Evans and WNCG Student Receive IEEE Top Paper Award
WNCG Ph.D. student Chao Jia and his advisor Prof. Brian L. Evans received a Top 10% Paper Award at the 14th IEEE Multimedia Signal Processing Workshop held in September. Their paper was entitled “Probabilistic 3D Motion Estimation for Rolling Shutter Video Rectification from Visual and Inertial Measurements.”
This paper addressed the long-standing problem of warping artifacts that appear during video recording on smart phones and many other handheld digital cameras. In these handheld products, the camera does not have a hardware shutter in order to save cost and reduce weight and physical volume. During video acquisition, light continually impinges on the camera's image sensor array. The sensor array is read one row at a time, which causes the sensor row to be reset. This arrangement is called a rolling shutter. The amount of time that elapses from reading the first row to reading the last row can lead to significant warping artifacts in the video frame, especially when there is fast motion in the scene relative to the camera.
In this paper, the authors fuse two sources of information to reduce rolling shutter artifacts. The first source is the gyroscope readings, which indicate camera rotation. These readings have to be synchronized with the video camera and interpolated to match the much higher rate at which rows in the image sensor array are read. The second source is finding and tracking feature points in the video frames. By developing a framework based on extended Kalman filtering, the authors were able to correct warping artifacts in several test video sequences taken on a smart phone.
View the award here (http://www.mmsp2012.org/top10p_paper_award.html) and the paper here (http://users.ece.utexas.edu/~bevans/papers/2012/rolling/index.html).
Sanghavi and Caramanis Receive NSF Grant
The National Science Foundation awarded WNCG Professors Sujay Sanghavi and Constantine Caramanis, along with a colleague at UC Berkeley, a $1.1 million grant to advance the frontiers of large-scale machine learning in the era of big and noisy data.
The proposed research focuses on the development of both fundamental new theory and algorithms for data that lives in very high-dimensional spaces; the dimensionality renders basic statistical tricks ineffective. Applications include, for instance, the problem of designing recommender systems, such as those used by Amazon, Netflix and other various online companies. This involves analyzing large matrices that describe users' behavior in past situations. In sociology, researchers are interested in fitting networks to large-scale data sets, involving hundreds or thousands of individuals. In medical imaging, the goal is to reconstruct complicated phenomena (e.g., brain images; videos of a beating heart) based on a minimal number of incomplete and possibly corrupted measurements.
Motivated by such applications, the goal of this research is to develop and analyze models and algorithms for extracting relevant structure from such high-dimensional data sets in a robust and scalable fashion. Robustness is a particular issue in high dimensions, and existing algorithms are especially fragile to corruption and model mismatch issues that are rife in modern large data sets.
The research leverages tools from convex optimization, signal processing, and robust statistics. It also further facilitates the recent addition of such ideas to the ECE curriculum, in the form of the two-class sequence on optimization and learning co-taught by the PIs in 2012-13.
Dr. Sujay Sanghavi is an assistant professor with WNCG, and Dr. Constantine Caramanis is an associate professor WNCG. Both of their work focuses on machine learning, statistics, large-scale networks and optimization.