HKUST Library Institutional Repository Banner

HKUST Institutional Repository >
Computer Science and Engineering >
CSE Technical Reports >

Please use this identifier to cite or link to this item:
Title: Automatic video data structuring through shot partitioning and key frame selection
Authors: Xiong, Wei
Lee, John Chung-Mong
Ma, Rui-Hua
Keywords: automatic video data structuring
Video partitioning
Net Comparison
Image similarity measure
Invariant parameters
Key frame selection
Two-stage search
Key frame pruning
Issue Date: 16-May-1996
Series/Report no.: Computer Science Technical Report ; HKUST-CS96-13
Abstract: In this paper, we propose automatic solutions to the problems: (i) video partitioning, (ii) key frame selection, (iii) key frame pruning. For the first problem, an algorithm called "Net Comparision" is devised. It is accurate and fast because it uses both statistical and spatial information in an image and does not have to process the entire image. Choice of number and size of subsampling regions is analyzed. we also introduce a novel searching approach - the step-variable search for video partitioning. For the last two problems, we develop an original image similarity criterion which considers both spatial layout and detail content in an image. For this purpose, coefficients of wavelet decomposition are used to derive parameter vectors accounting for these aspects. The parameters exhibit (quasi-) invariant properties, making the algorithm robust to many types of object/camera motion and scaling as well. A novel "Two-Stage Search" strategy is proposed for key frame selection which allows to obtain the largest representation range possible of key frames. Inter-shot redundancy of key frames is suppressed using the same image similarity measure. Experimental results demonstrate the effectiveness and effciency of our techniques.
Appears in Collections:CSE Technical Reports

Files in This Item:

File Description SizeFormat
tr9613.pdf543KbAdobe PDFView/Open

All items in this Repository are protected by copyright, with all rights reserved.