Google Earth on the Web lets you navigate and explore geographic data
on a 3D globe using a Web browser, without the need of a plug-in. It
is one of the, if not the largest WebGL application out there.
Read more about it, at the Siggraph 2014 talk by Kontkanen and Parker:
Earth in Google maps: rendering trillions of triangles in JavaScript
Google Maps Engine is a scalable, powerful and secure platform for geographic data and imagery using the power of the Google cloud. It enabled customers to upload massive geographical data, including vectors and imagery, and to customize the maps to their own needs. As a software engineer, I developed image-processing algorithms for massive geo-spatial imagery.
Explored neighborhood graphs to improve
the robustness and quality of data analysis tools.
Towards Robust Topology of Sparsely Sampled Data.
Locally-scaled Spectral Clustering using Empty Region Graphs.
Wrote an open source library to compute neighborhood graphs, re-written and now hosted at https://github.com/dhrieman/ngl
Topological Spines are new visual representations of high-dimensional data.
They encode both the topology and geometric structure of a scalar field.
We worked together with climate scientists at LLNL to quantify uncertainty
in climate models, and with physicists at the National Ignition Facility to
help them understand the results of their simulations.
Read more about it in our 2011 IEEE Visualization Paper:
Topological Spines:
A Structure-Preserving Visual Representation of Scalar Fields
Developed computer vision algorithms to support a program to monitor
tens of square kilometers of terrain from the skies.
Implemented robust image stitching algorithms on GPUs (BAVM 2012), and worked with researchers from UC Davis to automatically fuse aerial
optical imagery with LiDAR data, in our 2014 ICCP
paper:
Automatic Registration of LiDAR and Optical Imagery using Depth Map Stereo
Non-parametric Models for Large Scale Aerial Image Registration
Developed new visualizations for climate modeling, including
the statistical lighting model, with data from the
Coupled Model Intercomparison Project (CMIP5).
I also created the Mutual Information Diagram, a 2D diagram
that encodes information theoretic measures to visualize the statistical
relationships between random variables.
Created a new medium for sharing videos as compact narratives using a series of computer
vision techniques, including dense correspondence estimation, foreground-background
classification and min-cut image segmentation, to automatically split a
movie into seamless mosaics that can be played back interactively.
Dynamic Video Narratives, Siggraph 2010.
Project page
Explorable Images are a series of rendering techniques for remote and in-situ
visualization. These images encode intermediate representations that can be used to
re-color, re-light and transform the result in real-time without the need to
access the original data.
Visualization by Proxy, A Novel Framework for Deferred Interaction with Volume Data
An Exploratory Technique for Coherent Visualization of Time-varying Volume Data
Explorable Images for Visualizing Volume Data
Over the years I have investigated methods to speed up and improve
the quality of volume rendered images, including methods to fuse
multimodal imagery, gradient estimation methods and re-formation.
Visibility guided multimodal volume visualization
Gradient Estimation Methods on Unstructured Meshes
Curve-Centric Volume Reformation for Comparative Visualization
Developed visualizations and algorithms for analysis and understanding
of social networks. I coined the term centrality sensitivity and developed
algorithms to measure it, as described in:
Visual Reasoning about Social Networks using
Centrality Sensitivity
I developed netzen, an interactive tool for social network analysis and
visualization.
Co-PI of an NSF FODAVA project for the study of uncertainty in visual analytics.
Developed sensitivity plots for interactive local regression and the
understanding of variable uncertainty.
In our article
The Generalized Sensitivity Plot,
we summarize our study on showing sensitivity on scatter plots, a
relatively simple but very effective mechanism for visualizing
high-dimensional data sets.
Created a series of feature detection algorithms for improving the
visualization of 3D features in volumetric data sets.
These include the Size-based Transfer Function, using a
novel 3D non-linear diffusion algorithm, the Occlusion Spectrum,
which is capable of extracting features based on their spatial surroundings,
and the
Visibility Histogram, which computes in real time the
distribution of visibility.
Coined the term illustrative deformation to describe a series of
rendering techniques to simulate deformation in a non-realistic
manner, akin to illustrative rendering.
Wrote my Ph.D. dissertation
on that.
Wrote software to animate volumetric data, like the visible human
data set. Optimized an animation pipeline to run from hours to minutes
using GPUs.
Carlos D. Correa is a computer scientist and software engineer, interested in data analytics, visualization, computer vision and computer graphics.
He currently works at Google, rendering the Earth on the Web. Before joining Google, he was a computer scientist at LLNL and was a postdoc at the University of California, Davis.
He has authored over 50 articles in top tier conferences and journals, including Siggraph, KDD and IEEE Transactions on Visualization and Computer Graphics. He has also served as part of the international program committee of IEEE Vis and Infovis, Pacific Vis and Eurovis, and peer-reviews articles for IEEE TVCG, Siggraph, Eurographics, Computer Graphics Forum, among others.
| cdcorrea @ gmail | |
| google+ | https://plus.google.com/u/0/+CarlosCorreaO/ |
| http://www.facebook.com/cdcorrea1 | |
| http://www.linkedin.com/in/cdcorrea | |
| https://twitter.com/cdcorrea | |
| youtube | https://www.youtube.com/user/cdcorrea1 |
| Google scholar | https://scholar.google.com/citations?user=-Km_NFkAAAAJ&hl=en |
| Publications | 58 | Citations | 1478 |
|---|---|---|---|
| h-index | 22 | i10-index | 36 |
