Comparison of Multiepisode Video Summarization Algorithms

  • PDF / 723,295 Bytes
  • 8 Pages / 600 x 792 pts Page_size
  • 10 Downloads / 255 Views

DOWNLOAD

REPORT


Comparison of Multiepisode Video Summarization Algorithms Itheri Yahiaoui Department of Multimedia Communications, Institut Eur´ecom, BP 193, 06904 Sophia Antipolis, France Email: [email protected]

Bernard Merialdo Department of Multimedia Communications, Institut Eur´ecom, BP 193, 06904 Sophia Antipolis, France Email: [email protected]

Benoit Huet Department of Multimedia Communications, Institut Eur´ecom, BP 193, 06904 Sophia Antipolis, France Email: [email protected] Received 26 April 2002 and in revised form 30 September 2002 This paper presents a comparison of some methodologies for the automatic construction of video summaries. The work is based on the simulated user principle to evaluate the quality of a video summary in a way that is automatic, yet related to the user’s perception. The method is studied for the case of multiepisode video, where we do not describe only what is important in a video but rather what distinguishes this video from the others. Experimental results are presented to support the proposed ideas. Keywords and phrases: multimedia content analysis, video summaries, image similarity, automated evaluation.

1.

INTRODUCTION

The ever-growing availability of multimedia data creates a strong requirement for efficient tools to manipulate and present data in an effective manner. Automatic video summarization tools aim at creating, with little or no human interaction, short versions which contains the salient information of original video. The key issue here is to identify what should be kept in the summary and how relevant information can be automatically extracted. To perform this task, we consider several algorithms and compare their performance to define the most appropriate one for our application. 2.

RELATED WORK

A number of approaches have been proposed to define and identify what is the most important content in a video. However, most have two major limitations. First, evaluation is difficult in the sense that it is hard to judge the quality of a summary or, when a performance measure is available, it is hard to understand its interpretation. Secondly, while the summarization of a single video has received increasing attention [1, 2, 3, 4, 5, 6], little work has been devoted to the problem of multiepisode video summarization [7, 8] which raises other interesting difficulties.

Existing video summarization approaches can be classified in two categories. The rule-based approaches combine evidences from several types of processing (audio, video, text) to detect certain configuration of events to include in the summary. Examples of this approach are the “video skims” of the Informedia Project [3] and the movie trailers of the MoCA project [5]. The mathematically oriented approaches, on the other hand, use similarities within the video to compute a relevance value of video segments or frames. Possible relevance criteria include segments duration, intersegment similarities, and combination of temporal and positional measures. Examples of this approach include the use of sin