标；
(5)计算相邻于被提取方块角点的另一方块角点的世界坐标，并求得相邻的两方块角点的世界坐标的差值，再以差值与基准值比较出一比对误差，比对误差可供检测立体摄像机的内部及外部参数的准确性。
优选地，本发明提出的检测方法在上述步骤执行完毕后，可连续 执行以下步骤：
(1)将矩形的多个基准影像分别投影至左眼及右眼影像，并提取与立体摄像机的一主轴点(principle point)最邻近的方块角点影像，作为一中心基准点；
(2)依序计算各基准影像的一基准影像角点与中心基准点的世界坐标的差值，以分别记录为一畸变误差，并判断畸变误差是否符合一畸变阈值，若未符合，则可重新进行影像校正。
为达上述目的，本发明提出一种立体摄像机的校正品质的检测系统，包含一显示器、一立体摄像机及一校正品质检测装置；显示器可显示一校正板影像，校正板影像具有多个以矩阵方式排列的方块，方块具有彼此相邻的多个方块角点，且相邻的两方块角点的间的距离，被定义为一基准值；立体摄像机可置于荧幕平整的显示器前，其利用一左眼及一右眼影像获取单元对校正板影像进行拍摄，以分别获取出一左眼与一右眼影像；校正品质检测装置分别连耦接于显示器及立体摄像机，其具有相互呈信息连结的一特征检测单元、一坐标转换单元及一校正评价单元；特征检测单元可从立体摄像机所获取的影像检测出多个方块角点影像，并可提取方块角点成像于一影像坐标的一左眼及一右眼成像坐标；坐标转换单元可依据左眼及右眼成像坐标计算一视差，并可依据视差、立体摄像机的内部及外部参数，将方块角点的影像坐标转换为一世界坐标；校正评价单元可计算相邻的两方块角点的世界坐标的差值，也可就差值与基准值比较出一比对误差；再者，校正评价单元也可将矩形的多个基准影像分别投影至左眼及右眼影像，并提取与立体摄像机的一主轴点最邻近的方块角点影像，作为一中心基准点，再依序计算各基准影像的一基准影像角点与中心基准点的世界坐标的差值，以分别记录为一畸变误差。
采用上述技术方案带来的有益效果：
可同时检验立体摄像机之内外部参数与畸变参数的准确性、可让立体摄像机校正品质的检测结果更加准确，且即便在检测空间有限的情况下(如产线)，仍然得以本发明的解决方案进行校正品质的检测。
优选地，当所述分别从左眼及右眼影像检测出多个方块角点影像步骤执行完毕，执行下一步骤前，先判断所述方块角点影像于一时空域的一变异量是否符合一位移门槛值，若有则判断所述立体摄像机为一稳定状态。
优选地，定义所述基准值时，系以所述显示器的一荧幕尺寸、一荧幕解析度及所述校正板影像的一方块数量计算而得。
优选地，执行所述分别从左眼及右眼影像检测出多个方块角点影像步骤时，若判断所述立体摄像机为所述稳定状态，先计算于所述左眼影像及所述右眼影像中的所述方块角点影像的一标准差，以判断各所述方块角点影像的特征点分布状态，若所述标准差低于一标准差阈值，则判断所述左眼影像及所述右眼影像的特征点分布状态较为均匀，并连续执行下一步骤。
优选地，所述特征检测单元也供以判断所述方块角点影像于一时空域的一变异量是否符合一位移门槛值，以判断所述立体摄像机是否符合一稳定状态。
优选地，所述显示器的所述校正板影像的所述基准值，系以所述显示器的一荧幕尺寸、一荧幕解析度及所述校正板影像的一方块数量计算而得。
优选地，若所述立体摄像机符合所述稳定状态，则所述特征检测单元也供计算于所述左眼影像及所述右眼影像中的所述方块角点影像的一标准差，以判断各所述方块角点影像的特征点分布状态。
为使贵审查委员得以清楚了解本发明的目的、技术特征及其实施后的功效，兹以下列说明搭配图示进行说明，敬请参阅。
附图说明
图1为本发明的系统架构图。
图2为本发明的系统实施流程图。
图3为本发明的视差计算示意图。
图4为本发明的畸变误差计算示意图。
图5为本发明的另一实施例(一)的系统实施流程图。
图6为本发明的另一实施例(二)的系统实施流程图。
符号说明：
S立体摄像机的校正品质的检测方法
S10于显示器显示方块大小为已知的校正板影像
S20立体摄像机进行影像获取
S30检测方块角点的特征
S301于时空域分析方块角点影像稳定度
S302分析方块角点影像的标准差
S40计算角块角点视差及从影像坐标转换为世界坐标
S50将摄像机所提取两方块角点的世界坐标距离与基准值作比较，以求得一比对误差
S60投影多个基准影像以观察畸变情形
S70计算畸变误差
FG1方块影像        FG2方块影像
FC1方块角点影像    FC2方块角点影像
XL左眼成像坐标     XR右眼成像坐标
d视差
Pp主轴点           Fc中心基准点
R1基准影像         R2基准影像
R1C1基准影像角点   R2C1基准影像角点
1立体摄像机的校正品质的检测系统
10显示器           101校正板影像
20立体摄像机       201左眼影像获取单元
                   202右眼影像获取单元
30校正品质检测系统 301特征检测单元
                   302坐标转换单元
                   303校正评价单元
具体实施方式
请参阅「图1」，其为本发明的系统架构图，本发明提出一种立体摄像机的校正品质的检测系统1，包含一显示器10、一立体摄像机20及一校正品质检测装置30，其中：
(1)显示器10可显示一校正板影像101(pattern board)，校正板影像101可具有多个以矩阵方式排列的方块，方块可具有彼此相邻的多个方块角点，且相邻的两方块角点之间的距离，被定义为一基准值(known ground truth)；
(2)立体摄像机20置于荧幕平整的显示器10前，并具有相互呈信息连结的一左眼影像获取单元201及一右眼影像获取单元202，可分别对校正板影像101进行拍摄，以分别获取出一左眼影像与一右眼影像，另，立体摄像机20也可为一多目摄像机而具备二个以上的影像获取单元；
(3)校正品质检测装置30分别耦接于显示器10及立体摄像机20，以分别与显示器10及立体摄像机20进行通讯与数据传输，其可具有相互呈信息连结的一特征检测单元301、一坐标转换单元302、一校正评价单元303及一处理器(图中未绘示)，处理器供以运行校正品质检测装置30且可控制上述各单元作动，并具备逻辑运算、暂存运算结果、保存执行指令位置等功能。
(4)特征检测单元301可从立体摄像机20所获取的左眼及右眼影像检测出多个方块影像(FG1、FG2…)及多个方块角点影像(FC1、FC2…)，并可提取其中一方块角点成像于一影像坐标(image coordinate)的一左眼成像坐标及一右眼成像坐标，而检测或渲染方块影像及方块角点影像的技术手段，并非本发明的主要特征，于此不加以详述，相关技术可参OpenCV开放原始码函式库的findChessboardCorners()等函数，均不以此为限；
(5)坐标转换单元302可依据左眼成像坐标及右眼成像坐标计 算一视差，并可进一步依据视差、立体摄像机20的一内部参数及一外部参数，将方块角点的影像坐标转换至一世界坐标(world coordinate)，其中，内部参数可包含一焦距(focal length)、一位置偏差量(cx,cy，单位为像素，其分别表示在正x轴方向与正y轴方向中，左眼及右眼影像获取单元(201,202)的中心相对于透镜的光轴的偏差值)、一畸变参数(distortion coefficients)，而坐标转换过程中会需要先转换至一摄像机坐标(camera coordinate)，一般而言，影像坐标是为了描述成像过程中物体从摄像机坐标系到影像坐标系的投影透射关系而引入，是我们真正从立体摄像机20内读取到的影像所在的坐标系，单位为像素(pixel)；摄像机坐标就是以立体摄像机20为原点建立的坐标系，为了从立体摄像机20的角度描述物体位置而定义，作为沟通世界坐标系和影像坐标系的中间环节，单位为m；世界坐标为使用者定义的三维空间的坐标系，是为了描述某目标物在真实世界里的位置而被导入，单位为m；
(6)校正评价单元303可于计算相邻的两方块角点的世界坐标的差值后，以差值与基准值比较出一比对误差，而比对误差可供检测立体摄像机20的内部及外部参数的准确性，因此，可解决常规熟知校正品质检测方法或系统并无正解(即基准值)可供比对的问题；
(7)另，本实施例的校正评价单元303也可将矩形的多个基准影像分别投影至左眼影像及右眼影像，并提取与立体摄像机20的左眼及右眼影像获取单元(201,202)的一主轴点(principle point，即摄像机光轴与成像平面的交点)最邻近的方块角点影像，作为一中心基准点，再依序计算各基准影像的一基准影像角点与中心基准点的世界坐标的差值，以分别记录为一畸变误差，依此，可针对经过校正的立体摄像机20，评估被立体摄像机20获取的各方块影像(FG1、FG2…)及各角点影像(FC1、FC2…)距离主轴点(中心)越远，畸变状况越严重的情形是否已被修正(即各角点影像(FC1、FC2…)彼此的间的距离是否均一致)，故因此可判断畸变参数校正得是否够准确；
(8)另，本实施例的特征检测单元301也可判断方块角点影像 (FC1、FC2…)于一时空域的一变异量是否符合一位移门槛值，以判断立体摄像机20是否符合一稳定状态，若符合稳定状态，则特征检测单元301也可计算于左眼影像及右眼影像中的方块角点影像(FC1、FC2…)的一标准差(STD)，以判断各方块角点影像(FC1、FC2…)的特征点分布状态；
(9)其中，显示器10所显示校正板影像101的基准值，得以显示器10的一荧幕尺寸、一荧幕解析度及校正板影像的一方块数量计算而得；
(10)其中，校正板影像101可为一棋盘(Chessboard)图案、一ArUco图案或一ChArUco图案的其中一种，并不以此为限。
请继续参阅「图2」，其为本发明的系统实施流程图，并请搭配参阅「图1」、「图3」及「图4」，本发明提出一种立体摄像机的校正品质的检测方法S，可包括以下步骤：
(1)于显示器显示方块大小为已知的校正板影像(步骤S10)：一校正品质检测装置30使一显示器10显示一校正板影像101，校正板影像101具有多个以矩阵方式排列的方块，方块具有彼此相邻的多个方块角点，且相邻的两方块角点的间的距离，被校正品质检测装置30定义为一基准值，更具体而言，基准值可由荧幕尺寸与荧幕当下的解析度及校正板影像101的一方块数量来求得，举例而言，若荧幕大小为128x72公分，校正板影像101的解析度为3840x2160像素，宽度为16个方块，高度为9个方块，则每个方格为240x24像素，每公分的像素数量(pixel/cm)为30，也就是荧幕上的1公分就显示了30个像素，进而可求得每个方格的基准值(正解)为8x8公分；
(2)立体摄像机进行影像获取(步骤S20)：置于荧幕平整的显示器10前的一立体摄像机20，分别以一左眼影像获取单元201及一右眼影像获取单元202对校正板影像101进行拍摄，以分别获取出一左眼影像与一右眼影像；
(3)检测方块角点的特征(步骤S30)：校正品质检测装置30的一特征检测单元301分别从左眼影像及右眼影像检测出多个方块影像 (FG1、FG2…)及多个方块角点影像(FC1、FC2…)；
(4)计算角块角点视差及从影像坐标转换为世界坐标(步骤S40)：校正品质检测装置30的一坐标转换单元302分别从左眼影像与右眼影像，提取其中一方块角点(如目标点P)分别被左眼及右眼影像获取单元(201,202)观视后，成像于一影像坐标的一左眼成像坐标XL及一右眼成像坐标XR，并依据左眼成像坐标XL及右眼成像坐标XR的差值计算出一视差d，即计算目标点P成像于左眼及右眼影像的横向坐标的差值，从而实现了双目测距的原理，即如「图3」所示；
(5)承上，坐标转换单元302计算出视差d后，再依据视差d、立体摄像机20的一内部参数及一外部参数，将方块角点的影像坐标转换为一世界坐标(XW,YW,ZW)，再对相邻于被提取的方块角点的另一方块角点，重复执行步骤S40以计算另一方块角点的世界坐标，其中，本步骤S40于求得世界坐标时，得以下述公式进行演算，并且，下述公式的b为「图3」的左眼与右眼镜心(OL、OR)的中心距离差，
Figure PCTCN2020128352-appb-000001
表示方块角点(例如目标点P)于影像坐标系下的点坐标，d为视差，XW、YW、ZW则分别表示方块角点于世界坐标系下的点坐标，相关推导原理也可参见OpenCV API Reference于「Camera Calibration and 3D Reconstruction」所公开的教程文件，惟以下公式及OpenCV相关技术仅为举例，并不以此为限，特先陈明：
Figure PCTCN2020128352-appb-000002
Figure PCTCN2020128352-appb-000003
Figure PCTCN2020128352-appb-000004
Figure PCTCN2020128352-appb-000005
(6)另，执行步骤S40所需的内部参数可包含一焦距、一位置偏差量(cx,cy)、一畸变参数等参数；
(7)将摄像机所提取两方块角点的世界坐标距离与基准值作比较，以求得一比对误差(步骤S50)：校正品质检测装置30的一校正评价单元303计算相邻的两方块角点的世界坐标的差值，再以差值与基准值比较出一比对误差，比对误差可供检测立体摄像机20的内部参数及外部参数的准确性，因此，可解决常规熟知校正品质检测方法或系统并无正解(即基准值)可供比对的问题。
(8)承上，本实施例的检测方法S更可连续执行以下步骤：
(9)投影多个基准影像以观察畸变情形(步骤S60)：校正评价单元303将矩形的多个基准影像(R1、R2、…)分别投影至左眼影像及右眼影像，并提取与立体摄像机20的一主轴点Pp(principle point)最邻近的方块角点影像，作为一中心基准点Fc，即如「图4」所示，且由图4可知，本实施例的基准影像(R1、R2、…)以四个边均等长的矩形状呈现为佳，且各基准影像(R1、R2、…)均以相同的间距间隔排列为佳；
(10)计算畸变误差(步骤S70)：校正评价单元303依序计算各基准影像(R1、R2、…)的一基准影像角点(R1C1、R2C1…)与中心基准点Fc的世界坐标的差值，以分别记录为一畸变误差(distortion error)，并判断畸变误差是否符合一畸变阈值，若未符合，重新进行一影像校正动作，依此，可针对经过校正的立体摄像机20，评估被立体摄像机20获取的各方块影像(FG1、FG2…)及各角点影像(FC1、FC2…)距离主轴点(中心)越远，畸变越严重的情形是否已被修正(即各角点影像(FC1、FC2…)彼此的间的距离是否均一致)，故因此可判断畸变参数校正得是否够准确，而畸变主要可分为径向畸变(Radial Distortion)和切向畸变(tangential distortion)，其中，径向畸变又可分为Pincushion及Barrel畸变，于此不再赘述。
请参阅「图5」，其为本发明的另一实施例(一)的系统实施流程图，并请搭配参阅「图1」，本实施例与「图2」的技术类同，主要差异在于，由于立体摄像机20可于步骤S20以一致角度获取多张影像，故当步骤S30执行完毕，并于执行步骤S40前，可先执行步骤 S301(于时空域分析方块角点影像稳定度):特征检测单元301判断方块角点影像(FC1、FC2…)于一时空域的一变异量是否符合一位移门槛值，若有符合，则判断立体摄像机20为一稳定状态(stability)，并可连续执行步骤S40；反的，若未符合位移门槛值，也就是判断方块角点影像(FC1、FC2…)的影像坐标会因跳动而不一致；因此，本实施例据以实施后，可预先测试立体摄像机20所获取的影像是否模糊，如果影像模糊则无需连续执行步骤S40，而应先修正影像稳定度不佳的问题。
请参阅「第6图」，其为本发明的另一实施例(二)的系统实施流程图，并请搭配参阅「图1」，本实施例与「图5」的技术类同，主要差异在于，当步骤S301的分析结果为符合，即判断立体摄像机20为稳定状态时，可连续执行步骤S302(分析方块角点影像的标准差):特征检测单元301计算于左眼影像及右眼影像中的方块角点影像的一标准差(Standard Deviation)，以判断各方块角点影像的特征点分布状态，若标准差低于一标准差阈值，则判断左眼影像及右眼影像的特征点分布状态较为均匀，并可连续执行步骤S40。
综上可知，本发明据以实施后，至少可达成同时检验立体摄像机之内外部参数与畸变参数的准确性、可让立体摄像机校正品质的检测结果更臻准确的有益功效，并且，即便在检测空间有限的情况下(如产线)，仍然得以本发明的解决方案进行校正品质的检测，再者，由于本发明的校正板影像系由荧幕平整的显示器予以显示，故在光源均匀、校正板影像可控的情况下，作出的校正品质检测结果将更为准确。
以上所述者，仅为本发明较佳的实施例而已，并非用以限定本发明实施的范围；任何熟习此技艺者，在不脱离本发明的精神与范围下所作的均等变化与修饰，皆应涵盖于本发明的专利范围内。

Claims (10)

  1. 一种立体摄像机的校正品质的检测方法，包含：
    (A)一显示器显示一校正板影像，所述校正板影像具有多个以矩阵方式排列的方块，所述方块具有彼此相邻的多个方块角点，且相邻的两所述方块角点的间的距离，被定义为一基准值；
    (B)置于所述显示器前的一立体摄像机对所述校正板影像进行拍摄，分别获取出一左眼影像与一右眼影像；
    (C)分别从所述左眼影像及所述右眼影像检测出多个方块角点影像；
    (D)分别从所述左眼影像与所述右眼影像，提取其中一所述方块角点成像于一影像坐标的一左眼成像坐标及一右眼成像坐标，并依据所述左眼成像坐标及所述右眼成像坐标计算一视差；
    (E)依据所述视差、所述立体摄像机的一内部参数及一外部参数，将其中一所述方块角点的所述影像坐标转换为一世界坐标；
    (F)将相邻于被提取所述方块角点的另一所述方块角点，重复执行所述步骤(D)～所述步骤(E)以计算另一所述方块角点的所述世界坐标；以及
    (G)计算相邻的两所述方块角点的所述世界坐标的差值，再以所述差值与所述基准值比较出一比对误差，所述比对误差可供检测所述立体摄像机的所述内部参数及所述外部参数的准确性。
  2. 如权利要求1所述的立体摄像机的校正品质的检测方法，其特征在于，所述(G)步骤执行完毕后，连续执行以下步骤：
    (H)将矩形的多个基准影像分别投影至所述左眼影像及所述右眼影像，并提取与所述立体摄像机的一主轴点最邻近的所述方块角点影像，作为一中心基准点；
    (I)依序计算各所述基准影像的一基准影像角点与所述中心基准点的所述世界坐标的差值，以分别记录为一畸变误差，并判断所述畸变误差是否符合一畸变阈值，若未符合，重新进行一影像校正动作。
  3. 如权利要求1所述的立体摄像机的校正品质的检测方法，其特征在于，当所述(C)步骤执行完毕，执行所述(D)步骤前，先判断所述方块角点影像于一时空域的一变异量是否符合一位移门槛值，若有则判断所述立体摄像机为一稳定状态。
  4. 如权利要求1所述的立体摄像机的校正品质的检测方法，其特征在于，于所述(A)步骤定义所述基准值时，系以所述显示器的一荧幕尺寸、一荧幕解析度及所述校正板影像的一方块数量计算而得。
  5. 如权利要求3所述的立体摄像机的校正品质的检测方法，其特征在于，执行所述(C)步骤时，若判断所述立体摄像机为所述稳定状态，先计算于所述左眼影像及所述右眼影像中的所述方块角点影像的一标准差，以判断各所述方块角点影像的特征点分布状态，若所述标准差低于一标准差阈值，则判断所述左眼影像及所述右眼影像的特征点分布状态较为均匀，并连续执行所述(D)步骤。
  6. 一种立体摄像机的校正品质的检测系统，包含：
    一显示器，供以显示一校正板影像，所述校正板影像具有多个以矩阵方式排列的方块，所述方块具有彼此相邻的多个方块角点，且相邻的两所述方块角点的间的距离，被定义为一基准值；
    一立体摄像机，置于荧幕平整的所述显示器前，供以利用一左眼影像获取单元和一右眼影像获取单元对所述校正板影像进行拍摄，以分别获取出一左眼影像与一右眼影像；
    一校正品质检测装置，分别耦接于所述显示器及所述立体摄像机，所述校正品质检测装置具有相互呈信息连结的一特征检测单元、一坐标转换单元及一校正评价单元；
    该所述特征检测单元供以从所述立体摄像机所获取的影像检测出多个方块角点影像，并可提取所述方块角点成像于一影像坐标的一左眼成像坐标及一右眼成像坐标；
    该所述坐标转换单元供以依据所述左眼成像坐标及所述右眼成像坐标计算一视差，并依据所述视差、所述立体摄像机的一内部参数 及一外部参数，将所述方块角点的所述影像坐标转换为一世界坐标；以及
    该所述校正评价单元供以计算相邻的两所述方块角点的所述世界坐标的差值，也供以该所述差值与所述基准值比较出一比对误差，所述比对误差供检测所述立体摄像机的所述内部参数及所述外部参数的准确性。
  7. 如权利要求6所述的立体摄像机的校正品质的检测系统，其特征在于，所述校正评价单元也供将矩形的多个基准影像分别投影至所述左眼影像及所述右眼影像，并提取与所述立体摄像机的一主轴点最邻近的所述方块角点影像，作为一中心基准点，再依序计算各所述基准影像的一基准影像角点与所述中心基准点的所述世界坐标的差值，以分别记录为一畸变误差。
  8. 如权利要求6所述的立体摄像机的校正品质的检测系统，其特征在于，所述特征检测单元也供以判断所述方块角点影像于一时空域的一变异量是否符合一位移门槛值，以判断所述立体摄像机是否符合一稳定状态。
  9. 如权利要求6所述的立体摄像机的校正品质的检测系统，其特征在于，所述显示器的所述校正板影像的所述基准值，系以所述显示器的一荧幕尺寸、一荧幕解析度及所述校正板影像的一方块数量计算而得。
  10. 如权利要求8所述的立体摄像机的校正品质的检测系统，其特征在于，若所述立体摄像机符合所述稳定状态，则所述特征检测单元也供计算于所述左眼影像及所述右眼影像中的所述方块角点影像的一标准差，以判断各所述方块角点影像的特征点分布状态。
PCT/CN2020/128352 2019-11-15 2020-11-12 立体摄像机的校正品质的检测方法及检测系统 WO2021093803A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911119782.X 2019-11-15
CN201911119782.XA CN112822481A (zh) 2019-11-15 2019-11-15 立体摄像机的校正品质的检测方法及检测系统

Publications (1)

Publication Number Publication Date
WO2021093803A1 true WO2021093803A1 (zh) 2021-05-20

Family

ID=75851844

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/128352 WO2021093803A1 (zh) 2019-11-15 2020-11-12 立体摄像机的校正品质的检测方法及检测系统

Country Status (2)

Country Link
CN (1) CN112822481A (zh)
WO (1) WO2021093803A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114302121A (zh) * 2021-12-28 2022-04-08 深圳市火乐科技发展有限公司 图像校正检验方法、装置、电子设备及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104581106A (zh) * 2013-10-16 2015-04-29 深圳市云立方信息科技有限公司 一种对3d相机进行校正的装置及方法
CN106097367A (zh) * 2016-06-21 2016-11-09 北京格灵深瞳信息技术有限公司 一种双目立体相机的标定方法及装置
CN107909616A (zh) * 2017-12-14 2018-04-13 元橡科技（北京）有限公司 一种双目视觉系统的自动矫正方法
WO2019001410A1 (zh) * 2017-06-30 2019-01-03 杭州海康威视数字技术股份有限公司 一种相机参数标定方法、装置、设备及系统
CN109889820A (zh) * 2019-03-14 2019-06-14 深圳博时特科技有限公司 双目摄像头模组的检测方法、装置、存储介质及终端

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104581106A (zh) * 2013-10-16 2015-04-29 深圳市云立方信息科技有限公司 一种对3d相机进行校正的装置及方法
CN106097367A (zh) * 2016-06-21 2016-11-09 北京格灵深瞳信息技术有限公司 一种双目立体相机的标定方法及装置
WO2019001410A1 (zh) * 2017-06-30 2019-01-03 杭州海康威视数字技术股份有限公司 一种相机参数标定方法、装置、设备及系统
CN107909616A (zh) * 2017-12-14 2018-04-13 元橡科技（北京）有限公司 一种双目视觉系统的自动矫正方法
CN109889820A (zh) * 2019-03-14 2019-06-14 深圳博时特科技有限公司 双目摄像头模组的检测方法、装置、存储介质及终端

Also Published As

Publication number Publication date
CN112822481A (zh) 2021-05-18

Similar Documents

Publication Publication Date Title
JP6681470B2 (ja) 奥行き知覚３眼カメラシステム
RU2769303C2 (ru) Оборудование и способ для формирования представления сцены
US20120287287A1 (en) Display screen for camera calibration
CN108063940B (zh) 一种人眼跟踪裸眼3d显示系统的校正系统和方法
JP5633058B1 (ja) ３次元計測装置及び３次元計測方法
CN110809786A (zh) 校准装置、校准图表、图表图案生成装置和校准方法
EP2707838A1 (en) Camera calibration using an easily produced 3d calibration pattern
CN105306922B (zh) 一种深度相机参考图的获取方法和装置
CN109712232B (zh) 一种基于光场的物体表面轮廓三维成像方法
CN109285189B (zh) 一种无需双目同步的直线轨迹快速计算方法
JP2011242183A (ja) 画像処理装置、画像処理方法およびプログラム
JP2011160421A (ja) 立体画像の作成方法、作成装置及びプログラム
TWI712310B (zh) 立體相機之校正品質的檢測方法及檢測系統
CN112489109B (zh) 一种三维成像系统方法、装置及三维成像系统
WO2021093803A1 (zh) 立体摄像机的校正品质的检测方法及检测系统
JP5313187B2 (ja) 立体画像補正装置および立体画像補正方法
CN113411564A (zh) 一种人眼跟踪参数的测量方法、装置、介质及系统
KR20110025083A (ko) 입체 영상 시스템에서 입체 영상 디스플레이 장치 및 방법
KR101275127B1 (ko) 초점 가변 액체 렌즈를 이용한 3차원 이미지 촬영장치 및 방법
KR102547717B1 (ko) 3d 개발엔진을 이용한 다시점 콘텐츠 생성 시뮬레이션 장치 및 그 방법
TW202111666A (zh) 立體相機的校正方法及校正系統
TWM594320U (zh) 立體相機之校正品質的檢測系統
WO2016186211A1 (ja) ３次元計測システム、３次元計測方法及び３次元計測プログラム
JP5453328B2 (ja) 立体撮像システム、補正装置およびそのプログラム
KR102295987B1 (ko) 스테레오 카메라 모듈의 캘리브레이션 방법 및 장치, 및 컴퓨터 판독 가능한 저장매체

Legal Events

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

Ref document number: 20886465

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20886465

Country of ref document: EP

Kind code of ref document: A1