-meh- 

Kaushik Mitra

Assistant Professor, Department of Electrical Engineering
IIT Madras
office. CSD 303
email. kmitra -at- ee .dot. iitm .dot. ac .dot. in

Research

Traditionally, imaging optics design and image-based inference (image processing, computer vision) have been done independently. However, by jointly designing the imaging optics and inference algorithms, we can significantly enhance the performance of imaging systems. This is the philosophy behind the fast evolving field of Computational Imaging (CI). My research focuses on the following aspects of CI:

  • Develop theory to explore the performance limits of CI systems

  • Develop novel CI systems

  • Explore the promise of machine learning in CI and computer vision

Theory for computational imaging


Framework for analysis of CI systems
IEEE TPAMI 2014

Data-driven design for CI systems
ICCP 2014

Denoise or deblur: Parameter optimization for imaging systems
SPIE Electronic Imaging, 2014

Analyzing computational imaging systems
SPIE Newsroom, 2013

Novel CI systems and processing algorithms


Focal-sweep for large aperture time-of-flight cameras
ICIP 2016

Spatial Phase-Sweep: Increasing temporal resolution of transient imaging using a light source array
ICIP 2016

Improving resolution of light field using hybrid imagingdesign
ICCP 2014

Compressive epsilon photography
SIGGRAPH 2014

A common framework for light field processing
CVPR Workshop 2012

Towards compressive camera networks
IEEE Computer 2014

Machine learning for CI and Computer Vision


Compressive image recovery using recurrent generative model
Under review in ICIP 2017

Missing data matrix factorization
NIPS 2010

Analysis of sparsity based robust regressionalgorihms
IEEE TSP 2013

Robust regression using sparse learning techniques
ICASSP 2010

Robust RVM regression using sparse outlier model
CVPR 2010

Computer vision (Face recognition and SfM)


Saliency guided Wavelet compression forlow-bitrate Image and Video coding
IEEE GlobalSIP 2015

Blur and illumination robust face recognition
IEEE TIP 2013

Recognition of motion blurred faces
Book chapter in Motion Deblurring, CUP, 2014

Projective bundle adjustment using L-infty norm
ICVGIP 2008

Webpage credits

This webpage is modeled after Aswin's webpage.