Methods of Video Data Mining

Submitted by: Submitted by

Views: 193

Words: 583

Pages: 3

Category: Science and Technology

Date Submitted: 10/27/2013 08:01 AM

Report This Essay

1st Paper - Presents visual browsing as an alternative to visual indexing. Various techniques of videos browsing are presented – storyboard, slide shows, audio speedups, animation and adaptive accelerating fast playback

2nd Paper - Concentrates on a tool to extraction of semantic content using non temporal clustering of scene segments into conceptual form grounded in the recognition of real world back ground .

Used for browsing videos when people remember it as some idea. Physical Setting is the keyword. A tree like structure where the root is called the physical setting. Assumptions based on video genre made.

3rd Paper – Making assumption based on video genre also means a user provided metadata is available. But in case it is not available , this paper presents a way to automatically detect how the internal structure of the video is organized. This they achieve with the help of temporal boundaries. Temporal boundaries are classified into three types micro, macro and mega. Micro boundaries consist of “shortest observable temporal segments” in continuous video footage. Macro boundaries aim at conceptual partition between video segment. For instance different stories covered in a news broadcast represent different macro boundaries. Mega boundary attempts to separate the program material from the non program material. The paper cites a news broadcast and commercial aired in it as an example of mega boundary

4th Paper – The highlights of a video segment can be captured with help of motion activity descriptor. Perception wise highlights can be seen as more intense moments for example the presenting the headline in a news story. MPEG-7 sound recognition can help detect semantic fragmentation of a video. This two methods in combination can help produce summary of an entire video sequence.

5th Paper – Content based video analysis attempts to understand a video just as humans do. This is another method where summary of a video is made but here the attempt to is...