Paper Title
Modified Fuzzy C-Means Clustering For Video Summarization
Abstract
Due to the raised amount of video data, many problems have occurred such as limited storage and unsatisfying
video quality. To overcome this problem, a video summary with the least semantic information loss is proposed. Creating a
video summary involves segmenting the visual and audio from the original video and extracting representative information
from these videos. Modified Fuzzy C-Means Clustering (MFCM) method uses three audio features namely Mel Frequency
Cepstral Coefficients (MFCC), Short time energy (STE) and Zero Cross rate (ZCR) to segment the audio. For segmenting
video it involves (i) Shot detection (ii) Sub-shot classification (iii) Key frame extraction. A new key frame extraction based
on color histogram using fuzzy c means clustering is proposed. Finally all the above extracted components are integrated
into a compact video.