Luke Olsen

Technical Section: Sketch-based modeling: A survey

User interfaces in modeling have traditionally followed the WIMP (Window, Icon, Menu, Pointer) paradigm. Though functional and very powerful, they can also be cumbersome and daunting to a novice user, and creating a complex model requires considerable expertise and effort. A recent trend is toward more accessible and natural interfaces, which has lead to sketch-based interfaces for modeling (SBIM). The goal is to allow sketches-hasty freehand drawings-to be used in the modeling process, from rough model creation through to fine detail construction. Mapping a 2D sketch to a 3D modeling operation is a difficult task, rife with ambiguity. To wit, we present a categorization based on how a SBIM application chooses to interpret a sketch, of which there are three primary methods: to create a 3D model, to add details to an existing model, or to deform and manipulate a model. Additionally, in this paper we introduce a survey of sketch-based interfaces focused on 3D geometric modeling applications. The canonical and recent works are presented and classified, including techniques for sketch acquisition, filtering, and interpretation. The survey also provides an overview of some specific applications of SBIM and a discussion of important challenges and open problems for researchers to tackle in the coming years.

NaturaSketch: Modeling from Images and Natural Sketches

Sketching on paper is a quick and easy way to communicate ideas. However, many sketch-based systems require people to draw in contrived ways instead of sketching freely as they would on paper. NaturaSketch affords a more natural interface through multiple strokes that overlap, cross, and connect. It also features a meshing algorithm to support multiple strokes of different classifications, which lets users design complex 3D shapes from sketches drawn over existing images. To provide a familiar workflow for object design, a set of sketch annotations can also specify modeling and editing operations. NaturaSketch empowers designers to produce a variety of models quickly and easily.

Technical Section: Multiresolution for curves and surfaces based on constraining wavelets

We present a novel method for determining local multiresolution filters for a broad range of subdivision schemes. Our approach is based on constraining the wavelet coefficients such that the coefficients at even vertices can be computed from the coefficients of neighboring odd vertices. This constraint leads to an initial set of decomposition filters. To increase the quality of these initial filters, we use an optimization that reduces the size of the wavelet coefficients. The resulting multiresolution filters yield a biorthogonal wavelet system whose construction is similar to the lifting scheme. This approach is demonstrated in depth for cubic B-spline curves and Loop subdivision surfaces. Our filters are shown to perform comparably with existing filters.

A Taxonomy of Modeling Techniques using Sketch-Based Interfaces

Traditional user interfaces in modeling have followed the WIMP (Window, Icon, Menu, Pointer) paradigm. While functional and powerful, they can also be cumbersome and daunting to a novice user; creating a complex model requires much expertise and effort. A recent trend is toward more accessible and natural interfaces, which has lead to sketch-based interfaces for modeling (SBIM). The goal is to allow hand-drawn sketches to be used in the modeling process, from rough model creation through to fine detail construction. Mapping 2D sketches to a 3D modeling operation is a difficult and ambiguous task, so our categorization is based on how an SBIM application chooses to interpret a sketch, of which there are three primary methods: to create a 3D model, to add details to an existing model, or to deform and manipulate a model. In this STAR, we present a taxonomy of sketchbased interfaces focused on geometric modeling applications. The canonical and recent works are presented and classified, including techniques for sketch acquisition, filtering, and interpretation. The report also includes a discussion of important challenges and open problems for researchers to tackle in the coming years.

Stroke extraction and classification for mesh inflation

We provide a method for extracting and classifying stroke segments from a line drawing or sketch with the goal of producing perceptually-valid output in the context of mesh inflation. This is important as processing freehand sketch input is a fundamental task in sketch-based interfaces, yet many systems bypass the problem by forcing simplified, unnatural drawing patterns. Our stroke extraction combines contour tracing with feature-preserving post-processing. The extracted strokes are classified according to the objects and regions in the sketch: object and region boundaries, interior features, and suggestive lines. The outcome of this classification is demonstrated with examples in feature-sensitive mesh inflation.

Fast stroke matching by angle quantization

Determining similarity of two point sequences (strokes) is a fundamental task in gestural interfaces. Because the length of each stroke is arbitrary, mapping to a fixed-dimension feature space is often done to allow for direct comparison. In this paper, we propose a new feature space based on angle quantization. For each adjacent pair of points in a stroke, the vector between them defines an angle relative to a fixed axis. The sequence of these angles can be mapped to a k-dimensional feature space by quantizing the unit circle into k ranges, and taking a normalized count of the number of stroke angles in each range. The Euclidean distance between strokes in this feature space gives a measure of stroke similarity. The measure is scale invariant, and some degree of rotational invariance can be achieved with slight modification. Our method is shown to offer efficient and accurate gestural matching performance compared to traditional signal-processing and image-based methods.

A Discrete Approach to Multiresolution Curves and Surfaces

Subdivision surfaces have been widely adopted in modeling in part because they introduce a separation between the surface and the underlying basis functions. Such a separation allows for simple schemes that work on general topology surfaces. Multiresolution representations based on subdivision, however, incongruently return to continuous functional spaces in their construction and analysis. In this paper, we investigate a discrete approach to multiresolution construction for a variety of subdivision schemes, based only on the subdivision rules. Noting that a compact representation can only afford to store a subset of the detail information, our construction enforces a constraint between locally adjacent detail terms. In this way, all detail information is recoverable for reconstruction, and a decomposition approach is implied by the constraint. The construction is demonstrated with case studies in Dyn-Levin-Gregory curves and Catmull-Clark surfaces, each of which our method produces results as good as earlier methods. It is further shown that our construction can be interpreted as biorthogonal wavelet systems.

Content description servers for networked video surveillance

Advances in digital signal processing technology make it possible to embed the low-level functions performed by a video surveillance system into cameras. The result is metadata cameras, like MPEG-7 cameras, that provide descriptions of what they see in the form of XML documents served over a network. We use this concept to build video information servers that are conceptually similar to MPEG-7 cameras, but differ in that they interact with client applications and can be configured dynamically to perform different functions. We demonstrate the use of these video information servers in some simple surveillance applications. Dynamic configuration of the servers allows them to do more than describe video content. For example, we present a functioning surveillance system, built with content description servers, that describes the activity in a hockey game based on the observations of several cameras. Thus, we demonstrate that content description servers provide building blocks with which to create networked video surveillance systems.

Internet broadcast of hockey: A scale prototype

We present a system for the broadcast of hockey games over the internet. The system allows users to experience the hockey game while it is in progress. Our system uses generic content description servers that acquire information from an external source, process it, and serve the processed data to client systems. Dynamic configuration of the servers allows us to use them in a variety of roles. For example, video information servers, like an MPEG-7 camera, produce XML documents that describe the motion of objects in the scene in addition to unprocessed video. Unlike an MPEG-7 camera, our video information servers interact with client systems, and can change their behavior through dynamic configuration. In an alternate configuration, a content description server acts as a game server in our hockey broadcast system. The game server forms an environment model that encapsulates the state of the hockey game and serves data from the model to clients. We developed and tested our system using a 1/32-scale model of a hockey rink. Early results using data acquired at a real rink indicate that the system performs as expected.

Subjective trajectory characterization: acquisition, matching, and retrieval

We describe a system that automatically tracks moving objects in a scene and subjectively characterizes the object trajectories for storage and retrieval. A multi-target color-histogram particle filter combined with besthypothesis data association is the foundation of our trajectory acquisition algorithm. To improve computational performance, we use quasi-Monte-Carlo methods to reduce the number of particles required by each filter. The tracking system operates in real-time to produce a stream of XML documents that contain the object trajectories. To characterize trajectories subjectively, we form a set of shape templates that describes basic maneuvers (e.g., gentle turn right, hard turn left, straight line). Procrustes shape analysis provides a scaleand rotation-invariant mechanism to identify occurrences of these maneuvers within a trajectory. To add spatial information to our trajectory representation, we partition the two-dimensional space under surveillance into a set of mutually exclusive regions. A temporal sequence of region-to-region transitions gives a spatial representation of the trajectory. The shape and position descriptions combine to form a compact, high-level representation of a trajectory. We provide similarity measures for the shape, position, and combined shape and position representations.