CN113034572B - Epipolar extraction method based on eight-parameter epipolar model - Google Patents

Epipolar extraction method based on eight-parameter epipolar model Download PDF

Info

Publication number
CN113034572B
CN113034572B CN202110460890.4A CN202110460890A CN113034572B CN 113034572 B CN113034572 B CN 113034572B CN 202110460890 A CN202110460890 A CN 202110460890A CN 113034572 B CN113034572 B CN 113034572B
Authority
CN
China
Prior art keywords
epipolar
parameter
epipolar line
parameters
equation
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
Application number
CN202110460890.4A
Other languages
Chinese (zh)
Other versions
CN113034572A (en
Inventor
余俊鹏
沈刚
李洁明
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.)
Guangdong University of Technology
Original Assignee
Guangdong University of Technology
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 Guangdong University of Technology filed Critical Guangdong University of Technology
Priority to CN202110460890.4A priority Critical patent/CN113034572B/en
Publication of CN113034572A publication Critical patent/CN113034572A/en
Application granted granted Critical
Publication of CN113034572B publication Critical patent/CN113034572B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/60Analysis of geometric attributes
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/70Determining position or orientation of objects or cameras
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10004Still image; Photographic image
    • G06T2207/10012Stereo images

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Geometry (AREA)
  • Image Processing (AREA)

Abstract

基于八参数核线模型的核线提取方法,属于摄影测量与计算机视觉领域。本发明解决了现有近似核线方法提取的核线结果精度较低、核线提取过程较复杂的问题。本发明方法包括:获取立体影像数据;在立体影像数据中匹配同名点,记录同名点的像点坐标;根据同名点像点坐标建立八参数核线方程;根据八参数核线方程确定八参数的初值;对八参数核线方程线性化得到误差方程;基于误差方程,利用最小二乘法建立法方程,根据法方程求出八参数的修正量;判断修正量是否满足收敛要求,若收敛,则根据修正量对八参数进行修正,得到修正后的八参数;通过修正后的八参数计算生成目标区域立体影像的核线。本发明用于立体影像核线提取。

Figure 202110460890

The invention discloses an epipolar line extraction method based on an eight-parameter epipolar line model, which belongs to the field of photogrammetry and computer vision. The invention solves the problems that the accuracy of the epipolar line extracted by the existing approximate epipolar line method is low and the epipolar line extraction process is relatively complicated. The method of the present invention comprises: obtaining stereoscopic image data; matching the same-name point in the stereoscopic image data, and recording the image point coordinates of the same-name point; establishing an eight-parameter epipolar line equation according to the same-name point image point coordinates; determining the eight-parameter epipolar line equation according to the eight-parameter epipolar line equation Initial value; linearize the eight-parameter epipolar equation to obtain the error equation; based on the error equation, use the least square method to establish the normal equation, and obtain the correction amount of the eight parameters according to the normal equation; judge whether the correction amount meets the convergence requirement, and if it converges, then The eight parameters are corrected according to the correction amount to obtain the corrected eight parameters; the epipolar line of the stereoscopic image of the target area is calculated and generated through the corrected eight parameters. The invention is used for epipolar line extraction of stereo images.

Figure 202110460890

Description

基于八参数核线模型的核线提取方法Epipolar line extraction method based on eight-parameter epipolar line model

技术领域technical field

本发明涉及基于八参数核线模型的核线提取方法。属于摄影测量与计算机视觉领域。The invention relates to an epipolar line extraction method based on an eight-parameter epipolar line model. It belongs to the field of photogrammetry and computer vision.

背景技术Background technique

在摄影测量中,一般利用单幅影像无法确定物体上点的空间位置,只能确定物点所在摄影光线的空间方向,要获得物点的空间位置一般需利用两幅相互重叠的影像构成立体像对,它是立体摄影测量的基本单元,由其构成的立体模型是摄影测量三维定位的基础。In photogrammetry, it is generally impossible to determine the spatial position of a point on an object by using a single image, but only the spatial direction of the photographic light where the object point is located. To obtain the spatial position of an object point, it is generally necessary to use two overlapping images to form a stereoscopic image Yes, it is the basic unit of stereo photogrammetry, and the stereo model formed by it is the basis of photogrammetry three-dimensional positioning.

核线是摄影测量的一个基本概念。摄影测量学者Helava等在20世纪70年代初提出核线相关理论后,核线的作用在摄影测量领域很快得到重视。在摄影测量学中,摄影基线与地面任意点所构成的平面为核面,一个核面与具有立体重叠度的左右影像的交线为同名核线,在单中心投影的立体影像上,同名像点严格位于同名核线上。在影像匹配中,最重要的几何约束条件就是核线约束。利用核线约束在影像匹配中可以将同名点的搜索范围从二维降为一维,大大减少同名点搜索空间,提高影像同名点匹配效率和准确性。在多视影像匹配和计算机立体视觉中,核线约束条件对于保证密集匹配处理的可靠性起到不可替代的作用。Epipolar line is a basic concept of photogrammetry. After the photogrammetry scholar Helava et al. put forward the theory of epipolar line in the early 1970s, the role of epipolar line was quickly paid attention to in the field of photogrammetry. In photogrammetry, the plane formed by the photographic baseline and any point on the ground is the epipolar plane, and the intersection line between a nuclear plane and the left and right images with three-dimensional overlap is the epipolar line of the same name. The points lie strictly on the epipolar line of the same name. In image matching, the most important geometric constraints are epipolar constraints. The use of epipolar constraints in image matching can reduce the search range of homonymous points from two dimensions to one dimension, greatly reducing the search space of homonymous points, and improving the efficiency and accuracy of image homonymous point matching. In multi-view image matching and computer stereo vision, epipolar constraints play an irreplaceable role in ensuring the reliability of dense matching processing.

卫星摄影测量技术是一种空间对地观测技术,主要采用线阵推扫式传感器获取地表几何信息,在地形测绘、资源调查、数字城市、国防建设中应用广泛。线阵卫星影像的核线生成方法一直是摄影测量与遥感领域的难点。由于线阵影像每一行均对应着不同的外方位元素,立体影像间不存在唯一的摄影基线,不适用经典核线理论。已有研究表明,线阵立体影像的核线是类似双曲线的曲线,仅在小范围内可看作直线。目前卫星线阵影像提取近似核线的方法以投影轨迹法及其多种衍生方法为主。投影轨迹法在已知左右影像定向参数基础上,求左影像目标点a在右影像上的投影轨迹线即核曲线l′,a的同名点a′必在l′上,得出a与l′的点线对应关系。进一步研究表明,右影像核曲线l′上的邻近点b′,c′对应的左影像核曲线l1,l2很接近,可用同一曲线l替代,从而可建立局部范围内l与l′的线线对应关系。基于投影轨迹法的衍生方法有基于投影基准面的线阵推扫式卫星立体像对近似核线影像生成方法、沿水平像面的核线影像生成方法等。Satellite photogrammetry technology is a space-to-earth observation technology, which mainly uses linear array push-broom sensors to obtain surface geometric information, and is widely used in terrain surveying and mapping, resource investigation, digital cities, and national defense construction. The generation method of epipolar lines of linear array satellite images has always been a difficult point in the field of photogrammetry and remote sensing. Since each line of the linear array image corresponds to a different outer orientation element, there is no unique photographic baseline between stereoscopic images, and the classical epipolar theory does not apply. Existing studies have shown that the epipolar line of a linear stereoscopic image is a curve similar to a hyperbola, which can only be regarded as a straight line in a small range. At present, the methods for extracting approximate epipolar lines from satellite line array images are mainly based on the projection trajectory method and its various derivative methods. Based on the known orientation parameters of the left and right images, the projected trajectory method calculates the projected trajectory line of the target point a in the left image on the right image, that is, the nuclear curve l′. ’ corresponding relationship between dots and lines. Further studies show that the left image kernel curves l 1 and l 2 corresponding to the adjacent points b' and c' on the right image kernel curve l' are very close, and can be replaced by the same curve l, so that the relationship between l and l' in a local range can be established Correspondence between lines. The derivative methods based on the projection trajectory method include the method of generating the approximate epipolar image of the line array push-broom satellite stereo image pair based on the projection reference plane, and the generation method of the epipolar image along the horizontal image plane.

尽管以上方法提取的近似核线在核线影像匹配中能够起到核线约束的作用,但存在以下问题:Although the approximate epipolar line extracted by the above method can play the role of epipolar line constraint in epipolar line image matching, there are the following problems:

(1)核线影像与原始影像的坐标转换关系复杂,且依赖于影像的严格几何模型或者有理多项式函数模型,而这些模型参数有时并不可得,模型参数求解也存在一定困难。(1) The coordinate transformation relationship between the epipolar image and the original image is complicated, and depends on the strict geometric model of the image or the rational polynomial function model, and these model parameters are sometimes not available, and there are certain difficulties in solving the model parameters.

(2)不同于单中心投影影像严格的核线,近似核线存在一定的上下视差,且对不同卫星线阵影像所生成核线的精度存在差异。(2) Different from the strict epipolar line of the monocentric projection image, the approximate epipolar line has a certain upper and lower parallax, and the accuracy of the epipolar line generated by different satellite linear array images is different.

鉴于现有近似核线方法的不足,亟需提出一种精度更高、方便易用、普适性好的核线模型及核线提取方法,为摄影测量和计算机立体视觉数据处理提供技术支撑。In view of the deficiencies of the existing approximate epipolar methods, it is urgent to propose a epipolar model and epipolar extraction method with higher accuracy, convenience and universality, so as to provide technical support for photogrammetry and computer stereo vision data processing.

发明内容Contents of the invention

本发明是为了解决现有近似核线方法提取的核线结果精度较低、核线提取过程较复杂的问题。现提供基于八参数核线模型的核线提取方法。The invention aims to solve the problems of low accuracy of the epipolar line extracted by the existing approximate epipolar line method and complicated epipolar line extraction process. An epipolar line extraction method based on an eight-parameter epipolar line model is now provided.

基于八参数核线模型的核线提取方法,包括以下步骤:The epipolar line extraction method based on the eight-parameter epipolar line model comprises the following steps:

步骤一、获取覆盖同一目标区域的立体影像数据;Step 1. Obtain stereoscopic image data covering the same target area;

步骤二、在立体影像上匹配若干均匀分布的同名点,并记录同名点的像点坐标;Step 2, matching some evenly distributed homonymous points on the stereoscopic image, and recording the image point coordinates of the homonymous points;

步骤三、根据同名点像点坐标建立目标在左右影像的同名点的八参数核线方程;Step 3, establish eight parameter epipolar equations of the target at the same name point of the left and right images according to the image point coordinates of the same name point;

步骤四、根据八参数核线方程确定八参数的初值;Step 4, determine the initial value of eight parameters according to the eight-parameter epipolar equation;

步骤五、对八参数核线方程线性化处理,得到误差方程;Step 5, linearize the eight-parameter epipolar equation to obtain the error equation;

步骤六、基于误差方程,利用最小二乘法建立法方程,根据法方程求出八参数的修正量;Step 6, based on the error equation, utilize the least squares method to establish the normal equation, and obtain the correction amount of the eight parameters according to the normal equation;

步骤七、判断修正量是否满足收敛要求,若收敛,则根据修正量对八参数进行修正,得到修正后的八参数,若不收敛,则将修正后的八参数代替八参数的初值,执行步骤三至步骤六,直至收敛;Step 7. Determine whether the correction amount meets the convergence requirement. If it converges, modify the eight parameters according to the correction amount to obtain the corrected eight parameters. If not, replace the corrected eight parameters with the initial value of the eight parameters and execute Step 3 to Step 6 until convergence;

步骤八、通过修正后的八参数计算生成目标区域立体影像的核线。Step 8: Generate the epipolar line of the stereoscopic image of the target area through the calculation of the modified eight parameters.

有益效果Beneficial effect

(1)核线精度高。本发明基于线阵影像核线的倾角变化规律而提出,测试表明八参数核线模型提取核线的上下视差中误差可小于0.5像元,与基于投影轨迹法的常用核线模型相比精度可提高30%到70%,能够充分满足立体影像同名点几何约束要求。(1) The epipolar line has high precision. The present invention is proposed based on the variation law of the inclination angle of the epipolar line of the linear array image. The test shows that the error in the upper and lower parallax of the epipolar line extracted by the eight-parameter epipolar line model can be less than 0.5 pixels, and the accuracy can be compared with the common epipolar line model based on the projection trajectory method. The improvement is 30% to 70%, which can fully meet the geometric constraint requirements of the same-named point of the stereoscopic image.

(2)简便易用。八参数核线模型的参数求解只需少量同名连接点,无需已知任何影像成像模型参数。从原始影像到核线影像的坐标转换计算简单,对后续核线影像重采样处理十分有利。(2) Easy to use. The parameter solution of the eight-parameter epipolar model only needs a small number of connection points with the same name, and no imaging model parameters need to be known. The calculation of the coordinate conversion from the original image to the epipolar image is simple, which is very beneficial to the resampling of the subsequent epipolar image.

(3)适用范围广。八参数核线模型可同时适用于单中心投影和多中心投影影像的核线提取,可应用于航天、航空、近景摄影测量及计算机立体视觉处理。(3) The scope of application is wide. The eight-parameter epipolar model can be applied to the epipolar extraction of single-center projection and multi-center projection images at the same time, and can be applied to aerospace, aviation, close-range photogrammetry and computer stereo vision processing.

附图说明Description of drawings

图1为本发明方法流程图;Fig. 1 is a flow chart of the method of the present invention;

图2为影像核线精度对比图。Figure 2 is a comparison chart of image epipolar accuracy.

具体实施方式detailed description

具体实施方式一:参照图1具体说明本实施方式,本实施方式基于八参数核线模型的核线影像生成方法,包括:Specific embodiment 1: This embodiment is specifically described with reference to FIG. 1 . This embodiment is based on an epipolar image generation method of an eight-parameter epipolar model, including:

步骤一、获取覆盖同一目标区域的立体影像数据;Step 1. Obtain stereoscopic image data covering the same target area;

步骤二、在立体影像数据中匹配若干(不少于4个)均匀分布的同名点,并记录同名点的像点坐标;Step 2, matching some (not less than 4) evenly distributed homonymous points in the stereoscopic image data, and recording the image point coordinates of the homonymous points;

步骤三、根据同名点像点坐标建立目标在左右影像的同名点的八参数核线方程;Step 3, establish eight parameter epipolar equations of the target at the same name point of the left and right images according to the image point coordinates of the same name point;

步骤四、根据八参数核线方程确定八参数的初值;Step 4, determine the initial value of eight parameters according to the eight-parameter epipolar equation;

步骤五、对八参数核线方程线性化处理,得到误差方程;Step 5, linearize the eight-parameter epipolar equation to obtain the error equation;

步骤六、基于误差方程,利用最小二乘法建立法方程,根据法方程求出八参数的修正量;Step 6, based on the error equation, utilize the least squares method to establish the normal equation, and obtain the correction amount of the eight parameters according to the normal equation;

步骤七、判断修正量是否满足收敛要求,若收敛,则根据修正量对八参数进行修正,得到修正后的八参数,若不收敛,则将修正后的八参数代替八参数的初值,执行步骤三至步骤六,直至收敛。Step 7. Determine whether the correction amount meets the convergence requirement. If it converges, then correct the eight parameters according to the correction amount to obtain the corrected eight parameters. If not, replace the corrected eight parameters with the initial value of the eight parameters and execute Step 3 to Step 6 until convergence.

步骤八、通过修正后的八参数计算生成目标区域立体影像的核线。Step 8: Generate the epipolar line of the stereoscopic image of the target area through the calculation of the modified eight parameters.

具体实施方式二:本实施方式与具体实施方式一不同的是,所述步骤三中八参数核线方程通过如下公式表示:Specific embodiment two: the difference between this embodiment and specific embodiment one is that the eight-parameter epipolar equation in the step three is expressed by the following formula:

0=xL′-xR′ (1)其中,0=x L ′-x R ′ (1) where,

Figure BDA0003042189980000031
Figure BDA0003042189980000031

Figure BDA0003042189980000032
Figure BDA0003042189980000032

其中,(xL,yL)表示目标在左立体影像上的列坐标和行坐标;(xR,yR)表示目标在右立体影像上的列坐标和行坐标;(xL′,yL′)表示目标在左核线影像上的列坐标和行坐标;(xR′,yR′)表示目标在右核线影像上的列坐标和行坐标;a1、b1、c1、a2、b2、c2为右核线影像相对于右原始影像的仿射变换参数;α0表示过左立体影像上某参考点的左核线倾角,αv表示左核线倾角的变化率;Among them, (x L , y L ) represents the column coordinates and row coordinates of the target on the left stereo image; (x R , y R ) represents the column coordinates and row coordinates of the target on the right stereo image; (x L ′, y L ′) represents the column and row coordinates of the target on the left epipolar image; (x R ′, y R ′) represents the column and row coordinates of the target on the right epipolar image; a 1 , b 1 , c 1 , a 2 , b 2 , c 2 are the affine transformation parameters of the right epipolar image relative to the right original image; α 0 represents the inclination angle of the left epipolar line passing through a reference point on the left stereo image, and α v represents the inclination angle of the left epipolar line rate of change;

其中,α0、αv、a1、b1、c1、a2、b2、c2为左右同名核线纠正的八参数,为核线模型待求参数;八参数的初值用向量表示为:Among them, α 0 , α v , a 1 , b 1 , c 1 , a 2 , b 2 , c 2 are the eight parameters corrected by the left and right epipolar lines with the same name, which are the parameters to be sought for the epipolar line model; the initial values of the eight parameters are vectors Expressed as:

Figure BDA0003042189980000041
Figure BDA0003042189980000041

本实施方式中,当核线方向在列方向上时,目标在左核线影像上的列坐标和行坐标为(xL′,yL′);若核线方向在行方向上时,左核线影像上的列坐标和行坐标为(xL″,yL″),表示为:In this embodiment, when the epipolar direction is in the column direction, the column and row coordinates of the target on the left epipolar image are (x L ', y L '); if the epipolar direction is in the row direction, the left epipolar The column and row coordinates on the line image are (x L ″, y L ″), expressed as:

Figure BDA0003042189980000042
Figure BDA0003042189980000042

其它步骤及参数与具体实施方式一相同。Other steps and parameters are the same as those in Embodiment 1.

具体实施方式三:本实施方式与具体实施方式一或二不同的是,所述步骤四中的误差方程为:Specific embodiment three: the difference between this embodiment and specific embodiment one or two is that the error equation in the step four is:

V=Bx-l,P (5)V=Bx-l,P (5)

其中,x=[Δα0 Δαv Δa1 Δb1 Δc1 Δa2 Δb2 Δc2]T为八参数的修正量;B为式(1)对八参数的偏导系数矩阵;l为不符值向量,即将八参数初值X0代入式(1)的值;V为改正数向量;P为权矩阵。Among them, x=[Δα 0 Δα v Δa 1 Δb 1 Δc 1 Δa 2 Δb 2 Δc 2 ] T is the correction amount of the eight parameters; B is the partial derivative coefficient matrix of the formula (1) to the eight parameters; l is the non-conforming value vector , which is to substitute the initial value X 0 of the eight parameters into the value of formula (1); V is the correction number vector; P is the weight matrix.

其它步骤及参数与具体实施方式一或二相同。Other steps and parameters are the same as those in Embodiment 1 or Embodiment 2.

具体实施方式四:本实施方式与具体实施方式一至三之一不同的是,所述步骤五基于误差方程,利用最小二乘法建立法方程;根据法方程求解八参数的修正量;具体过程为:Specific embodiment four: this embodiment is different from one of specific embodiments one to three in that said step five is based on the error equation, and utilizes the least squares method to establish a normal equation; solve the correction amount of eight parameters according to the normal equation; the specific process is:

按最小二乘原理建立法方程:Establish the normal equation according to the principle of least squares:

BTPBx-BTPl=0 (6)B T PBx - B T Pl = 0 (6)

其中,T表示转置;Among them, T means transpose;

令:make:

NBB=BTPB,W=BTPlN BB = B T PB, W = B T Pl

可简写成Can be abbreviated as

NBBx-W=0(7) NBB xW=0(7)

式中,NBB为满秩矩阵,此时x有唯一解亦是最优解,

Figure BDA0003042189980000043
或x=(BTPB)-1BTPl。In the formula, N BB is a full-rank matrix, and at this time, x has a unique solution and is also an optimal solution.
Figure BDA0003042189980000043
Or x=(B T PB ) -1 B T Pl.

其它步骤及参数与具体实施方式一至三之一相同。Other steps and parameters are the same as those in Embodiments 1 to 3.

具体实施方式五:本实施方式与具体实施方式一至四之一不同的是,所述步骤六根据八参数的修正量对八参数进行修正;具体过程为:Embodiment 5: This embodiment is different from Embodiment 1 to Embodiment 4 in that the step 6 corrects the eight parameters according to the correction amount of the eight parameters; the specific process is:

X=X0+x (8)X=X 0 +x (8)

其中,X表示修正后的八参数。Among them, X represents the modified eight parameters.

其它步骤及参数与具体实施方式一至四之一相同。Other steps and parameters are the same as in one of the specific embodiments 1 to 4.

具体实施方式六:本实施方式与具体实施方式一至五之一不同的是,所述收敛要求为:Embodiment 6: The difference between this embodiment and one of Embodiments 1 to 5 is that the convergence requirement is:

设定一个阈值10-8,若x的每个参数改正数绝对值小于阈值,根据此时X的数值计算八参数;Set a threshold of 10 -8 , if the absolute value of each parameter correction number of x is less than the threshold, calculate the eight parameters according to the value of X at this time;

若x中每个参数的改正数的绝对值均大于等于阈值,则令X=X0,重复步骤三至步骤六。If the absolute value of the correction number of each parameter in x is greater than or equal to the threshold, set X=X 0 , and repeat steps 3 to 6.

其它步骤及参数与具体实施方式一至五之一相同。Other steps and parameters are the same as one of the specific embodiments 1 to 5.

实施例Example

为了验证本发明方法的有效性,本发明采用仿真数据进行核线模型参数估计实验。分别利用Pleiades、SPOT-5、QuickBird和ZY3卫星影像附带的RPC参数,模拟生成1000×1000、2000×2000、3000×3000、4000×4000和5000×5000像素影像范围内均匀分布的同名点对,使用八参数核线得到不同卫星影像在不同影像范围内得到的上下视差的中误差。In order to verify the effectiveness of the method of the present invention, the present invention uses simulation data to conduct epipolar model parameter estimation experiments. Using the RPC parameters attached to the Pleiades, SPOT-5, QuickBird, and ZY3 satellite images, simulate and generate uniformly distributed point pairs with the same name in the range of 1000×1000, 2000×2000, 3000×3000, 4000×4000 and 5000×5000 pixel images, Using the eight-parameter epipolar line to obtain the median error of the upper and lower parallax obtained by different satellite images in different image ranges.

表1试验卫星线阵立体影像参数表Table 1 Test satellite linear array stereoscopic image parameter list

Figure BDA0003042189980000051
Figure BDA0003042189980000051

八参数核线模型结果见表2,各影像核线精度对比如图2所示。The results of the eight-parameter epipolar model are shown in Table 2, and the comparison of the epipolar accuracy of each image is shown in Figure 2.

表2通过八参数核线模型获得的上下视差中误差Table 2 The error in the upper and lower disparity obtained by the eight-parameter epipolar model

Figure BDA0003042189980000052
Figure BDA0003042189980000052

从表2可以得出,不同的卫星立体影像在八参数模型中得到的上下视差中误差随影像范围扩大而递增。四种影像的核线上下视差中误差比较结果为:ZY3<SPOT-5<Pleiades<QuickBird。其中,ZY3、Pleiades、SPOT-5和QuickBird卫星影像数据在影像范围为5000×5000时,上下视差中误差分别为0.0015像素、0.1216像素、0.0255像素和0.7883像素。It can be concluded from Table 2 that the error in the upper and lower parallax obtained by the eight-parameter model of different satellite stereoscopic images increases with the expansion of the image range. The comparison results of the errors in the parallax above and below the epipolar line of the four images are: ZY3<SPOT-5<Pleiades<QuickBird. Among them, when the image range of ZY3, Pleiades, SPOT-5 and QuickBird satellite image data is 5000×5000, the errors in the upper and lower parallax are 0.0015 pixels, 0.1216 pixels, 0.0255 pixels and 0.7883 pixels respectively.

应该说明的是,对于本领域的普通技术人员,依据本发明的思想,在具体实施方式及应用范围上均会有改变之处,而所有这些改进和变换都应属于本发明所附权利要求的保护范围。It should be noted that, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation and application scope, and all these improvements and transformations should belong to the appended claims of the present invention. protected range.

Claims (8)

1. The epipolar line extraction method based on the eight-parameter epipolar line model is characterized by comprising the following steps of:
step one, acquiring three-dimensional image data covering the same target area;
matching a plurality of uniformly distributed homonymy points on the stereo image, and recording image point coordinates of the homonymy points;
step three, establishing an eight-parameter epipolar line equation of the target at the homonymous points of the left image and the right image according to the homonymous point image point coordinates, specifically:
0=x L ′-x R ′ (1)
wherein:
Figure FDA0003758125940000011
Figure FDA0003758125940000012
wherein (x) L ,y L ) Representing column coordinates and row coordinates of the target on the left stereoscopic image; (x) R ,y R ) The column coordinate and the row coordinate of the target on the right original image are represented; (x) L ′,y L ') column coordinates and row coordinates of the object on the left epipolar line image; (x) R ′,y R ') column coordinates and row coordinates of the object on the right epipolar line image; a is 1 、b 1 、c 1 、a 2 、b 2 、c 2 Affine transformation parameters of the right epipolar line image relative to the right original image; alpha is alpha 0 Indicating the inclination, alpha, of the left epipolar line passing through a reference point on the left stereoscopic image v Representing the rate of change of the left epipolar line dip;
determining initial values of the eight parameters according to an eight-parameter epipolar equation;
step five, carrying out linearization processing on the eight-parameter epipolar equation to obtain an error equation;
step six, establishing a normal equation by using a least square method based on an error equation, and solving correction quantity of eight parameters according to the normal equation;
step seven, judging whether the correction quantity meets the convergence requirement, if so, correcting the eight parameters according to the correction quantity to obtain corrected eight parameters, if not, replacing the initial values of the eight parameters with the corrected eight parameters, and executing the steps three to six until convergence;
and step eight, calculating and generating an epipolar line of the three-dimensional image of the target area through the corrected eight parameters.
2. The epipolar line extraction method based on an eight-parameter epipolar line model according to claim 1, wherein the homonymy points in the second step are not less than four.
3. The epipolar extraction method based on the eight-parameter epipolar model according to claim 2, wherein the initial values of the eight parameters are specifically:
α 0 、α v 、a 1 、b 1 、c 1 、a 2 、b 2 、c 2 eight parameters corrected for left and right homonymic epipolar lines the initial values of the eight parameters are expressed in vector form as:
Figure FDA0003758125940000013
4. the method for extracting epipolar lines based on an eight-parameter epipolar line model according to claim 3, wherein the error equation in the fourth step is expressed as:
V=Bx-l,P (4)
wherein, the vector is a correction number vector; x is correction quantity with eight parameters; b is a partial derivative coefficient matrix of eight parameters by an eight-parameter epipolar equation; l is an inconsistency vector, namely substituting the initial value X0 of the eight parameters into the value of an eight-parameter epipolar equation; p is a weight matrix.
5. The epipolar line extraction method based on an eight-parameter epipolar line model according to claim 4, wherein the eight-parameter correction quantity x is expressed as:
x=[Δα 0 Δα v Δa 1 Δb 1 Δc 1 Δaa 2 Δb 2 Δc 2 ]T。
6. the epipolar line extraction method based on the eight-parameter epipolar line model according to claim 5, wherein the fifth step is based on an error equation, a normal equation is established by using a least square method, and an eight-parameter correction quantity is obtained according to the normal equation; the specific process is as follows:
establishing a normal equation:
B T PBx-B T Pl=0 (5)
wherein T represents transpose;
let N BB =B T PB,W=B T Pl, then the equation of the law is abbreviated as:
N BB x-W=0 (6)
wherein N is BB Representing a full rank matrix, in this case eight parametersHas a unique solution to the correction quantity x of (c),
Figure FDA0003758125940000021
or x = (B) T PB) -1 B T Pl。
7. The epipolar line extraction method based on an eight-parameter epipolar line model according to claim 6, wherein step six corrects the eight parameters according to correction amounts of the eight parameters; the specific process is as follows:
X=X 0 +x (7)
wherein X represents the eight parameters after correction.
8. The epipolar line extraction method based on an eight-parameter epipolar line model according to claim 7, wherein the convergence requirement is:
setting a threshold value 10 -8 If the absolute value of the correction number of each parameter of X is smaller than the threshold value, calculating eight parameters according to the numerical value of X at the moment;
if the absolute value of the correction number of each parameter in X is greater than or equal to the threshold value, let X = X 0 And repeating the third step to the sixth step.
CN202110460890.4A 2021-04-27 2021-04-27 Epipolar extraction method based on eight-parameter epipolar model Active CN113034572B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110460890.4A CN113034572B (en) 2021-04-27 2021-04-27 Epipolar extraction method based on eight-parameter epipolar model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110460890.4A CN113034572B (en) 2021-04-27 2021-04-27 Epipolar extraction method based on eight-parameter epipolar model

Publications (2)

Publication Number Publication Date
CN113034572A CN113034572A (en) 2021-06-25
CN113034572B true CN113034572B (en) 2022-12-23

Family

ID=76454718

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110460890.4A Active CN113034572B (en) 2021-04-27 2021-04-27 Epipolar extraction method based on eight-parameter epipolar model

Country Status (1)

Country Link
CN (1) CN113034572B (en)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106023147A (en) * 2016-05-06 2016-10-12 西安航天天绘数据技术有限公司 GPU-based linear array remote sensing image DSM rapid extraction method

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101598550B (en) * 2009-07-03 2011-06-15 武汉大学 Rapid rearrangement method of satellite stereoscopic image approximate epipolar line based on projection trajectory
CN101915913B (en) * 2010-07-30 2012-09-19 中交第二公路勘察设计研究院有限公司 Steady automatic matching method for high-resolution satellite image connecting points
CN103630120B (en) * 2013-07-16 2016-09-14 中国人民解放军信息工程大学 Martian surface linear array image core line method for resampling based on tight geometric model
CN106991643B (en) * 2017-03-13 2020-08-25 广东南方数码科技股份有限公司 Real-time line checking method and real-time line checking system with low resource consumption

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106023147A (en) * 2016-05-06 2016-10-12 西安航天天绘数据技术有限公司 GPU-based linear array remote sensing image DSM rapid extraction method

Also Published As

Publication number Publication date
CN113034572A (en) 2021-06-25

Similar Documents

Publication Publication Date Title
CN111126148B (en) DSM (digital communication system) generation method based on video satellite images
CN103823981B (en) A kind of satellite image block adjustment method of digital elevation model auxiliary
CN107705329B (en) High-resolution optical satellite staring image registration method based on geometric constraints
CN110223222B (en) Image stitching method, image stitching device, and computer-readable storage medium
CN109709551B (en) Area network plane adjustment method for satellite-borne synthetic aperture radar image
CN110660099B (en) Rational function model fitting method for remote sensing image processing based on neural network
CN111003214B (en) Attitude and Orbit Refinement Method of Domestic Land Observation Satellite Based on Cloud Control
CN112270698B (en) Non-rigid geometric registration method based on nearest curved surface
CN106910238A (en) Color texture method for reconstructing based on high inclination-angle close-range image
CN111724465A (en) Method and device for satellite image adjustment based on plane constraints to select virtual control points
CN104048649B (en) A kind of multi-view images and the rapid registering method of threedimensional model
CN118089666A (en) Photogrammetry method and system suitable for low-overlapping-degree unmanned aerial vehicle images
CN113379648B (en) High-resolution seven-number and resource three-number stereoscopic image joint adjustment method
CN112070891B (en) Image area network adjustment method and system using digital ground model as three-dimensional control
CN101598785A (en) A Rational Function Imaging Model Generation Method for Satellite Imagery Products at All Levels
CN115719320B (en) Tilt correction dense matching method based on remote sensing image
CN113947638A (en) Image orthorectification method for fisheye camera
CN106980601A (en) The high-precision method for solving of basis matrix based on three mesh epipolar-line constraints
CN109029379B (en) High-precision small-base-height-ratio three-dimensional mapping method
CN115713548A (en) Automatic registration method for multi-stage live-action three-dimensional model
CN106251349A (en) A kind of SAR stereoscopic image dense Stereo Matching method
CN113034572B (en) Epipolar extraction method based on eight-parameter epipolar model
CN107330934A (en) The boundling Adjustable calculation method and system of low dimensional
CN103644897B (en) A 3D measurement method based on super-resolution image reconstruction
CN114998496B (en) A method for rapid generation of orthophotos based on scene aerial images and sparse point clouds

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