Symmetry Structured Analysis Sparse Coding for Key Frame Extraction

Abstract

AbstractThe efficiency of sparse coding based key frame extraction algorithm is influenced by various sparse regularization and optimization strategies. However, sparse coding with an analytical model for key frame extraction is still a challenging task. In this paper, we present a new analysis sparse coding algorithm for key frame extraction using minimax concave penalty (MCP). Analysis sparse coding has low computation complexity compared to the common synthesis model. Furthermore, analysis sparse coding can automatically lead to symmetry structured for key frame extraction. In this context, the MCP sparse regularization is non-convex that can promote the sparsity of solutions. Unlike conventional non-convex sparse regularization in formulating a non-convex sparse coding cost function, MCP can maintain the convexity that can be used to solve the optimization problem for obtaining the global minimum. The proposed key frame extraction algorithm leads into the following: 1) provides more compressed key frames, 2) decreases the computational complexity and 3) accelerates the process tasks. Our results demonstrate the effectiveness of the proposed symmetry structured with analysis sparse coding algorithm that is validated with both simulations and a number of challenging real-world scenarios, outperforming the state-of-the-art techniques.KeywordsKey frame extractionAnalysis sparse codingMinimax Concave Penalty (MCP)Non-convex