Four Eyes Lab Open House

Thursday, October 26, 2006, 10am-1pm

Location: Four Eyes Lab, UCSB Campus, Trailer 935

(Directly east of Phelps Hall, the middle of three trailers)

Information and Directions at http://ilab.cs.ucsb.edu

At the "Four Eyes" Lab, directed by Matthew Turk and Tobias Höllerer, we pursue research in the four I's of Imaging, Interaction, and Innovative Interfaces. During the open house, we will be describing and demonstrating several ongoing research projects. Feel free to drop by any time from 10:00am to 1:00pm and have a look at any projects that might interest you, talk to the lab's faculty, students, and visitors, and partake of some refreshments.

List of Presented Projects and Presenters:

Mobile Augmented Reality - Towards "Anywhere Augmentation"

Jason Wither, Stephen DiVerdi

The goal of this project is to allow users working in augmented reality to interact with and augment a real world outdoor scene that has not already been modeled. We strive to accomplish this using affordable commercial GPS and orientation trackers, as well as computer vision, and simple and intuitive user input. We also make use of aerial photographs to accurately annotate 3D features from a single position. Aerial photographs provide a rich set of features that can be automatically extracted to create best guesses of intended annotations with minimal user input.

GroundCam - An Anywhere Augmentation Tracking Modality

Stephen DiVerdi

Anywhere Augmentation pursues the goal of lowering the initial investment of time and money necessary to participate in mixed reality work. In this spirit, the GroundCam is a new outdoor, 2D person-tracking modality that operates on the same principles as an optical mouse. Our results show that the GroundCam has favorable performance characteristics, making it a good choice for Anywhere Augmentation work.

Classification of Abnormal Activities in Video

Justin Muncaster

We present a method to classify activities based on the trajectory of feature measurements. We propose a hierarchically organized graphical model. We also propose to use deterministic annealing to partition the feature space and automatically discover low-level states in our model. We apply our algorithm to real world data to demonstrate its effectiveness.

Interactive Stereoscopic FogScreen: Towards a 3D Walkthrough Display

Cha Lee, Stephen DiVerdi, Thomas Klemmer

We are working towards a 3D walk-through display based on the FogScreen, an "immaterial" indoor 2D projection screen. We extend the basic 2D FogScreen setup with head tracking to provide correct perspective rendering for a single user. Second, we add support for multiple types of stereoscopic imagery. Third, we present the front and back views of a graphics scene on the two sides of the FogScreen, so that one can cross the screen to see the content from the back.

Simulation of a 3D Display based on Transparent Screens

Cha Lee

We present a simulation of a volumetric 3D scene based on immaterial displays. We base our simulation on using multiple transparent displays with head tracking to generate view-dependent blended images which are projected onto the front and back of the screens. This creates the illusion of depth and gives a stereoscopic feel to the image. Using the simulator, we are exploring various options of constructing a pseudo-volumetric display from multiple FogScreens.

InViTe - Distance Collaboration using an Augmented Video Approach

Chris Coffin

In collaboration with the Korea Institute for Science and Technology (KIST), we are working towards tangible interfaces for telecommunication. We present InViTe, the Interactive Video Teleconferencing system, which tackles integration of real and virtual imagery in a video-based augmented reality approach. Our goal is to provide a highly interactive tangible experience with augmented videos.

Multi-Flash Stereopsis

Longbin Chen, Rogerio Feris

We describe a novel method for acquiring high quality depth maps from real-world scenes. Our approach is based on a compact stereo camera with small baseline multi-flash illumination, which allows discontinuity preserving correspondence maps from scenes that are extremely challenging for conventional stereo methods.

Scalable Graph Visualization

Brynjar Gretarsson

The goal of this project is to make it easier for data analysts to visualize and interact with large graphs of data. We use a spring model to lay out the graph and allow the user to see how the graph gradually reaches it's resting position. There are three main parts to this project. The first one is to be able to visualize graphs of up to hundreds of thousands of nodes. The second is to have constrained interaction with the graph layout to make it easier for the user to modify the layout. The third part is to strategically place node labels on the screen with no overlapping.

Simulation of a 3D Display based on Transparent Screens

Cha Lee

Multi-view Appearance-based 3D Hand Pose Estimation

Haiying Guan

We describe a novel approach to appearance-based hand pose estimation which relies on multiple cameras to improve accuracy and resolve ambiguities caused by self-occlusions. Rather than estimating 3D geometry, our approach uses multiple views to extend current exemplar-based methods for estimating hand pose by matching a probe image with a large discrete set of labeled hand pose images. We formulate the problem in a MAP framework, where the information from multiple cameras is fused to provide reliable hand pose estimation. Our quantitative experimental results show that the correct estimation rate is much higher using our multi-view approach than using a single-view approach.

A Sketch Interface to Aid 3D Scene Reconstruction

Jonathan Ventura, Taehee Lee

We propose a semi-automatic scene reconstruction method for a single photograph which uses a sketch interface as user input. This sketch system uses image features in several ways in order to lessen the the cognitive load on the user. Because strokes on the image form the only input to the system, it is easily adaptable to a tablet display with a pen device for input. In related work, we are striving towards acquiring rough 3D geometry from live video feeds in real time.

Viewpoint Stabilization for Live Collaborative Video Augmentations

Taehee Lee

We present a method for stabilizing live video from a moving camera for the purpose of a tele-meeting, in which a participant with an AR view onto a shared canvas collaborates with a remote user. The AR view is established without markers and using no other tracking equipment than a head-worn camera. The remote user is allowed to directly annotate the local user's view in real time on a desktop or tablet PC.