W. M. Silver. Determining shape and reflectance using multiple images. Master's thesis, MIT, Cambridge, MA, 1980.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of shape from shading is that reflectance has to be uniform or known. In the photometric stereo paradigm, by using multiple images, each taken with the same viewpoint, but a di#erent illuminant, it is possible to recover a non uniform albedo, as well as the surface orientation. In early work [ 33, 35, 18 ] illuminants needed to be known. A linear method to recover Lambertian surface albedo from three images, each taken with a di#erent point light source, has been proposed [ 1 ] Instead of relying on an explicit knowledge of the three illuminants, this method, in the spirit of the shape from ....

W. Silver. Determining shape and reflectance using multiple images. PhD thesis, MIT, Cambridge, MA, 1990.

....pointing unit surface normal n(x; y) From multiple images of the object seen from a fixed viewpoint but with different light source direction, we can solve Eq. 4 for b when the light source strengths and directions are known. This, of course, is the standard photometric stereo technique, see [8, 24, 29]. However, if the light source strengths and directions are not known, then we can only determine the vector field b(x; y) of surface normals and albedos up to a 3 Theta 3 linear transformation. For any invertible 3 Theta 3 linear transformation A 2 GL(3) 7, 3, 21] b T s = Ab) T A ....

W. Silver. Determining Shape and Reflectance Using Multiple Images. PhD thesis, MIT, Cambridge, MA, 1980.

....and, as we will describe, robust estimation can be used to remove shadows as outliers. Our approach has been strongly influenced by the linear lighting models suggested on both theoretical [28] and experimental [13, 5] grounds. A second important source has been the photometric stereo literature [30], 41] 33] 18] The closest work, which we will refer to throughout the paper, is the application of SVD to this problem by Hayakawa [15] and the recent work by 3 Fan and Wolff [8] Both consider the situation with spatially varying albedo and unknown light sources. Hayakawa uses SVD in ....

....the GBR. Finally, in section (4) we demonstrate an iterative algorithm which is able to improve the analysis by finding and rejecting shadows. 2 Lambertian Lighting Models and SVD In this section we describe the SVD approach to lighting [15] It builds on previous work on photometric stereo [30], 41] 33] 18] and on linear lighting models for objects [13] 28] See also [31] for the use of linear re rendering for interactive lighting design. Suppose we have a set of images generated by a Lambertian model where the lighting conditions vary. We use x to label positions in the image ....

[Article contains additional citation context not shown here]

W. Silver. Determining Shape and Reflectance Using Multiple Images. PhD Thesis. MIT, Cambridge, MA. 1980.

.... correspondence, namely scenes that are highly textured (i.e. where image features are not sparse) or scenes that completely lack texture (i.e. where there are insufficient image features) At the core of our approach for generating novel viewpoints is a variant of photometric stereo [27, 29, 13, 12, 30] which simultaneously estimates geometry and albedo across the scene. However, the main limitation of classical photometric stereo is that the light source positions must be accurately known, and this necessitates a fixed lighting rig as might be possible in an industrial setting. Instead, the ....

W. Silver. Determining Shape and Reflectance Using Multiple Images. PhD thesis, MIT, Cambridge, MA, 1980.

....on s 2 IR 3 . Therefore, in the absence of self shadowing, given three images of a Lambertian surface from the same viewpoint taken under three known, linearly independent light source directions, the albedo and surface normal can be recovered; this is the well known method of photometric stereo [50, 46]. Alternatively, one can reconstruct the image of the surface under a novel lighting direction by a linear combination of the three original images [43] In other words, if the surface is purely Lambertian and there is no shadowing, then all images under varying illumination lie within a 3 D ....

W.M. Silver. Determining Shape and Reflectance Using Multiple Images. PhD thesis, MIT, Cambridge, MA, 1980.

....of specular reflection, they are able to subtract it from the images to provide intrinsic Lambertian images for analysis. A four light illumination has also been used by Solomon and Ikeuchi [123] to determine the roughness of a specularly reflecting Torrance Sparrow [134] type surface. Silver [122] has also developed ways of applying the basic photometric stereo method to surfaces of differing reflectance properties. This work also showed that photometric stereo could be conducted using experimentally measured reflectance maps. Chapter 2. Related Work 14 The similar concept of photometric ....

W. M. Silver. Determining shape and reflectance using multiple images. Master's thesis, Department of Electrical Engineering and Computer Science, MIT, Cambridge, MA, 1980.

.... often written as: I(x; y) a(x; y)n(x; y) Delta s j b(x; y) Delta s; 1) where a(x; y) is the albedo of the object, n(x; y) is its surface normal, b(x; y) j a(x; y)n(x; y) and s is the light source direction (the light is assumed to be at infinity) If this equation applies then it is clear [26], 28] 25] 20] that the space of images of the object, as the light source direction changes, spans a three dimensional subspace. In other words, any image of the object can be expressed as: I(x; y) 3 X i=1 ff i b i (x; y) 2) 6 In fact it can shown to lie within a convex cone inside ....

....lie on a two dimensional grid but it is convenient to represent them as a vector) Our first scheme assumes that we have multiple images of the object 7 and the light sources are known. This is of least interest since it is a strong assumption and corresponds to standard photometric stereo [26, 28, 14, 17], though with nonconstant albedo. We investigated this scheme mainly to test the Lambertian asumptions about our data. We concluded that the model is a good approximation though robust techniques are needed to reduce the influence of shadows and specularities. If, however, there are multiple ....

[Article contains additional citation context not shown here]

W. Silver. Determining Shape and Reflectance Using Multiple Images. PhD Thesis. MIT, Cambridge, MA. 1980.

....on s 2 IR 3 . Therefore, in the absence of self shadowing, given three images of a Lambertian surface from the same viewpoint taken under three known, linearly independent light source directions, the albedo and surface normal can be recovered; this is the well known method of photometric stereo [26, 30]. Alternatively, one can reconstruct the image of the surface under an arbitrary lighting direction by a linear combination of the three original images, see [25] For classification, this fact has great importance: It shows that for a fixed viewpoint, all images of a Lambertian surface lie in a ....

W. Silver. Determining Shape and Reflectance Using Multiple Images. PhD thesis, MIT, Cambridge, MA, 1980.

....pointing unit surface normal n(x; y) From multiple images of the object seen from a fixed viewpoint but with different light source direction, we can solve Eq. 4 for b when the light source strengths and directions are known. This, of course, is the standard photometric stereo technique, see [6, 20, 24]. However, if the light source strengths and directions are not known, then we can only determine the vector field b(x; y) of surface normals and albedos up to a 3 Theta 3 linear transformation. For any invertible 3 Theta 3 linear transformation A 2 GL(3) 2, 5, 17] b T s = b T A T A ....

W. Silver. Determining Shape and Reflectance Using Multiple Images. PhD thesis, MIT, Cambridge, MA, 1980.

....values of p and q. This permits us to solve for the gradient. Becanse of the nonline rity of the equations, however, a number of solutions may be found at times. In this case, a third image may be used to disambiguate the remaining possibilities. Tiffs is the principle of plotometric stereo [ 3, . rthographic projection can simplify the calculation considerably. If we can assume that tile object is small compared with the distance to the source and the image forming system, then tile viewer direction can be approximated as the axis of the image forming system, and we can treat the ....

....the approximation fixes tile viewing direction. So we can, for example, rotate the source keeping the phase angle constant (The phase angle is the angle between the source and the viewer, measured at tile object) This means that we can obtain a new reflectance map just by rotating the old ote [3]. 2 Basic Consideration 2.1 Relationship between Radiance and Irradiance One of tile main points of our discussion here is that we consider only specular components of reflectance when we calculate the reflectance map, since many industrial mater,als are made of metal and have strong: ....

B. I(. P. Horn, R.J. Woodham, and W, Silver: "Determining Shape and Reflectance Using Multiple Images," AI-Memo 't90, 1978, AI Lab, MIT.

....surface orientation might be given for points on the surface on some regular sampling grid. This grid may conveniently correspond to the picture cells in an image. Such a simple representation is called a needle map (Figure 1) 13] Photometric stereo produces surface descriptions in this form [7, 12, 15, 28, 32, 33]. This representation too is not directly helpful when it comes to comparing surfaces of objects that may be rotated relative to one another (Both depth maps and needle maps depend on the position of the object as well as its attitude) The extended Gaussian image, on the other hand, does make it ....

Horn, B. K. P., R. J. Woodham, and W. M. Silver (1978) "Determining Shape and Reflectance using Multiple Images," M.I.T.A.I. Laboratory Memo 490, August.

No context found.

W. M. Silver. Determining shape and reflectance using multiple images. Master's thesis, MIT, Cambridge, MA, 1980.

No context found.

Silver, W.M.: Determining shape and reflectance using multiple images. Master's thesis, MIT, Cambridge, MA (1980)

No context found.

B. K. P. Horn, R. J. Woodham, and W. M. Silver. Determining shape and reflectance using multiple images. Technical Report 490, Artificial Intelligence Lab Memo, MIT, Cambridge, MA 02139, 1978.

No context found.

W. Silver. Determining Shape and Reflectance Using Multiple Images. PhD thesis, MIT, Cambridge, MA, 1980.

No context found.

W. M. Silver. Determining shape and reflectance using multiple images. Master's thesis, MIT, Cambridge, MA, 1980.

No context found.

Silver, W.M.: Determining shape and reflectance using multiple images. Master's thesis, MIT, Cambridge, MA (1980)

No context found.

W.M. Silver, Determining Shape and Reflectance Using Multiple Images, PhD thesis, Massachusetts Institute of Technology, 1980.

No context found.

B. K. P. Horn, R. J. Woodham, and W. M. Silver. Determining shape and reflectance using multiple images. Technical Report 490, Artificial Intelligence Lab Memo, MIT, Cambridge, MA 02139, 1978.

No context found.

W.M. Silver, Determining Shape and Reflectance Using Multiple Images, PhD thesis, MIT, Cambridge, MA, 1980.

No context found.

W. Silver. Determining shape and reflectance using multiple images. PhD thesis, MIT, Cambridge, MA, 1990.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC