Design reuse oriented partial retrieval of CAD models

As a huge number of 3D CAD models is generated each year, retrieval of 3D CAD models is becoming more and more important for achieving design reuse. However, the existing methods for partial retrieval of 3D CAD models are very few and far from the requirements of design reuse. In this paper, we present an approach to partial retrieval of 3D CAD models for design reuse. The criteria for determining whether a subpart of 3D CAD models is reusable for design is defined first. Based on the criteria defined, all the design reusable subparts involved in the 3D CAD models in the library are automatically extracted and stored in the library as reference models. Moreover, each design reusable subpart in the library is represented by all its local matching regions in a hierarchical way so as to support multi-mode partial retrieval. In our approach, three partial retrieval modes including normal retrieval, exact retrieval and relaxed retrieval are defined to meet various partial retrieval requirements of design reuse such as the incomplete and vague queries during the early design stage. And the multi-mode partial retrieval is achieved by performing multi-mode matching and similarity assessment between the query and the design reusable subparts in the library indexed by bitmap. Experimental results are presented to demonstrate the effectiveness of the approach.     

A dynamic priority-based approach to concurrent toolpath planning for multi-material layered manufacturing

This paper presents an approach to concurrent toolpath planning for multi-material layered manufacturing (MMLM) to improve the fabrication efficiency of relatively complex prototypes. The approach is based on decoupled motion planning for multiple moving objects, in which the toolpaths of a set of tools are independently planned and then coordinated to deposit materials concurrently. Relative tool positions are monitored and potential tool collisions detected at a predefined rate. When a potential collision between a pair of tools is detected, a dynamic priority scheme is applied to assign motion priorities of tools. The traverse speeds of tools along the x-axis are compared, and a higher priority is assigned to the tool at a higher traverse speed. A tool with a higher priority continues to deposit material along its original path, while the one with a lower priority gives way by pausing at a suitable point until the potential collision is eliminated. Moreover, the deposition speeds of tools can be adjusted to suit different material properties and fabrication requirements. The proposed approach has been incorporated in a multi-material virtual prototyping (MMVP) system. Digital fabrication of prototypes shows that it can substantially shorten the fabrication time of relatively complex multi-material objects. The approach can be adapted for process control of MMLM when appropriate hardware becomes available. It is expected to benefit various applications, such as advanced product manufacturing and biomedical fabrication.     

Multi-level hierarchical interface-based supervisory control

Hierarchical Interface-Based Supervisory Control employs interfaces that allow properties of a monolithic system to be verified through local analysis. By avoiding the need to verify properties globally, significant computational savings can be achieved. In this paper we provide local requirements for a multi-level architecture employing command-pair type interfaces. This multi-level architecture allows for a greater reduction in complexity and improved reconfigurability over the two-level case that has been previously studied since it allows the global system to be partitioned into smaller modules. This paper also provides results for synthesizing supervisors in the multi-level architecture that are locally maximally permissive with respect to a given specification and set of interfaces.     

A secure collaboration service for dynamic virtual organizations

Nowadays, various promising paradigms of distributed computing over the Internet, such as Grids, P2P and Clouds, have emerged for resource sharing and collaboration. To enable resources sharing and collaboration across different domains in an open computing environment, virtual organizations (VOs) often need to be established dynamically. However, the dynamic and autonomous characteristics of participating domains pose great challenges to the security of virtual organizations. In this paper, we propose a secure collaboration service, called PEACE-VO, for dynamic virtual organizations management. The federation approach based on role mapping has extensively been used to build virtual organizations over multiple domains. However, there is a serious issue of potential policy conflicts with this approach, which brings a security threat to the participating domains. To address this issue, we first depict concepts of implicit conflicts and explicit conflicts that may exist in virtual organization collaboration policies. Then, we propose a fully distributed algorithm to detect potential policy conflicts. With this algorithm participating domains do not have to disclose their full local privacy policies, and is able to withhold malicious internal attacks. Finally, we present the system architecture of PEACE-VO and design two protocols for VO management and authorization. PEACE-VO services and protocols have successfully been implemented in the CROWN test bed. Comprehensive experimental study demonstrates that our approach is scalable and efficient.     

Semantic modeling of natural scenes based on contextual Bayesian networks

This paper presents a novel approach based on contextual Bayesian networks (CBN) for natural scene modeling and classification. The structure of the CBN is derived based on domain knowledge, and parameters are learned from training images. For test images, the hybrid streams of semantic features of image content and spatial information are piped into the CBN-based inference engine, which is capable of incorporating domain knowledge as well as dealing with a number of input evidences, producing the category labels of the entire image. We demonstrate the promise of this approach for natural scene classification, comparing it with several state-of-art approaches.     

LPP solution schemes for use with face recognition

Locality preserving projection (LPP) is a manifold learning method widely used in pattern recognition and computer vision. The face recognition application of LPP is known to suffer from a number of problems including the small sample size (SSS) problem, the fact that it might produce statistically identical transform results for neighboring samples, and that its classification performance seems to be heavily influenced by its parameters. In this paper, we propose three novel solution schemes for LPP. Experimental results also show that the proposed LPP solution scheme is able to classify much more accurately than conventional LPP and to obtain a classification performance that is only little influenced by the definition of neighbor samples.     

Inter-image outliers and their application to image classification

Image variability that is impossible or difficult to restore by intra-image processing, such as the variability caused by occlusions, significantly reduces the performance of image-recognition methods. To address this issue, we propose that the pixels associated with large distances obtained by inter-image pixel-by-pixels comparisons should be considered as inter-image outliers and should be removed from the similarity calculation used for the image classification. When this method is combined with the template-matching method for image recognition, it leads to state-of-the-art recognition performance: 91% with AR database that includes occluded face images, 90% with PUT database that includes pose variations of face images and 100% with EYale B database that includes images with large illumination variation.     

Clustering of temporal gene expression data by regularized spline regression and an energy based similarity measure

Clustering analysis of temporal gene expression data is widely used to study dynamic biological systems, such as identifying sets of genes that are regulated by the same mechanism. However, most temporal gene expression data often contain noise, missing data points, and non-uniformly sampled time points, which imposes challenges for traditional clustering methods of extracting meaningful information. In this paper, we introduce an improved clustering approach based on the regularized spline regression and an energy based similarity measure. The proposed approach models each gene expression profile as a B-spline expansion, for which the spline coefficients are estimated by regularized least squares scheme on the observed data. To compensate the inadequate information from noisy and short gene expression data, we use its correlated genes as the test set to choose the optimal number of basis and the regularization parameter. We show that this treatment can help to avoid over-fitting. After fitting the continuous representations of gene expression profiles, we use an energy based similarity measure for clustering. The energy based measure can include the temporal information and relative changes of the time series using the first and second derivatives of the time series. We demonstrate that our method is robust to noise and can produce meaningful clustering results.     

A review of vision-based systems for soccer video analysis

This paper presents a survey of soccer video analysis systems for different applications: video summarization, provision of augmented information, high-level analysis. Computer vision techniques have been adapted to be applicable in the challenging soccer context. Different semantic levels of interpretation are required according to the complexity of the corresponding applications. For each application area we analyze the computer vision methodologies, their strengths and weaknesses and we investigate whether these approaches can be applied to extensive and real time soccer video analysis.     

Robust metric calibration of non-linear camera lens distortion

Camera lens distortion is crucial to obtain the best performance cameral model. Up to now, different techniques exist, which try to minimize the calibration error using different lens distortion models or computing them in different ways. Some compute lens distortion camera parameters in the camera calibration process together with the intrinsic and extrinsic ones. Others isolate the lens distortion calibration without using any template and basing the calibration on the deformation in the image of some features of the objects in the scene, like straight lines or circles. These lens distortion techniques which do not use any calibration template can be unstable if a complete camera lens distortion model is computed. They are named non-metric calibration or self-calibration methods.Traditionally a camera has been always best calibrated if metric calibration is done instead of self-calibration. This paper proposes a metric calibration technique which computes the camera lens distortion isolated from the camera calibration process under stable conditions, independently of the computed lens distortion model or the number of parameters. To make it easier to resolve, this metric technique uses the same calibration template that will be used afterwards for the calibration process. Therefore, the best performance of the camera lens distortion calibration process is achieved, which is transferred directly to the camera calibration process.