Abstract
We describe novel methods to automatically augment content into video highlights detected from soccer and tennis video. First, we extract generic and domain-specific features from the video to isolate key audio-visual events that we have empirically found to correlate well with the ground-truth highlights. Next, based on a set of heuristics-driven rules to minimize view disruption, spatial regions in the image frames of these video highlight segments are segmented for content augmentation. Preliminary trials from subjective viewing indicate a high level of acceptance for the content insertions. © Springer-Verlag Berlin Heidelberg 2004.
Cite
CITATION STYLE
Wan, K., Wang, J., Xu, C., & Tian, Q. (2004). Automatic sports highlights extraction with content augmentation. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 3332, 19–26. https://doi.org/10.1007/978-3-540-30542-2_3
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
