Social media video summarization using multi-Visual features and Kohnen's Self Organizing Map

作者：

Highlights：

• Social media video summarization technique based upon multi-visual feature fusion and Kohonen's Self Organizing Map is proposed.

• Both color and structure-based features are used to enhance the quality of key frame extraction. An adaptive formula is used to handle gradual changes occur due to lightening conditions.

• Fidelity and SRD values are used to compare the proposed system with the existing techniques.

• Phase based motion interpolation scheme is used to find the reconstructed interpolated frames between two key frames for determining the value of shot reconstruction degree.

摘要

•Social media video summarization technique based upon multi-visual feature fusion and Kohonen's Self Organizing Map is proposed.•Both color and structure-based features are used to enhance the quality of key frame extraction. An adaptive formula is used to handle gradual changes occur due to lightening conditions.•Fidelity and SRD values are used to compare the proposed system with the existing techniques.•Phase based motion interpolation scheme is used to find the reconstructed interpolated frames between two key frames for determining the value of shot reconstruction degree.

论文关键词：Video summarization,Key frames,Kohonen SOM,Feature fusion,Phase based interpolation motion scheme

论文评审过程：Received 11 August 2019, Revised 23 November 2019, Accepted 22 December 2019, Available online 28 December 2019, Version of Record 28 December 2019.

论文官网地址：https://doi.org/10.1016/j.ipm.2019.102190