Content-based image retrieval by integration of metadata encoded multimedia (image and text) features in constructing a video summarizer application.

Description:

Content-based image retrieval (CBIR) is the retrieval of images from a collection by means of internal feature measures of the information content of the images. In CBIR systems, text media is usually used only to retrieve exemplar images for further searching by image feature content. This research work describes a new method for integrating multimedia text and image content features to increase the retrieval performance of the system. I am exploring the content-based features of an image extracted from a video to build a storyboard for search retrieval of images. Metadata encoded multimedia features include extracting primitive features like color, shape and text from an image. Histograms are built for all the features extracted and stored in a database. Images are searched based on comparing these histogram values of the extracted image with the stored values. These histogram values are used for extraction of keyframes from a collection of images parsed from a video file. Individual shots of images are extracted from a video clip and run through processes that extract the features and build the histogram values. A keyframe extraction algorithm is run to get the keyframes from the collection of images to build a storyboard of images. In video retrieval, speech recognition and other multimedia encoding could help improve the CBIR indexing technique and makes keyframe extraction and searching effective. Research in area of embedding sound and other multimedia could enhance effective video retrieval.

Creator(s): Anusuri, Ramprasad
Creation Date: May 2003
Partner(s):
UNT Libraries
Collection(s):
UNT Theses and Dissertations
Usage:
Total Uses: 524
Past 30 days: 6
Yesterday: 0
Creator (Author):
Publisher Info:
Publisher Name: University of North Texas
Place of Publication: Denton, Texas
Date(s):
  • Creation: May 2003
  • Digitized: September 9, 2003
Description:

Content-based image retrieval (CBIR) is the retrieval of images from a collection by means of internal feature measures of the information content of the images. In CBIR systems, text media is usually used only to retrieve exemplar images for further searching by image feature content. This research work describes a new method for integrating multimedia text and image content features to increase the retrieval performance of the system. I am exploring the content-based features of an image extracted from a video to build a storyboard for search retrieval of images. Metadata encoded multimedia features include extracting primitive features like color, shape and text from an image. Histograms are built for all the features extracted and stored in a database. Images are searched based on comparing these histogram values of the extracted image with the stored values. These histogram values are used for extraction of keyframes from a collection of images parsed from a video file. Individual shots of images are extracted from a video clip and run through processes that extract the features and build the histogram values. A keyframe extraction algorithm is run to get the keyframes from the collection of images to build a storyboard of images. In video retrieval, speech recognition and other multimedia encoding could help improve the CBIR indexing technique and makes keyframe extraction and searching effective. Research in area of embedding sound and other multimedia could enhance effective video retrieval.

Degree:
Level: Master's
Discipline: Computer Science
Note:

Problem in Lieu of Thesis

Language(s):
Subject(s):
Keyword(s): CBIR | video retrieval | storyboard | metadata encoded | multimedia
Contributor(s):
Partner:
UNT Libraries
Collection:
UNT Theses and Dissertations
Identifier:
  • ARK: ark:/67531/metadc4238
Resource Type: Thesis or Dissertation
Format: Text
Rights:
Access: Public
License: Copyright
Holder: Anusuri, Ramprasad
Statement: Copyright is held by the author, unless otherwise noted. All rights reserved.