We describe an experimental interface that anticipates a user's intentions and accommodates predicted changes in advance. Our canonical example is an interactive version of ``magnetic poetry'' in which rectangular blocks containing single words can be juxtaposed to form arbitrary sentences or ``poetry.'' The user can rearrange the blocks at will, forming and dissociating word sequences. A crucial attribute of the blocks in our system is that they anticipate insertions and gracefully rearrange themselves in time to make space for a new word or phrase. The challenges in creating such an interface are three fold: 1) the user's intentions must be inferred from noisy input, 2) arrangements must be altered smoothly and intuitively in response to anticipated changes, and 3) new and changing goals must be handled gracefully at any time, even in mid animation. We describe a general approach for handling the dynamic creation and deletion of organizational goals. Fluid motion is achieved by continually applying and correcting goal-directed forces to the objects. Future applications of this idea include the manipulation of text and graphical elements within documents and the manipulation of symbolic information such as equations.

Fish tank virtual reality (FTVR) systems use eye tracking to present perspective projections to a user that are tailored to his or her view position. These perspective views can create the illusion of a world behind the display screen-a virtual fish tank. However, the illusion is not perfect "users do not feel that the virtual objects are real. Despite this, users clearly are getting some indication of depth. What exactly are they perceiving? We review previous work in FTVR and identify key differences between the virtual world and the real world. We show that we still have an incomplete understanding of perception in fish tank virtual reality and that some important questions may be impossible to answer.

We describe a system for freehand entry and editing of mathematical expressions using a pen and tablet. The expressions are entered in the same way that they would be written on paper. The system interprets the results and generates output in a form suitable for use in other applications, such as word processors or symbolic manipulators. Interpretation includes character segmentation, character recognition, and formula parsing. Our interface incorporates easy to use tools for correcting interpretation errors at any stage. The user can also edit the handwritten representation and ask the system to reinterpret the results. By recovering the formula's structure directly from its handwritten form, the user is free to use common conventions of mathematical notation without regard to internal representation. We report the results of a small user study, which indicate that the new style of interaction is effective.

Computer vision is a very broad and rapidly changing field. A major challenge for teachers of this subject is to cover a large body of basic theory while at the same time giving a good sense of state-of-the-art research. Traditionally our fourth year undergraduate course in computer vision was taught using lectures and textbooks. We report on the results of two different alterations of the course, both of which focus on our students' own research into recent primary source literature. In the first alteration we used a seminar-based format. It was a qualified success. The students found the course interesting, gained a high-level understanding of the state-of-the-art in computer vision, and developed critical analysis skills. However, despite the fact that they worked hard, student learning of details of the techniques suffered. They were able to complete the course without a good understanding of the theoretical and mathematical foundations of computer vision. In response to this, our second alteration was to base the course around six problem-based learning units. The students benefitted from the reduced set of topics and higher level of practical work.

Long image sequences provide a wealth of information, which means that a compact representation is needed to efficiently process them. In this paper a novel representation for motion segmentation in long image sequences is presented. This representation – the feature interval graph – measures the pairwise rigidity of features in the scene. The feature interval graph is re-cursively computed, making it a compact representation, and uses an interval model of uncertainty. The feature interval graph forms the basis for new al-gorithms for motion segmentation and occlusion analysis. Results of these algorithms are presented on synthetic and laboratory scenes. 1

We describe a new sketching interface in which shape recognition and morphing are tightly coupled. Raw input strokes are continuously morphed into ideal geometric shapes, even before the pen is lifted. By means of smooth and continual shape transformations the user is apprised of recognition progress and the appearance of the final shape, yet always retains a sense of control over the process. At each time t the system uses the trajectory traced out thus far by the pen coupled with the current appearance of the time-varying shape to classify the sketch as one of several pre-defined basic shapes. The recognition operation is performed using shape-specific fits based on least-squares or relaxation, which are continuously updated as the user draws. We describe the time-dependent transformation of the sketch, beginning with the raw pen trajectory, using a family of first-order ordinary differential equations that depend on time and the current shape of the sketch. Using this formalism, we describe several possible behaviors that result by varying the relative significance of new and old portions of a stroke, changing the “viscosity” of the morph, and enforcing different end conditions. A preliminary user study suggests that the new interface is particularly effective for rapidly constructing diagrams consisting of simple shapes.

In the past, evaluation of the effectiveness of optical flow algorithms on real scenes has been difficult due to a lack of complex test data for which the correct flow is known. We give a detailed description of how a simple ray tracing technique can be used to generate ground truth motion fields for the evaluation of optical flow algorithms on scenes of arbitrary complexity.

We present several methods for converting solid models to volume models whose boundaries are defined by an iso-surface. By storing samples of a signed distance function at each point in the volume, we show that we can reconstruct the original object with greater accuracy than methods based on an inclusion function. Our methods are fast, and require no more memory than conventional point sampling. These approaches apply to all solid objects for which ray intersection and point classification methods are defined. If a space subdivision test is available, we can quickly sample complex objects. We also present enhancements that allow for sampling of the object's original colour and texture.

We introduce virtual snakes for generating occlusion hypotheses. Initially, snakes are clustered based on their motion to form object hypotheses-a type of motion segmentation. When two snakes intersect, four virtual snakes are generated-a background and a foreground snake for each of the original two. The two foreground virtual snakes are allowed to relax, while the two background virtual snakes move in accordance with their previous motion. The combined energies of the snakes in the two colliding objects are examined after the collision to determine the occlusion relationship, and the inconsistent virtual snakes are deleted. We show that this heuristic can be used to correctly track objects in the presence of strong occlusion.

Evaluating the performance of optical flow algorithms has been difficult because of the lack of ground-truth data sets for complex scenes. We describe a simple modification to a ray tracer that allows us to generate ground-truth motion fields for scenes of arbitrary complexity. The resulting flow maps are used to assist in the comparison of eight optical flow algorithms using three complex, synthetic scenes. Our study found that a modified version of Lucas and Kanade's algorithm has superior performance but produces sparse flow maps. Proesmans et al.'s algorithm performs slightly worse, on average, but produces a very dense depth map. 1 Introduction Seventeen years have passed since Horn and Schunck published their influential paper on the calculation of optical flow [4]. Since then, a substantial amount of research has been devoted to finding ways to calculate optical flow more efficiently and more accurately. Optical flow extraction has been proposed as a preprocessing step...