A Method of Obtaining the Relative Positions of 4 Points from 3 Perspective Projections

نویسنده

  • Hugh Christopher Longuet-Higgins
چکیده

According to Ullman's Structure-from-Motion Theorem [U79], three orthogonal projections of four points in a rigid non-planar configuration uniquely determine their structure, and the relative orientations of the three views, up to a reflection in the image plane. It is here shown that a corresponding result holds for the more general "para-perspective" case, and leads to a rapidly convergent algorithm for the fully perspective case. Unless the four points are nearly coplanar, or the images closely similar, the output of this algorithm is not unduly sensitive to errors in the image coordinates. 1 The orthogonal case We adopt one of the points PQ as origin and denote the 3D coordinates of the others by (Xn , Yn , Zn), (Xn ' ,Yn ' ,Zn ') and (Xn",Yn",Zn") in the three projection frames, (Xn, Yn), (Xn' Yn') and (Xn",Yn") being the (relative) image coordinates and Z ^ Z n ' and Zn" the (relative) depth coordinates. Then there will exist rigid rotation matrices U = [uy] and V = [vy] such that X n ' = u l l x n + 12n + 13n' x n " = v l l x n + v 1 2 Y n + 13n' (1\1") Yn ' = «21 n + "22Yn + u 2 3 Z n , Yn" = v 2 1 X n + v 2 2 Y n + v^Z , , , (2\2") Z n = "31n + 32n + U33n. z n " = 31n + 32n + The problem is to find U and V and the depth coordinates from the three sets of image coordinates. As each rotation involves 3 unknown parameters, and there are 9 depth coordinates, we have 18 equations for only 15 unknowns, and may expect to encounter 3 consistency conditions, useful for checking purposes. Elimination of Zj, between (I) and (2') gives the three equations x n " 3 2 Y n u 3 l + x n ' u 2 3 Y n ' u 1 3 = 0, (n = 1, 2, 3), (4) BMVC 1991 doi:10.5244/C.5.12 87 from which one can obtain the ratios of the "border elements" U32 : U31 : U23 : U13. (The computation fails if either (i) the four points are coplanar, in which case equations (4) are no longer independent, or (ii) the Z and Z axes coincide, in which case U33 = ± 1 and all four border elements vanish.) For U to be a rigid rotation of the first frame into the second, its border elements must satisfy U 1 3 2 + U 23 2 ( l ~ 33) = U 3 1 2 + 32This is one of the three consistency conditions mentioned above, and can be checked as soon as the ratios U32 : U31 : U23 : n^ have been obtained from (4). Introducing the normalized quaternion Q = ip + jq + kr + s, p+q+r+s = 1, (5) related to U by the equation u l l 12 13 p-<i-r+s, 2(pq-rs) , 2(pr + qs) U(Q) = u 2 1 u 2 2 «23 = (Pq + s). -p 2 +q 2 r 2 +s 2 , 2(qr-ps) (6) 31 32 33 2(pr-qs), 2(qr + ps), -p 2 -q 2 +r 2 +s 2 we see that the ratios U32 :113^ : U23 : UJ3 determine the ratio of p to q and the ratio of r to s, but not the ratio of p to r. It follows that if Q is written in the parametric form Q = (i sin A + j cos A) sin C + (k sin B + cos B) cos C, (7) then the two images (Xn, Yn) and (Xn',Yn') yield the values of the two parameters A and B, but not the "vergence" parameter C (equal to half of the angle between the Z and Z' axes). To compute C we need all three images, and the A and B parameters of the rotations connecting them, which we now denote by Uj (= U), U2 (= V~*) and U 3 (=VU~ 1 ) , satisfying (see figure at top of next page):

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Some Conditions for Characterizing Minimum Face in Non-Radial DEA Models with Undesirable Outputs

The problem of utilizing undesirable (bad) outputs in DEA models often need replacing the assumption of free disposability of outputs by weak disposability of outputs. The Kuosmanen technology is the only correct representation of the fully convex technology exhibiting weak disposability of bad and good outputs. Also, there are some specific features of non-radial data envelopment analysis (DEA...

متن کامل

Initial blank design of deep drawn orthotropic materials using inverse finite element method

In this work, an inverse finite element formulation was modified for considering material anisotropy in obtaining blank shape and forming severity of deep drawn orthotropic parts. In this procedure, geometry of final part and thickness of initial blank sheet were known. After applying ideal forming formulations between material points of initial blank and final shape, an equation system was obt...

متن کامل

Introducing an Optimized Method for Obtaining X-ray Diffraction Patterns of Biological Tissues

Introduction Individual X-Ray diffraction patterns of biological tissues are obtained via interference of coherent scattering with their electrons. Many scientists have distinguished normal and cancerous breast tissue, bone density, and urinary stone types using the X-Ray diffraction patterns resulting from coherent scattering. The goal of this study was to introduce an optimized method for obt...

متن کامل

Simulation Studies for the Performance Analysis of the Reconstruction of a Line in 3-D from Two Arbitrary Perspective Views Using Two Plane Intersection Method

Objects in nature as well as man made are bound by planar as well as curvilinear surfaces. Hence the reconstruction process of three dimensional objects involves obtaining the geometrical attributes of these planar and curvilinear surfaces, lines, points etc. If a line in 3-D space is viewed from two arbitrary positions, two different images of the same line are obtained. The correspondence bet...

متن کامل

A novel Local feature descriptor using the Mercator projection for 3D object recognition

Point cloud processing is a rapidly growing research area of computer vision. Introducing of cheap range sensors has made a great interest in the point cloud processing and 3D object recognition. 3D object recognition methods can be divided into two categories: global and local feature-based methods. Global features describe the entire model shape whereas local features encode the neighborhood ...

متن کامل

A model based, anatomy dependent method for ultra-fast creation of primary SPECT projections

  Introduction: Monte Carlo (MC) is the most common method for simulating virtual SPECT projections. It is useful for optimizing procedures, evaluating correction algorithms and more recently image reconstruction as a forward projector in iterative algorithms; however, the main drawback of MC is its long run time. We introduced a model based method considering the eff...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 1991