The Discriminative Power of Shape in Time Series Matching
Shape provides significant discriminating power in time series matching of visual or geometric data as required in many important applications in graphics and vision. We present a new shape-aware algorithm which uses time and shape correspondence (TSC) at increasing levels of detail to define a similarity measure which is robust to noise and missing data. An L0 norm is used and implicitly regularised using a shapebased error. Through extensive experiments we empirically show that our algorithm performs better than existing state of the art algorithms and works more effectively with high dimensional data. We demonstrate its versatile applicability and comparative performance in human gait data, human action data from a Kinect, hand movements in quaternion stream data using a Myo armband and visual password detection using face tracker data.
Dr. Sudhir Mudur is professor and chair of the department of computer science and software engineering at Concordia University in Montreal, Canada. He completed his Bachelor’s degree from IIT Bombay and his PhD from the Tata Institute of Fundamental Research in Mumbai, India. He has over 4 decades of research, development and teaching experience. His primary research focus is in 3D Graphics and Virtual Worlds.