Guest Editorial: Computational Vision at Brown

International Journal of Computer Vision -     

视觉认知计算模型综述

视觉认知计算模型作为联系视觉认知和信息计算的有效手段,其研究涉及到认知科学、信息科学等多个交叉学科,具有复杂性和多样性等特点。为能更好地把握其发展规律,文中从视觉计算角度系统总结视觉认知计算模型,以其两个主要来源为主线分别从生物视觉机制和视觉计算理论回顾视觉认知计算模型的发展。根据其研究的特点,对视觉认知计算模型的发展做出一定评述,并指出视觉认知计算模型的发展必将对计算视觉理论和生物视觉机制产生深远影响。     

Vision ergonomics at recycling centres

All municipalities in Sweden offer their inhabitants a service for disposing of large-size and hazardous waste at local recycling centres. Opening hours at these centres include hours of darkness. The aims of this study were to 1) describe user and employee experiences of lighting and signs at Swedish recycling centres, 2) measure and assess the lighting system at the two recently built recycling centres in Linköping and to assess the legibility and visibility of the signs used and 3) propose recommendations regarding lighting and signs for recycling centres. Interviews and questionnaires were used to assess experiences of employees and users, and light measurements were performed. By observing users, activities with different visual demands at different areas within the recycling centres were identified. Based on the literature, standards and stakeholder experiences, recommendations regarding lighting systems and sign design, illuminance, luminance and uniformity are proposed for recycling centres.     

Introduction: Learning and Vision at CBCL

This special issue is dedicated to some of the recent work at the Center for Biological and Computational Learning at MIT in applying machine learning techniques to computer vision.     

Introduction: Computer Vision Research at NECI

This special issue of the International Journal of Computer Vision highlights research on computer vision at the NEC Research Institute. The purpose of this preface is to provide some context to how this work reflects our broader goals and themes of computer vision research at NECI.     

耶鲁枪击问题之分析

本文以耶鲁枪击问题为例,分析了典型的非单调理论在解决框架问题上的局限性;在综述了已有的解决方案所存在的问题后,指出了非单调的推理与逻辑程序设计系统GKD—NMRS在描述和解决框架问题上的简洁性和直观性。     

基于视觉的路口车辆排队长度检测方法

现有基于视觉的交叉路口车辆排队长度测量方法,不能给出实际排队长度,且摄像机标定复杂。为此,提出基于连通区域的车辆排队长度检测算法及以停车线为参照物的摄像机快速标定方法。在停车线后设定停车检测区域,并利用相邻帧差法判断是否存在车辆停车。在排队检测区域内用最大类间方差法自动设定阈值并分割图像,以检测车辆存在。利用连通区域各种参数确定车辆所在的连通域并检测出队尾。运用针孔成像模型论证以停车线为参照物的摄像机标定方法。实验结果证明,该方法可以准确检测出队首和队尾像素,计算出实际长度,可满足实时性要求。     

An overview of the Yale Gem system

The Gem system is an experimental computing facility that provides low cost, high speed, graphics-oriented computing to between ten and sixteen simultaneous users. It provides many unusual facilities to its users and presents a user interface that is unique in its convenience and flexibility. The motivation for the system, its design and user experience are described. Possible future avenues of research are also outlined.     

Vision at work visual defect and the visual demand of tasks

Several related studies of visual tasks in a large communications undertaking have revealed much defect of vision and many sources of visual fatigue and inefficiency. The nature of visual fatigue is described briefly and a programme to facilitate visual tasks is outlined.     

Guest Editorial: Statistical and Computational Theories of Vision: Modeling,Learning, Sampling and Computing,Part I

International Journal of Computer Vision -     

Computational aspects of the coarsening at random model and the Shapley value

This paper is concerned with computational aspects of coarse data in information technology. A method for calculating the CAR (coarsening at random) model of a finite random set is obtained, and a formula for computing the Shapley value using the distribution of the random set is given. The relationship between the CAR model solution and the Shapley value is also discussed. Several examples are given to illustrate our results.     

Early development of implicit methods for Computational Fluid Dynamics at NASA Ames

The development of implicit finite-difference methods at the Computational Fluid Dynamics Branch (NASA Ames Research Center) in the 1970 to early 1980s timeframe is presented. The seminal work of Drs. Robert F. Warming and Richard M. Beam is highlighted. Their contribution (along with that of others in the Ames CFD Branch) paved the way for modern large scale application codes. The focus will be on a few of the significant developments (e.g., approximate factorization) and results from those early pioneering days.     

Computational ludics

We reformulate the theory of ludics introduced by J.-Y. Girard from a computational point of view. We introduce a handy term syntax for designs, the main objects of ludics. Our syntax also incorporates explicit cuts for attaining computational expressivity. In addition, we consider design generators that allow for finite representation of some infinite designs. A normalization procedure in the style of Krivine’s abstract machine directly works on generators, giving rise to an effective means of computation over infinite designs.The acceptance relation between machines and words, a basic concept in computability theory, is well expressed in ludics by the orthogonality relation between designs. Fundamental properties of ludics are then discussed in this concrete context. We prove three characterization results that clarify the computational powers of three classes of designs. (i) Arbitrary designs may capture arbitrary sets of finite data. (ii) When restricted to finitely generated ones, designs exactly capture the recursively enumerable languages. (iii) When further restricted to cut-free ones as in Girard’s original definition, designs exactly capture the regular languages.We finally describe a way of defining data sets by means of logical connectives, where the internal completeness theorem plays an essential role.