... this paper, six well-known SFS algorithms are implemented and compared. The performance of the algorithms was analyzed on synthetic images using mean and standard deviation of depth (Z) error, mean of surface gradient (p, q) error, and CPU timing. Each algorithm works well for certain images, but performs poorly for others. In general, minimization approaches are more robust, while the other approaches are faster. The implementation of these algorithms in C and images used in this paper are available by anonymous ftp under the pub/tech_paper/survey directory at eustis.cs.ucf.edu (132.170.108.42). These are also part of the electronic version of paper.

A new surface radiance model for diffuse lighting is presented which incorporates shadows, interreflections, and surface orientation. An algorithm is presented that uses this model to compute shape--from--shading under diffuse lighting. The algorithm is tested on both synthetic and real images, and is found to perform more accurately than the only previous algorithm for this problem. Index Terms: shape--from--shading, diffuse lighting, interreflections, shadows, visual events, horizon, skyline I. Introduction The classical formulation of the shape--from--shading problem has been to assume that surface radiance is determined entirely by the surface orientation relative to a point light source at infinity [5], [7]. Shadows and interreflections are usually ignored. When the light source is diffuse, however, this model is not applicable [9], [10]. To see this, consider the following two examples: The first is a scene consisting of a convex Lambertian object resting on a ground plane, il...

Diffuse interreflections cause effects that make current theories of shape from shading unsatisfactory. We show that distant radiating surfaces produce radiosity effects at low spatial frequencies. This means that, if a shading pattern has a small region of support, unseen surfaces in the environment can only produce effects that vary slowly over the support region. It is therefore relatively easy to construct matching processes for such patterns that are robust to interreflections. We call regions with these patterns "shading primitives." Folds and grooves on surfaces provide two examples of shading primitives; the shading pattern is relatively independent of surface shape at a fold or a groove, and the pattern is localised. We show that the pattern of shading can be predicted accurately by a simple model, and derive a matching process from this model. Both groove and fold matchers are shown to work well on images of real scenes.

This paper describes the fully automatic creation of an environment's description using an image-based representation. This representation is a collection of cylindrical sample images combined into an "image-based virtual reality". The locations at which the environment will be sampled are chosen automatically using an operator inspired by models of human visual attention and saccadic motion. The image acquisition is performed by a mobile robot. The selection of vantage points is based on an analysis of the edge structure of sampled panoramic images. In order to trade o# the optimality of the generated description of the navigation e#ort required in solving the on-line problem, a concept referred to as alpha-backtracking is introduced. The paper illustrates sample data acquired by the procedure. Keywords--- mobile robotics, virtual reality, robotics I.

Diffuse interreflections mean that surface shading and shape are related in ways that are difficult to untangle; in particular, distant and invisible surfaces may affect the shading field that one sees. The effects of distant surfaces are confined to relatively low spatial frequencies in the shading field, meaning that we can expect signatures, called shading primitives, corresponding to shape properties. We demonstrate how these primitives can be used to support the construction of useful shape representations. Approaches to this include testing hypotheses of geometric primitives for consistency with the shading field, and looking for shading events that are distinctive of some shape event. We show that these approaches can be composed, leading to an attractive process of representation that is intrinsically bottom up. This representation can be extracted from images of real scenes, and that the representation is diagnostic.

We describe an approach to the automated construction of visual maps of an unknown environment. These maps take the form of image-based "walk-throughs" rather than 2D or 3D models. Our approach is based on the selection of informative viewpoints within the environment. These viewpoints are locations in the environment associated with views containing maximal visual interest. This approach to environment representation is analogous to image compression. Our goal is to obtain a set of representative views resembling those that would be selected by a human observer given the same task. Our computational procedure is inspired by models of human visual attention appearing in the literature on human psychophysics. We make use of the underlying edge structure of a scene, as it is largely una #ected by variations in illumination. Our implementation uses a mobile robot to traverse the environment, and then builds an image-based virtual representation of the environment, only keeping the views whose responses were highest. We demonstrate the e#ectiveness of our attention operator on both single images, and in viewpoint selection within an unknown environment. Keywords: image-based virtual reality, environment representation, visual attention, mobile robotics I.

Most local feature detectors/descriptors implicitly assume that the scene is (locally) planar, an assumption that is violated at surface discontinuities. We show that this restriction is, at least in theory, un-necessary, as one can construct local features that are viewpoint-invariant for generic non-planar scenes. However, we show that any such feature necessarily sacrifices shape information, in the sense of being non shape-discriminative. Finally, we show that if viewpoint is factored out as part of the matching process, rather than explicitly in the representation, then shape is discriminative indeed. We illustrate our theoretical results empirically by showing that, even for simplistic scenes, current affine descriptors fail where even a naive 3-D viewpoint invariant

The computation of surface data based on visual information is an important sub-component in the computer-graphical surface reconstruction of solids and in the control of 3-D environments. Different methodologies can be used for that as, e.g., static stereo, shape from motion, shape from shading, photometric stereo, or structured lightening. There exist different basic approaches in literature often based on simplifying assumptions. However, it is well known that such assumptions may not be true if surface reconstruction is applied under practical circumstances. In this paper, several problems are mentioned which are related to practical applications of surface reconstruction approaches following the methodologies of static stereo, shape from motion, and photometric stereo. We present specific solutions to cope with these problems, or the solution state what was reachable in our work. Some problems are ill-posed and limitations of approaches have to be accepted. As a second contribution of this paper, we discuss the evaluation problem of surface reconstruction algorithms. It is important to answer such questions as 1) For what kind of surfaces and 3-D objects an algorithm behaves either well or bad? 2) How accurate are the reconstruction results of an algorithm under specified circumstances? What measure can be used to evaluate reconstruction accuracy? 3) How to compare reconstruction results following different methodologies? 4) What algorithm can be suggested for a specific application project? and so on. So far we present some proposals and first quantitative or qualitative results for answering such questions. In our opinion a methodology for evaluating surface reconstruction algorithms is still at its beginning. However a critical evaluation of poten...

How may the shape of a smooth surface be inferred from an image? Traditional methods in computer vision for inferring shape-from-shading assume that surface shading depends entirely on surface orientation. In many illumination scenarios, however, shading may occur independently of surface orientation. For example, when an extended light source such as the sky casts a continuous shadow on a flat ground, the resulting shading is due to illumination variations only. In this thesis, a new approach to shape-from-shading is taken in which shading variations are attributed entirely to spatially varying illumination. This leads to a new analysis of the shape-from-shading problem, and to a new algorithm for solving it. In particular, a model of spatially varying illumination is developed which is in terms of the set of light rays in free space. This set is shown to be a four dimensional smooth manifold, called the radiation manifold. Local transformations between coordinate systems on this mani...

While virtual reality is a powerful tool for a range of applications, it has the following two associated overheads that fundamentally limit its usefulness: (1) The creation of realistic synthetic virtual environment models is di#cult and labour intensive; (2) The computing resources needed to render realistic complex environments in real time are substantial. In this paper, we describe an approach to the fully automated creation of image based virtual reality (VR) models: collections of panoramic images (cylindrical or spherical images) that illustrate an environment. Traditionally, a key bottleneck for this kind of modelling is the selection and acquisition of sample data. Our approach is based on using a small mobile robot to navigate in the environment and collect the image data of interest. A critical issue is selecting the appropriate sample locations of the modelling process: this is addressed using a computational mechanism that resembles human attention. Our objective is to select regions that di#er from the surrounding environment. We do this using statistical properties of the output of an edge operator. Specifically, we guide a camera-carrying mobile robot through an environment and have it acquire data with which we construct a VR model. We then demonstrate the e#ectiveness of our approach using real data. Keywords: virtual reality, image registration, visual attention, autonomous mobile robotics, attention, interest operator, environment modelling 1.