MIME-Version: 1.0 Server: CERN/3.0 Date: Monday, 25-Nov-96 00:17:04 GMT Content-Type: text/html Content-Length: 2536 Last-Modified: Thursday, 09-May-96 19:26:21 GMT Feature-Based Algorithms for Detecting and Classifying Scene Breaks

Feature-Based Algorithms for Detecting and Classifying Scene Breaks


Ramin Zabih
Justin Miller
Kevin Mai


Abstract

We describe a new approach to the detection and classification of scene breaks in video sequences. Our methods can detect and classify a variety of scene breaks, including cuts, fades, dissolves and wipes, even in sequences involving significant motion. We use a feature-based approach, and detect the appearance of intensity edges far from previous edges. A global motion computation is used to handle camera or object motion. The algorithms we propose withstand compression artifacts such as those introduced by JPEG and MPEG, even at very high compression rates. Experimental evidence demonstrates that our method can detect and classify scene breaks that are difficult to detect with conventional approaches. An initial implementation runs at about 2 frames per second on a Sun workstation.


If you wish to view the paper, you can choose from two options:

HTML
The paper that will appear in the 1995 ACM Multimedia Conference. Also available in postscript form.
Postscript
The rather large (4.5 MB!) postscript paper submitted to IEEE PAMI. Note: some printers have trouble printing this out. If you send us email we will send you a hardcopy.
In addition, Rob Szewczyk has built an MPEG browser, which uses our algorithm to detect scene breaks. It allows a user to search for a variety of scene breaks, in combination with motion. For instance, it is possible to skip forward to the next time a left pan is followed by a dissolve. Source code for the browser and the algorithm are available on an "as is" basis. In addition, there is a binary version of the browser compiled under SunOS.

rdz@cs.cornell.edu
Mon Jul 10 19:37:03 EDT 1995