Poster Session 4: Face/Gesture/Action Detection and Recognition.- Palmprint Recognition Under Unconstrained Scenes.- Comparative Studies on Multispectral Palm Image Fusion for Biometrics.- Learning Gabor Magnitude Features for Palmprint Recognition.- Sign Recognition Using Constrained Optimization.- Poster Session 4: Image and Video Processing.- Depth from Stationary Blur with Adaptive Filtering.- Three-Stage Motion Deblurring from a Video.- Near-Optimal Mosaic Selection for Rotating and Zooming Video Cameras.- Video Mosaicing Based on Structure from Motion for Distortion-Free Document Digitization.- Super Resolution of Images of 3D Scenecs.- Learning-Based Super-Resolution System Using Single Facial Image and Multi-resolution Wavelet Synthesis.- Poster Session 4: Segmentation and Classification.- Statistical Framework for Shot Segmentation and Classification in Sports Video.- Sports Classification Using Cross-Ratio Histograms.- A Bayesian Network for Foreground Segmentation in Region Level.- Efficient Graph Cuts for Multiclass Interactive Image Segmentation.- Feature Subset Selection for Multi-class SVM Based Image Classification.- Evaluating Multi-class Multiple-Instance Learning for Image Categorization.- Poster Session 4: Shape.- TransforMesh : A Topology-Adaptive Mesh-Based Approach to Surface Evolution.- Microscopic Surface Shape Estimation of a Transparent Plate Using a Complex Image.- Shape Recovery from Turntable Image Sequence.- Shape from Contour for the Digitization of Curved Documents.- Improved Space Carving Method for Merging and Interpolating Multiple Range Images Using Information of Light Sources of Active Stereo.- Shape Representation and Classification Using Boundary Radius Function.- A Convex Programming Approach to the Trace Quotient Problem.- Learning a Fast Emulator of a Binary Decision Process.- Multiplexed Illumination for Measuring BRDF Using an Ellipsoidal Mirror and a Projector.- Analyzing the Influences of Camera Warm-Up Effects on Image Acquisition.- Simultaneous Plane Extraction and 2D Homography Estimation Using Local Feature Transformations.- A Fast Optimal Algorithm for L 2 Triangulation.- Adaptively Determining Degrees of Implicit Polynomial Curves and Surfaces.- Determining Relative Geometry of Cameras from Normal Flows.- Poster Session 5: Geometry.- Highest Accuracy Fundamental Matrix Computation.- Sequential L ??? Norm Minimization for Triangulation.- Initial Pose Estimation for 3D Model Tracking Using Learned Objective Functions.- Multiple View Geometry for Non-rigid Motions Viewed from Translational Cameras.- Visual Odometry for Non-overlapping Views Using Second-Order Cone Programming.- Pose Estimation from Circle or Parallel Lines in a Single Image.- An Occupancy-Depth Generative Model of Multi-view Images.- Poster Session 5: Matching and Registration.- Image Correspondence from Motion Subspace Constraint and Epipolar Constraint.- Efficient Registration of Aerial Image Sequences Without Camera Priors.- Simultaneous Appearance Modeling and Segmentation for Matching People Under Occlusion.- Content-Based Matching of Videos Using Local Spatio-temporal Fingerprints.- Automatic Range Image Registration Using Mixed Integer Linear Programming.- Accelerating Pattern Matching or How Much Can You Slide?.- Poster Session 5: Recognition.- Detecting, Tracking and Recognizing License Plates.- Action Recognition for Surveillance Applications Using Optic Flow and SVM.- The Kernel Orthogonal Mutual Subspace Method and Its Application to 3D Object Recognition.- Viewpoint Insensitive Action Recognition Using Envelop Shape.- Unsupervised Identification of Multiple Objects of Interest from Multiple Images: dISCOVER.- Poster Session 5: Stereo, Range and 3D.- Fast 3-D Interpretation from Monocular Image Sequences on Large Motion Fields.- Color-Stripe Structured Light Robust to Surface Color and Discontinuity.- Stereo Vision Enabling Precise Border Localization Within a Scanline Optimization Framework.- Three Dimensional Position Measurement for Maxillofacial Surgery by Stereo X-Ray Images.- Total Absolute Gaussian Curvature for Stereo Prior.- Fast Optimal Three View Triangulation.- Stereo Matching Using Population-Based MCMC.- Dense 3D Reconstruction of Specular and Transparent Objects Using Stereo Cameras and Phase-Shift Method.- Identifying Foreground from Multiple Images.- Image and Video Matting with Membership Propagation.- Temporal Priors for Novel Video Synthesis.- Content-Based Image Retrieval by Indexing Random Subwindows with Randomized Trees.- Poster Session 6: Face/Gesture/Action Detection and Recognition.- Analyzing Facial Expression by Fusing Manifolds.- A Novel Multi-stage Classifier for Face Recognition.- Discriminant Clustering Embedding for Face Recognition with Image Sets.- Privacy Preserving: Hiding a Face in a Face.- Face Mosaicing for Pose Robust Video-Based Recognition.- Face Recognition by Using Elongated Local Binary Patterns with Average Maximum Distance Gradient Magnitude.- An Adaptive Nonparametric Discriminant Analysis Method and Its Application to Face Recognition.- Discriminating 3D Faces by Statistics of Depth Differences.- Kernel Discriminant Analysis Based on Canonical Differences for Face Recognition in Image Sets.- Person-Similarity Weighted Feature for Expression Recognition.- Converting Thermal Infrared Face Images into Normal Gray-Level Images.- Recognition of Digital Images of the Human Face at Ultra Low Resolution Via Illumination Spaces.- Poster Session 6: Math for Vision.- Crystal Vision-Applications of Point Groups in Computer Vision.- On the Critical Point of Gradient Vector Flow Snake.- A Fast and Noise-Tolerant Method for Positioning Centers of Spiraling and Circulating Vector Fields.- Interpolation Between Eigenspaces Using Rotation in Multiple Dimensions.- Conic Fitting Using the Geometric Distance.- Poster Session 6: Segmentation and Classification.- Efficiently Solving the Fractional Trust Region Problem.- Image Segmentation Using Iterated Graph Cuts Based on Multi-scale Smoothing.- Backward Segmentation and Region Fitting for Geometrical Visibility Range Estimation.- Image Segmentation Using Co-EM Strategy.- Co-segmentation of Image Pairs with Quadratic Global Constraint in MRFs.- Shape Reconstruction from Cast Shadows Using Coplanarities and Metric Constraints.- Evolving Measurement Regions for Depth from Defocus.- A New Framework for Grayscale and Colour Non-lambertian Shape-from-Shading.- A Regularized Approach to Feature Selection for Face Detection.- Iris Tracking and Regeneration for Improving Nonverbal Interface.- Face Mis-alignment Analysis by Multiple-Instance Subspace.

Computer vision - ACCV 2007 : 8th Asian Conference on Computer Vision, Tokyo, Japan, November 18-22, 2007, proceedings

The efficient development of complex algorithms for mobile robot applications is strongly encouraged by appropriate testing and validation opportunities. In this paper, we present SIMORE (Simulation of Multirobot Environments), a simulation framework for simulating sensors, actuators and entire robots, in virtual 3D environments. It provides synthesized sensor data but also additional information emanating from the intrinsic properties of given virtual environments. SIMORE is built upon open-source libraries such as the OpenSceneGraph toolkit for scene rendering and the Open Dynamics Engine to guarantee a solid physical behavior. It provides easy to use modeling tools and plug-ins for creating and modifying the appearance as well as the dynamic properties of objects to be included in the scene. Our simulation framework possess a module based design capable of supporting further extensions and enhancements such as the definition of individual sensors and actuators. Beyond that, SIMORE offers exemplary interfaces for available robotic middleware applications, such as Player or ROS, allowing for a fast and flexible transition between simulation and real hardware platforms. Although still under active development, SIMORE is used to support several research propositions in robot vision, navigation and recognition as well as being a proven prototyping tool for robotic competitions, such as the Robocup (Rescue Ligue).

Validating vision and robotic algorithms for dynamic real world environments

A quick estimation of depth is required by artificial vision systems for their self survival and navigation through the environment. Following the selection strategy of biological vision, known as visual attention, can help in accelerating extraction of depth for important and relevant portions of given scenes. Recent studies on depth perception in biological vision indicate that disparity is computed using object detection in the brain. The proposed method uses concepts from these studies and determines the shift that objects go through in the stereo frames using data regarding their borders. This enables efficient creation of depth saliency map for artificial visual attention. Results of the proposed model have shown success in selecting those locations from stereo scenes that are salient for human perception in terms of depth.

Pre-attentive detection of depth saliency using stereo vision

Depth is an important feature channel for natural vision organisms that helps in focusing attention on important locations of the viewed scene. Artificial visual attention systems require a fast estimation of depth to construct a saliency map based upon distance from the vision system. Recent studies on depth perception in biological vision indicate that disparity is computed using object detection in the brain. The proposed method exploits these studies and determines the shift that objects go through in the stereo frames using data regarding their borders. This enables efficient creation of depth saliency map for artificial visual attention. Results of the proposed model have shown success in selecting those locations from stereo scenes that are salient for human perception in terms of depth.

Fast Depth Saliency from Stereo for Region-Based Artificial Visual Attention

New interaction techniques are coming more than ever before, but are they meaningful? Is there a purpose behind these novel technologies or is it the technology itself that is emphasized? In this paper we have taken the approach of embodied interaction and evaluated current interaction modalities in order to design a next generation input device focusing on the user instead of the technology behind it. We have established the relevance of embodied interaction and our result brings interesting insights into design considerations when designing novel interaction mediums. We have also discussed the technical considerations around our proposed interaction technique.

/pdf/pushing-towards-embodied-interactions-p5oif772v3.pdf

Pushing Towards Embodied Interactions

In this paper we present a real-time scan-line segment based stereo vision for the estimation of biologically motivated classifier cells in an active vision system. The system is challenged to overcome several problems in autonomous mobile robotic vision such as the detection of incoming moving objects and estimating its 3D motion parameters in a dynamic environment. The proposed algorithm employs a modified optimization module within the scan-line framework to achieve valuable reduction in computation time needed for generating real-time depth map. Moreover, the experimental results showed high robustness against noises and unbalanced light condition in input data.

Real-time scan-line segment based stereo vision for the estimation of biologically motivated classifier cells

In this work, we propose a saliency-based approach for estimating and segmenting 3D motions of multiple moving objects represented by 2D motion vector fields (MVF). In order to overcome typical problems in autonomous mobile robotic vision such as noise, occlusions, and inhibition of the ego-motion defects of a moving camera head, a classification module has been implemented to define the global motion of the mounted camera. The proposed method achieves valuable reduction in computational time by applying a guided control module which limits the segmentation output to a flexible predefined threshold value. The computational enhancement is very important since the output of the motion segmentation approach is implemented in an active vision system.

Fast Saliency-Based Motion Segmentation Algorithm for an Active Vision System

In this paper, we present a fast 3D motion parameter estimation approach integrating the depth information acquired by a stereo camera head mounted on a mobile robot. Afterwards, the resulting 3D motion parameters are used to generate and accurately position motion vectors of the generated depth sequence in the 3D space using the geometrical information of the stereo camera head. The proposed approach has successfully detected and estimated predefined motion patterns such as motion in the Z direction and motion vectors pointing to the robot which is very important to overcome typical problems in autonomous mobile robotic vision such as collision detection and inhibition of the ego-motion defects of a moving camera head. The output of the algorithm is part of a multi-object segmentation approach implemented in an active vision system.

Fast depth-integrated 3d motion estimation and visualization for an active vision system

In this paper, we present a real time biologically motivated 3D motion classifier cells integrating the depth information generated from a stereo input implemented in an active vision system. The proposed approach is accurately able to detect and estimate multiple interfered 3D complex motions under the absence of predefined spatial coherence. Moreover, the system has ability to examine the response of input 3D motion vector fields to a certain 3D motion patterns (3D motion classifier cells) such as motion in the Z direction representing movements towards the system, which is very important to overcome typical problem in autonomous mobile robotic vision such as collision detection and inhibition of the ego-motion defects of a moving camera head. The output of the algorithm is part in a multi-object segmentation approach implemented in an active vision system.

Real time stereo-based biologically inspired 3D motion classifier cells

Segmentation isavery iniportant step inmostqfthe nmachine vision applications. A method forcolor segmentation isproposed asabasic requirement qf visuial attention system ofmobile robots. After experiments with different color spaces, theHSIcolor spaceisselected forsegmentation withnminimtm influence ofshades andshadows. Theproposed method first classifies thecolor ofseedpixel andthen selects a method forregion growing according to nature oftheseedpixel Results obtained arevery encouraging andaccor-ding torequtiremcents qfthe application area. Thispaperaddresses theproblem ofobtaining regions fromcolored images suchthat theregions may haveminimum influence ofshades andshadows in order toproduce goodinput forhigher level processing invisual attention systems ofmobile robots. HSI(HueSaturation-Inten sity) space, sometimes alsocalled HSV(Hue-Saturation-Value) orHSL(Hue-SaturationLuminance), hasagoodrepresentation ofthecolors of humanperception [3]andhasgoodcapability of dealing withhighlights, shades, andshadows [2]. As these features suite therequirements ofourapplication area hence ourapproach performs segmentation inHSI color space andproposes animproved algorithm to obtain better results.

M. Salah E.-N. Shafik

Papers

Real-time scan-line segment based stereo vision for the estimation of biologically motivated classifier cells

Fast Saliency-Based Motion Segmentation Algorithm for an Active Vision System

Fast depth-integrated 3d motion estimation and visualization for an active vision system

Real time stereo-based biologically inspired 3D motion classifier cells

Color Segmentation forVisual Attention ofMobile Robots