Automated content-based video analysis and management

Rapid expansion in the use of digital videos as an information source has led to a significant increase in the availability and the amount of video data. Multimedia applications use and generate large volume of complex video data sets. Since manual indexing, searching, browsing, and retrieval of relevant information are both computationally expensive and time consuming, efficient and reasonable mechanisms that can perform these operations are needed. This dissertation sorts out the challenges of video processing for automated content-based video analysis and management. More specifically, the work presented here include temporal video segmentation to partition video sequences into shots, extracting subset of representative key frames from both compressed and uncompressed video sequences to create video summaries and enable content-based video browsing and retrieval, construction of video indexing schemas for easily browsable and accessible video content in both web and mobile environments, video classification framework to categorize the video segments that will ease in accessing the relevant video content without sequential scanning and hierarchical clustering based schema for video annotation to organize the video data in a tree-based story structure. This research aims helping to facilitate effective video analysis to provide better understanding of video content. The techniques that we propose could enable a more reliable and efficient video content description.