In-source video summarization
Kousik Sankar Ramasubramaniam, Ganesankumar Annamalai · 2016
State-of-the-art broadcast/transmission standards such as ATSC/DVB/ISDB are based on MPEG as the underlying compression standard. As AV content increases and as newer applications/services - under the Internet-of-Things paradigm-get enabled, which require different handling of audio/video metadata; it becomes increasingly infeasible to use the traditional video summarization algorithms due to the disparate nature of the AV content and the associated meta-data. This paper presents a mechanism to define and efficiently utilize the meta-data for summarization without any heavy-duty processing on the video encoding / decoding pipeline. For production house content as well as user generated content, the meta-data can be inserted into the stream. The main advantage of this approach is the portability across various platforms and decoder implementations. The overheads on the bit-stream, processing and memory due to such a change are also depicted.