Ioana M. Boier-Martin

Parameterization of triangle meshes over quadrilateral domains

We present a method for parameterizing irregularly triangulated input models over polyhedral domains with quadrilateral faces. A combination of center-based clustering techniques is used to generate a partition of the model into regions suitable for remeshing. Several issues are addressed: the size and shape of the regions, their positioning with respect to features of the input geometry, and the amount of distortion introduced by approximating each region with a coarse polygon. Region boundaries are used to define a coarse polygonal mesh which is quadrangulated to obtain a parameterization domain. Constraints can be optionally imposed to enforce a strict correspondence between input and output features. We use the parameterization for multiresolution Catmull-Clark remeshing and we illustrate two applications that take advantage of the resulting representation: interactive model editing and texture mapping.

Differentiable parameterization of Catmull-Clark subdivision surfaces

Subdivision-based representations are recognized as important tools for the generation of high-quality surfaces for Computer Graphics. In this paper we describe two parameterizations of Catmull-Clark subdivision surfaces that allow a variety of algorithms designed for other types of parametric surfaces (i.e., B-splines) to be directly applied to subdivision surfaces. In contrast with the natural parameterization of subdivision surfaces characterized by diverging first order derivatives around extraordinary vertices of valence higher than four, the derivatives associated with our proposed methods are defined everywhere on the surface. This is especially important for Computer-Aided Design (CAD) applications that seek to address the limitations of NURBS-based representations through the more flexible subdivision framework.

Adaptive graphics

This article presents the idea of a unifying framework that allows visual representations of information to be customized and mixed together into new ones. The net result is a fine-grained approach to representing data, better suited to accessing and rendering it over networks. Although the focus is on geometric models and 3D shape representations, many issues discussed are relevant to network-based visualization in general.

Domain decomposition for multiresolution analysis

This paper describes a method for converting an arbitrary mesh with irregular connectivity to a semi-regular multiresolution representation. A shape image encoding geometric and differential properties of the input model is computed. Standard image processing operations lead to an initial decomposition of the model that conforms to its salient features. A triangulation step performed on the resulting partition in image space, followed by resampling and multiresolution analysis in object space, complete the procedure. The conversion technique is automatic, takes into account surface properties for deriving a base domain, and is computationally efficient as the bulk of the processing is carried out in image space. Besides domain decomposition, our image-based approach to handling geometry may be used in the context of related applications, including model simplification, remeshing, and wireframe generation.

Using predictive analysis to improve invoice-to-cash collection

It is commonly agreed that accounts receivable (AR) can be a source of financial difficulty for firms when they are not efficiently managed and are underperforming. Experience across multiple industries shows that effective management of AR and overall financial performance of firms are positively correlated. In this paper we address the problem of reducing outstanding receivables through improvements in the collections strategy. Specifically, we demonstrate how supervised learning can be used to build models for predicting the payment outcomes of newly-created invoices, thus enabling customized collection actions tailored for each invoice or customer. Our models can predict with high accuracy if an invoice will be paid on time or not and can provide estimates of the magnitude of the delay. We illustrate our techniques in the context of real-world transaction data from multiple firms. Finally, simulation results show that our approach can reduce collection time up to a factor of four compared to a baseline that is not model-driven.

Predictive modeling for collections of accounts receivable

It is commonly agreed that accounts receivable (AR) can be a source of financial difficulty for firms when they are not efficiently managed and are underperforming. Experience across multiple industries shows that effective management of AR and overall financial performance of firms are positively correlated. In this paper we address the problem of reducing outstanding receivables through improvements in the collections strategy. Specifically, we demonstrate how supervised learning can be used to build models for predicting the payment outcomes of newly-created invoices, thus enabling customized collection actions tailored for each invoice or customer. Our models can predict with high accuracy if an invoice will be paid on time or not and can provide estimates of the magnitude of the delay. We illustrate our techniques in the context of transaction data from multiple firms.

A Model-Driven Development Approach to Integrating Requirements, Design and Simulations in the Early Stages of Product Development

Decisions made in the early stages of the product development lifecycle have significant impact on the downstream activities. However, existing tools supporting decision-making and product verification at these stages are very limited. One of the obvious reasons is the lack of a common understanding between the system-level design activity and the design activities within the various participating engineering disciplines. In this paper, we propose a collaboration solution which we have developed and commercialized based on the model-driven development platform that allows numerous engineers from heterogeneous engineering disciplines to collaborate on the development of a complex system, such as an automobile. It helps engineers apprehend the system holistically and collectively thus make better architectural decisions. More specifically, this solution connects discipline-specific designs and simulations with the system-level requirements that trigger them in order to facilitate the integration of development efforts and to enable system-level evaluation of the design concepts early in the product development processes. Our approach provides an effective way to trace and analyze the impact of requirements and design changes, facilitates reuse of simulation artifacts for the optimization of future product designs, and supports decision-making activities at the system level. We illustrate our approach in the context of a automotive use case involving mechanical, requirement and safety engineers respectively using their own authoring environments but collectively in synch on the total system thanks to an SOA (Service-Oriented Architecture) based integration between their authoring environments.© 2007 ASME

Detail-preserving variational surface design with multiresolution constraints

We present a variational framework for rapid shape prototyping. The modeled shape is represented as a Catmull-Clark multiresolution subdivision surface which is interactively deformed by direct user input. Free-form design goals are formulated as constraints on the shape and the modeling problem is cast into a constrained optimization one. The focus of this paper is on handling multiresolution constraints of different kinds and on preserving surface details throughout the deformation process. Our approach eliminates the need for an explicit decomposition of the input model into frequency bands and the overhead associated with saving and restoring high-frequency detail after global shape fairing. Instead, we define a deformation vector field over the model and we optimize its energy. Surface details are considered as part of the rest shape and are preserved during free-form model editing. We explore approximating the solution of the optimization problem to various degrees to balance trade-offs between interactivity and accuracy of the results.

Parameterization for Remeshing Over Dynamically Changing Domains

We present a method for incrementally maintaining a global, bijective mapping between an arbitrary triangular input mesh and a dynamically changing domain mesh by use of a meta-mesh. The domain mesh is derived from the input mesh through an arbitrary sequence of local mesh connectivity operations. The mapping can be maintained independent of the choice of metrics for driving the simplification process to derive the domain mesh. In addition, due to the dynamic nature of the mapping, user post-processing can be applied to adjust the final domains

System Data Management: An Inter-Disciplinary Collaboration Architecture for Systems Engineering

This paper presents a novel approach to integrating systems engineering (SE) artifacts and methods with discipline-specific detailed design artifacts and processes, for the purpose of facilitating inter-disciplinary collaboration. In particular, it addresses the lifecycle management of complex products involving mechanical, electrical, electronics and software aspects, and being designed following a formal product development methodology. The primary motivation of the approach is to capture and maintain the traceability between the concrete artifacts stored in discipline-specific “repositories” and the abstract artifacts used to support system-level decisions as well as system integration. The proposed approach acknowledges the fundamental differences that exist between the various engineering disciplines and therefore favors a loose coupling based on a process-centric management of the artifacts traceability links. These considerations lead to an inter-disciplinary collaboration and infrastructure pattern called “system data management” (SDM), with the role of enforcing the integrity of the inter-disciplinary traceability between artifacts. As a byproduct, this approach suggests a novel perspective on product data management (PDM) and software configuration management (SCM) integration that sharply contrasts with point-to-point integration solutions. The authors have implemented a prototype based on a service-oriented architecture (SOA) and existing PDM and SCM technologies.© 2005 ASME