Technical Papers

Operating on Images

Thursday, 29 November 09:00 - 10:45 |  Peridot 201

Quality Prediction for Image Completion - Picture

Quality Prediction for Image Completion

We present a data-driven method to predict the quality of an image completion method. We use our algorithm to first crop and then complete stitched panoramas. Our predictive ability is used to find an optimal crop shape before the completion is computed, potentially saving significant amounts of computation.


Johannes Kopf, Microsoft Research Redmond
Wolf Kienzle, Microsoft Research
Steven Drucker, Microsoft Research
Sing Bing Kang, Microsoft Research


Manifold Preserving Edit Propagation - Picture

Manifold Preserving Edit Propagation

We propose a novel edit propagation algorithm for interactive image
and video manipulations. We use the locally linear embedding
(LLE) to represent each pixel as a linear combination of its
neighbors in a feature space.We demonstrate our manifold preserving edit propagation with various applications.


Xiaowu Chen, Beihang University
Dongqing Zou, Beihang University
Qinping Zhao, Beihang University
Ping Tan, National University of Singapore


Sparse PDF Maps for Non-Linear Multi-Resolution Image Operations - Picture

Sparse PDF Maps for Non-Linear Multi-Resolution Image Operations

sPDF-maps are a new type of image pyramid that enables the accurate evaluation of non-linear image operators on gigapixel images directly at any output resolution. They are computed in a pre-processing step, but enable run time evaluation of a variety of interactive image operations on GPUs.


Markus Hadwiger, King Abdullah University of Science and Technology
Ronell Sicat, King Abdullah University of Science and Technology
Johanna Beyer, King Abdullah University of Science and Technology
Jens Krueger, IVDA, DFKI, Intel VCI
Torsten Moeller, Simon Fraser University


DressUp! Outfit Synthesis Through Automatic Optimization - Picture

DressUp! Outfit Synthesis Through Automatic Optimization

An automatic optimization approach to outfit synthesis is introduced. The proposed system can suggest a set of outfits subject to a particular dress code, given the body color attributes and a wardrobe of clothing items. It will be very useful for personal outfit suggestion and virtual character modeling.


Lap-Fai Yu, University of California, Los Angeles
Sai-Kit Yeung, The Singapore University of Technology and Design
Demetri Terzopoulos, University of California, Los Angeles
Tony F. Chan, The Hong Kong University of Science and Technology