使用動態規劃法之內涵式視訊搜尋
Other Title
A Dynamic Programming Approach to Content-based Video Searching
Date Issued
2000
Date
2000
Author(s)
DOI
892213E002015
Abstract
In this project, we propose a dynamic
programming approach to content-based
video searching. First, each video is
segmented into several shots, each of
which is represented by a key frame.
Each key frame is characterized by a
feature vector, including color, texture,
shape, etc. Second, a user may query the
video database by a query video to search
the best matched video or a set of videos
with the similarity greater than a threshold.
The proposed content-based video
searching algorithm has the following
properties:
1. Query a video database by the content
of an example video.
2. Use a dynamic programming approach
to find the best matched video.
3. The complexity of the algorithm is
bounded by O(mn), where m is the
number of key frames in the query
video and n is the total number of key
frames of videos in the database.
4. Keep the temporal constraints on the
query.
5. Provide similarity search.
6. The more shots are contained in the
query video, the higher precision and
recall are.
Subjects
Video data
Video frame
Shot
Scene
Index
Dynamic Programming
Publisher
臺北市:國立臺灣大學資訊管理學系暨研究所
Type
other
File(s)![Thumbnail Image]()
Loading...
Name
892213E002015.pdf
Size
32.63 KB
Format
Adobe PDF
Checksum
(MD5):3942175e76e2bb60d6e86b2383fde668
