TEL AVIV UNIVERSITY

אוניברסיטת תל-אביב

The Iby and Aladar Fleishman Faculty of Engineering

School of Electrical Engineering 

Dept of Physical Electronics

הפקולטה להנדסה ע"ש איבי ואלדר פליישמן              

בית הספר להנדסת חשמל

המחלקה לאלקטרוניקה פיזיקאלית

From Camera-Array to CrowdCam

Multi-camera systems  are often used to solve challenging computer vision problems, such as 3D reconstruction, 3D motion estimation, tracking, and super resolution. We discuss the benefits and challenges of multi-camera systems under traditional "restricted" setups, e.g., calibrated and synchronized camera-array. Then, we will consider the scenario of a group of people capturing asynchronously a dynamic scene, using probably the most popular means of photographing today – cellphones. The data obtained this way can be regarded as the output of a new type of extended camera, which we call a crowd-based camera (or CrowdCam).  This data is very rich in content, but lacks accurate temporal information. We define a novel problem of temporally ordering a set of still images taken by CrowdCam. We term this problem photo sequencing and present a geometry-based solution to it. Photo sequencing is an essential step in the analysis of dynamic scenes form a set of still images.