We present a technique for building a three-dimensional description of a static scene from a dense sequence of images. These images are taken in such rapid succession that they form a solid block of data in which the temporal continuity from image to image is approximately equal to the spatial continuity in an individual image. The technique utilizes knowledge of the camera motion to form and analyze slices of this solid. These slices directly encode not only the three-dimensional positions of objects, but also such spatiotemporal events as the occlusion of one object by another. For straight-line camera motions, these slices have a simple linear structure that makes them easier to analyze. The analysis computes the threedimensional positions of object features, marks occlusion boundaries on the objects, and builds a threedimensional map of "free space. " In our article, we first describe the application of this technique to a simple camera motion, and then show how projective duality is used to extend the analysis to a wider class of camera motions and object types that include curved and moving objects. 1

this paper we present & new method for separating and recovering the motion and shape of multiple independently moving objects in sequence of images. The method does not require prior knowledge of the number of objects, nor is dependent on any grouping of features into an object at the image level. For this purpose, we introduce a mathematical construct of object shapes, called the shape interaction matrix, which is invariant to both the object motions and the selection of coordinate systems. This invariant structure is computable solely from the observed trajectories of image features without grouping them into individual objects

The structure-from-motion problem has been extensively studied in the field of computer vision. Yet, the bulk of the existing work assumes that the scene contains only a single moving object. The more realistic case where an unknown number of objects move in the scene has received little attention, especially for its theoretical treatment. In this paper we present a new method for separating and recovering the motion and shape of multiple independently moving objects in a sequence of images. The method does not require prior knowledge of the number of objects, nor is dependent on any grouping of features into an object at the image level. For this purpose, we introduce a mathematical construct of object shapes, called the shape interaction matrix, which is invariant to both the object motions and the selection of coordinate systems. This invariant structure is computable solely from the observed trajectories of image features without grouping them into individual objects. Once the matr...

Scene flow is the three-dimensional motion field of points in the world, just as optical flow is the twodimensional motion field of points in an image. Any optical flow is simply the projection of the scene flow onto the image plane of a camera. In this paper, we present a framework for the computation of dense, non-rigid scene flow from optical flow. Our approach leads to straightforward linear algorithms and a classification of the task into three major scenarios: (1) complete instantaneous knowledge of the scene structure, (2) knowledge only of correspondence information, and (3) no knowledge of the scene structure. We also show that multiple estimates of the normal flow cannot be used to estimate dense scene flow directly without some form of smoothing or regularization. 1

We show how to automatically acquire similarity-invariant shape representations of objects from noisy image sequences under weak perspective. The proposed method is linear and incremental, requiring no more than pseudo-inverse. It is based on the observation that the trajectories that points on the object form in weak-perspective image sequences are linear combinations of three of the trajectories themselves, and that the coefficients of the linear combinations represent shape in an affine-invariant basis. A nonlinear but numerically sound preprocessing stage is added to improve the accuracy of the results even further. Experiments show that attention to noise and computational techniques improve the shape results substantially with respect to previous methods proposed for ideal images. 1 Introduction In model-based recognition, images are matched against stored libraries of three-dimensional object representations, so that a good match implies recognition of the object. The recogniti...

Abstract. We cast the problem of motion segmentation of feature trajectories as linear manifold finding problems and propose a general framework for motion segmentation under affine projections which utilizes two properties of trajectory data: geometric constraint and locality. The geometric constraint states that the trajectories of the same motion lie in a low dimensional linear manifold and different motions result in different linear manifolds; locality, by which we mean in a transformed space a data and its neighbors tend to lie in the same linear manifold, provides a cue for efficient estimation of these manifolds. Our algorithm estimates a number of linear manifolds, whose dimensions are unknown beforehand, and segment the trajectories accordingly. It first transforms and normalizes the trajectories; secondly, for each trajectory it estimates a local linear manifold through local sampling; then it derives the affinity matrix based on principal subspace angles between these estimated linear manifolds; at last, spectral clustering is applied to the matrix and gives the segmentation result. Our algorithm is general without restriction on the number of linear manifolds and without prior knowledge of the dimensions of the linear manifolds. We demonstrate in our experiments that it can segment a wide range of motions including independent, articulated, rigid, non-rigid, degenerate, non-degenerate or any combination of them. In some highly challenging cases where other state-of-the-art motion segmentation algorithms may fail, our algorithm gives expected results. 2 1

Invariance under a group of 3D transformations seems a desirable component of an efficient 3D shape representation. We propose representations which are invariant under weak perspective to either rigid or affine 3D transformations, and we show how they can be computed efficiently from a sequence of images with a linear and incremental algorithm. We show simulated results with perspective projection and noise, and the results of model acquisition from a real sequence of images. The use of linear computation, together with the integration through time of invariant representations, offers improved robustness and stability. Using these invariant representations, we derive model-based projective invariant functions of general 3D objects. We discuss the use of the model-based invariants with existing recognition strategies: alignment without transformation, and constant time indexing from 2D images of general 3D objects.

This paper describes methods for recovering time-varying shape and motion of non-rigid 3D objects from uncalibrated 2D point tracks. For example, given a video recording of a talking person, we would like to estimate the 3D shape of the face at each instant, and learn a model of facial deformation. Time-varying shape is modeled as a rigid transformation combined with a non-rigid deformation. Reconstruction is ill-posed if arbitrary deformations are allowed, and thus additional assumptions about deformations are required. We first suggest restricting shapes to lie within a lowdimensional subspace, and describe estimation algorithms. However, this restriction alone is insufficient to constrain reconstruction. To address these problems, we propose a reconstruction method using a Probabilistic Principal Components Analysis (PPCA) shape model, and an estimation algorithm that simultaneously estimates 3D shape and motion for each instant, learns the PPCA model parameters, and robustly fills-in missing data points. We then extend the model to model temporal dynamics in object shape, allowing the algorithm to robustly handle severe cases of missing data.

This paper examines the construction of a 3-D surface model of an object rotating in front of a camera. Previous research in depth from motion has demonstrated the power of using an incremental approach to depth estimation. In this paper, we extend this approach to more general motion and use a full 3-D surface model instead of a 2 1 = 2 -D sketch. The algorithm starts with a flow field computed using local correlation. It then projects individual measurements into 3-D points with associated uncertainties. Nearby points from successive frames are merged to improve the position estimates. These points are then used to construct a finite element surface model, which is itself refined over time. We demonstrate the application of our new techniques to several real image sequences. Keywords: Computer vision, 3-D model construction, image sequence (motion) analysis, optic flow, Kalman filter, surface interpolation, computer aided design, computer graphics animation. c flDigital Equipment C...

We introduce an objective shape-identification task for measuring the kinetic depth effect (KDE). A rigidly rotating surface consisting of hills and valleys on an otherwise flat ground was defined by 300 randomly positioned dots. On each trial, 1 of 53 shapes was presented; the observer's task was to identify the shape and its overall direction of rotation. Identification accuracy was an objective measure, with a low guessing base rate, of the observer's perceptual ability to extract 3D structure from 2D motion via KDE. (1) Objective accuracy data were consistent with previously obtained subjective rating judgments of depth and coherence. (2) Along with motion cues, rotating real 3D dot-defined shapes inevitably produced a cue of changing dot density. By shortening dot lifetimes to control dot density, we showed that changing density was neither necessary nor sufficient to account for accuracy; motion alone sufficed. (3) Our shape task was solvable with motion cues from the 6 most relevant locations. We extracted the dots from these locations and used them in a simplified 2D direction-labeling motion task with 6 perceptually flat flow fields. Subjects ' performance in the 2D and 3D tasks was equivalent, indicating that the information processing capacity of KDE is not unique. (4) Our proposed structure-from-motion algorithm for the shape task first finds relative minima and maxima of local velocity and then assigns 3D depths proportional to velocity. In 1953, Wallach and O'Connell described a depth percept