Comments
Description
Transcript
Kaushik Mitra
Kaushik Mitra Department of Electrical and Computer Engineering 2050 Duncan Hall, Rice University, Houston, Texas 77025, USA Research Interests Academic Experience Publications Cell Phone: +1(301)741-3953 Email: [email protected] Web: http://www.ece.rice.edu/ km23/ Computational Imaging, Computer Vision, Machine Learning and Optimization Techniques Postdoctoral Research Associate Aug 2011 - Electrical and Computer Engineering Rice University, Houston Research Area: Computational Imaging Advisor: Prof. Ashok Veeraraghavan M.S. and Ph.D. Aug 2004 - Aug 2011 Electrical and Computer Engineering University of Maryland, College Park Research Area: Computer Vision Advisor: Prof. Rama Chellappa M.E. Aug 2001 - Jan 2003 Electrical Communication Engineering Indian Institute of Science, Bangalore, India Research Area: Computer Network Advisor: Prof. Anurag Kumar B.Tech. Aug 1998 - Aug 2001 Radiophysics and Electronics University of Calcutta, Kolkata, India Project: Analytic Study of mm-Wave Oscillator Advisor: Prof. Subal Kar Book Chapters • Recognition of Motion Blurred Faces, K. Mitra, P. Vageeswaran and R. Chellappa, Motion Deblurring: Algorithms and Systems, Cambridge University Press, A. N. Rajagopalan and R. Chellappa (Editors), 2014. Journal Publications • Generalized Assorted Camera Arrays: Robust Cross-channel Registration and Applications, J. Holloway, K. Mitra, S. Koppal, A. Veeraraghavan, in preparation for submission to IEEE Transactions on Image Processing (TIP), to be submitted in March 2014. • Compressive Epsilon Photography for Post-Capture Control in Digital Imaging, A. Ito, S. Tambe, K. Mitra, A. Sankaranarayanan, A. Veeraraghavan, accepted in SIGGRAPH (ACM Transactions on Graphics), 2014. • A Framework for Analysis of Computational Imaging Systems: Role of Signal Prior, Sensor Noise and Multiplexing, K. Mitra, O. Cossairt and A. Veeraraghavan, accepted in IEEE Transactions on Pattern Analysis and Machine Learning (TPAMI), 2014. • Towards Compressive Camera Networks, K. Mitra, A. Veeraraghavan, A. Sankaranarayanan and R. G. Baraniuk, IEEE Computer, May 2014. • Blur and Illumination Robust Face Recognition via Set-Theoretic Characterization, P. Vageeswaran, K. Mitra and R. Chellappa, IEEE Transactions on Image Processing (TIP), 2013. • Analysis of Sparse Regularization Based Robust Regression Algorithms, K. Mitra, A. Veeraraghavan and R. Chellappa, IEEE Transactions on Signal Processing (TSP), 2013. Conference Publications • Can we Beat Hadamard Multiplexing? Data-driven Design and Analysis for Computational Imaging Systems, K. Mitra, O. Cossairt and A. Veeraraghavan, IEEE International Conference on Computational Photography (ICCP), 2014. • Improving Resolution and Depth-of-Field of Light Field Cameras Using a Hybrid Imaging System, V. Boominathan, K. Mitra and A. Veeraraghavan, IEEE International Conference on Computational Photography (ICCP), 2014. • To Denoise or Deblur: Parameter Optimization for Imaging Systems, K. Mitra, O. Cossairt and A. Veeraraghavan, SPIE Electronic Imaging, 2014. • Performance Bounds for Computational Imaging, O. Cossairt, A. Veeraraghavan, K. Mitra and M. Gupta, Imaging and Applied Optics Technical Papers, OSA, 2013. • Performance Limits for Computational Photography, O. Cossairt, K. Mitra and A. Veeraraghavan, International Workshop on Advanced Optical Imaging and Metrology, Springer, 2013. • Light Field Denoising, Light Field Superresolution and Stereo Camera Based Refocussing using a GMM Light Field Patch Prior, K. Mitra and A. Veeraraghavan, CVPR Workshop on Computational Cameras and Displays, 2012. • A Hierarchical Approach For Human Age Estimation, P. Thukral, K. Mitra and R. Chellappa, IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) 2012. • Large-Scale Matrix Factorization with Missing Data under Additional Constraints, K. Mitra, S. Sheorey, R. Chellappa, Advances in Neural Information Processing Systems (NIPS) 2010. • Robust RVM Regression Using Sparse Outlier Model, K. Mitra, A. Veeraraghavan and R. Chellappa, IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2010. • Robust Regression Using Sparse Learning for High Dimensional Parameter Estimation Problem, K. Mitra, A. Veeraraghavan and R. Chellappa, IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) 2010. • A Scalable Projective Bundle Adjustment Algorithm using the L∞ Norm, K. Mitra and R. Chellappa, Proceedings of the Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP) 2008. Research Experience Rice University, Houston, TX. Postdoctoral Research Associate, 2011 I am working in the new and emerging field of Computational Imaging (CI). Traditionally, imaging/camera design and computer vision (inference from imaging data) are considered as independent fields. However, significant advantage can be obtained by designing the imaging devices and inference algorithms in a unified framework. Recently, many CI systems have been proposed for tackling the problem of motion blurring (for example Flutter Shutter), focus blurring (for example coded aperture cameras and extended depth of field cameras) , light field capture, etc. Given that there are so many CI systems, it is imperative to compare and characterize their performance. I proposed a framework for analysis of CI systems that characterizes the performance of the systems by taking into account the effect of optical coding, image prior and sensor noise (TPAMI 2014). I then used this framework to find optimal camera parameters for conventional camera (SPIE 2014) and to design optimal optical codes for CI systems (ICCP 2014). I am also interested in designing novel CI systems. Current light field cameras (such as Lytro) have very poor spatial resolution. I have proposed a hybrid imaging system, consisting of a standard light field camera and a high resolution conventional camera, to generate high resolution light fields (ICCP 2014). I have also proposed a general framework for light field processing tasks such as denoising and (angular and spatial) superresolution (CVPR workshop 2012). Recently, I have proposed compressive epsilon photography, a technique for achieving post-capture control of focus, iso, exposure and aperture in a traditional camera by acquiring a carefully selected set of images and computationally reconstructing images corresponding to other focus and aperture settings (under review in SIGGRAPH, 2014). University of Maryland, College Park, MD. Graduate Research Assistant, 2004 - 2011 I proposed robust and efficient machine learning algorithms for a large class of computer vision problems. Many computer vision problems can be formulated as learning problems such as regression and matrix factorization. However, because of variations due to viewpoint, occlusion, shadows, etc., the visual data suffers from significant outliers and missing data, and this makes the learning problems very challenging. Towards solving the problem of outliers, I have proposed robust regression algorithms, based on sparse regularization and Bayesian techniques, that can handle large amount of outliers (CVPR 2010, ICASSP 2010). I have performed analysis of the robust algorithms to provide estimates of the fraction of outliers in a dataset that the proposed algorithms can successfully handle (TSP 2013). Towards solving the problem of missing data in matrix factorization, I have formulated it as a low-rank semidefinite programming problem with the advantage that it can handle large-scale vision datasets (NIPS 2010). I have also worked in the traditional vision problems of Structure from Motion (SfM) and face recognition. In SfM I have proposed a scalable algorithm for projective bundle adjustment (ICVGIP 2008) and in face recognition I have proposed an algorithm to identify blurred and poorly illuminated faces (TIP 2013). Indian Institute of Science, Bangalore, India. Master’s thesis research, 2002 I worked on queuing models for HTTP traffic under the supervision of Prof. Anurag Kumar. Teaching Experience Rice University, Houston, TX. Designed assignments and projects for graduate-level computer vision course (Elec 547). Delivered guest lectures in computer vision (Elec 345, Elec 547) and computation photography (Elec 549) courses. University of Maryland, College Park, MD. Delivered guest lectures in the graduate-level image understanding course (ENEE 731). Work Experience Samsung India Software Operations, Bangalore, India. Senior Software Engineer, 2003-2004 I worked on the multimedia software (audio, video, streaming players) of the Samsung mobile phone platform. Honors • Ranked Second in M.E. (IISc, Bangalore). • All India Rank Second in GATE (Admission test for graduate studies in Engineering). • Ranked First in B.Tech (University of Calcutta). Talks and Poster Presentations • To Denoise or Deblur: Parameter Optimization for Imaging Systems, SPIE Electronic Imaging, San Francisco January 2014. • Light Field Denoising, Light Field Superresolution and Stereo Camera Based Refocussing using a GMM Light Field Patch Prior, CVPR Workshop on Computational Cameras and Displays, Providence, Rhode Island, June 2012. • Handling Outliers and Missing Data in Statistical Data Models, talk at Electronics and Communication Sciences Unit (ECSU), Indian Statistical Institute (ISI), Kolkata, India,January 2011. • Large-Scale Matrix Factorization with Missing Data under Additional Constraints, poster presentation at Advances in Neural Information Processing Systems (NIPS), Vancouver, B.C., Canada, December 2010. • Fitting Models to Data: Handling Outliers and Missing Data, talk at Mitsubishi Electric Research Laboratories (MERL), Cambridge, July 2010. • Fitting Models to Data: Handling Outliers and Missing Data, talk at MIT Media Lab, Camera Culture Lab, July 2010. • Robust RVM Regression Using Sparse Outlier Model, poster presentation at the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), San Francisco,June 2010. • Robust Regression Using Sparse Learning for High Dimensional Parameter Estimation Problem, poster presentation at the International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Dallas, March 2010. • Scalable Bayesian Robust Regression for High Dimensional Applications, talk at Statistics Department, Florida State University, March 2009. • A Scalable Projective Bundle Adjustment Algorithm using the L∞ Norm, talk at the Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP), Bhubaneswar, India, December 2008. References Rama Chellappa ECE Department University of Maryland College Park, MD 20742 Phone: 301-405-4526 [email protected] Ashok Veeraraghavan ECE Department Rice University Houston, Tx 77005 Phone: 713-348-5104 [email protected] Richard Baraniuk ECE Department Rice University Houston, Tx 77005 Phone: 713-348-5132 [email protected] B. S. Manjunath ECE Department University of California Santa Barbara, CA 93106 Phone: 805-893-7112 [email protected] Amit Roy-Choudhury EE Department University of California Riverside, CA 92521 Phone: 951-827-7886 [email protected] Oliver Cossairt EECS Department Northwestern University Evanston, IL 60208 Phone: 847-491-0895 [email protected]