Feature Extraction for Temporal Signal Recognition: An Overview
This is an incremental review paper aimed at researchers in pattern recognition, summarizing existing taxonomies to help identify suitable features for specific tasks.
The paper provides an overview of feature extraction methods for temporal signal recognition, focusing on audio and human behavior analysis, without presenting new experimental results or specific performance metrics.
Due to the huge progress of the recording devices, data from heterogeneous nature can be recorded, such as spatial, temporal and spatio-temporal. Nowadays, time-based data is of particular interest since it has the ability to capture the characteristics evolution of the data over time. The temporal data could be gait, auditory scene, piece of music, and so on. In this paper, we are particularly interested in feature extraction for two different temporal recognition applications namely, audio and human behavior analysis and recognition. Indeed, relevant and discriminative features are of critical and fundamental importance to achieve high performances in any automatic pattern recognition system. This work is intended to provide researchers with a brief overview of the different existing features through an understanding of basic taxonomies which may serve as a reference to identify the adequate features for a specific task.