http://hdl.handle.net/1721.1/7614 http://hdl.handle.net/1721.1/42411 HeadLock : wide-range head pose estimation for low resolution video DeCamp, Philip (Philip James) This thesis focuses on data mining technologies to extract head pose information from low resolution video recordings. Head pose, as an approximation of gaze direction, is a key indicator of human behavior and interaction. Extracting head pose information from video recordings is a labor intensive endeavor that severely limits the feasibility of using large video corpora to perform tasks that require analysis of human behavior. HeadLock is a novel head pose annotation and tracking tool. Pose annotation is formulated as a semiautomatic process in which a human annotator is aided by computationally generated head pose estimates, significantly reducing the human effort required to accurately annotate video recordings. HeadLock has been designed to perform head pose tracking on video from overhead, wide-angle cameras. The head pose estimation system used by HeadLock can perform pose estimation to arbitrary precision on images that reveal only the top or back of a head. This system takes a 3D model-based approach in which heads are modeled as 3D surfaces covered with localized features. The set of features used can be reliably extracted from both hair and skin regions at any resolution, providing better performance for images that may contain small facial regions and no discernible facial features. HeadLock is evaluated on video recorded for the Human Speechome Project (HSP), a research initiative to study human language development by analyzing longitudinal audio-video recordings of a developing child. Results indicate that HeadLock may enable annotation of head pose at ten times the speed of a manual approach. In addition to head tracking, this thesis describes the data collection and data management systems that have been developed for HSP, providing a comprehensive example of how very large corpora of video recordings may be used to research human development, health and behavior. Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, February 2008. Includes bibliographical references (p. 85-87). http://hdl.handle.net/1721.1/42410 Activity recognition with end-user sensor installation in the home Rockinson, Randy Joseph Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2008. Includes bibliographical references (v. 2, leaves 248-253). http://hdl.handle.net/1721.1/42409 Transformative copy Offenhuber, Dietmar The ability to create an unlimited number of identical copies is a privilege of digital documents. What if that would not be the case, if each copy of a digital file would go along with some sort of transformation? This thesis examines the implications of such a scenario on information ecologies and map out the design space for a new type of decentralized authoring applications. The concept of the copy is inherently transformative, even if just as the transformation from "the One into the Many ". However, until the recent transition from analog to digital media, the perfectly identical copy was not possible, every reproduction resulted in transformation. While this transformation usually is associated with loss of information, it also creates new information, traces of an objects history. As a result of the ability to create of unlimited numbers of perfectly identical copies from digital files, this dimension of context information is lost - digital files do not have a history. In the course of this thesis, five examples were implemented that illustrate important properties of the concept of transformative copying, in order to propose a general framework for a sociable, transformative file format. They investigate two cases of transformation: First, transformation as a deliberate process by humans. A second dimension is transformation as an autonomous process, either in form of an imperfect copy or as a result of an objects age and usage, such as wear. Both points seem like independent cases, but in our scenario they are closely interconnected and inform each other. Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2008. Page 76 blank. Includes bibliographical references (p. 72-75). http://hdl.handle.net/1721.1/42408 Co-channel digital signal separation : application and practice Shen, Dawei This thesis studies the theory and application of co-channel digital signal separation techniques. We set up a test-bed with the GNU Software Defined Radio (SDR) platform where we implement and experiment with single-antenna signal separation algorithms. We mainly investigate linearly-modulated digital signals. To do this, we design a multiple RFID card reader capable of decoding multiple commodity ID cards simultaneously. These passive RFID cards transmit DBPSK waveforms once activated. A signal separation function at the receiver delivers great convenience to the users without increasing the complexity and cost of the cards. Second, we derive the optimal criteria for deciding the start of an RFID frame. We show that the commonly utilized correlation rule is suboptimal and that a correction term needs to be considered to achieve the best detection performance. Several rules for frame synchronization are proposed and analyzed numerically using Monte Carlo simulation. These signal separation techniques present an opportunity to improve the capacity of wireless systems and combat interference. This thesis documents design issues in the physical and application layers, thereby demonstrating the great flexibility and strength of the GNU SDR system. Thesis (S.M.)--Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2008. Includes bibliographical references (leaves 84-86).