Searching for Repetitive Patterns in 3D Human Motion Captured Data

Description

Repetitive patterns are segments that appear frequently in a given signal. The signal data can be either discrete or continuous. For a discrete signal, repetitive patterns can be found by exact matching with string matching techniques or inexact matching with dynamic programming. Unlike discrete signals, a continuous signal contains an infinite number of states, so it is impractical to identify exactly matched states. Hence, a point cloud similarity approach is proposed to inexactly match similar segment pairs and then approximate the optimal matching.