Continuous Wavelet Transform for time-varying motion extraction

Widespread use of digital multimedia data has led to the development of advanced processing techniques for them. Motion estimation is a fundamental step in the extraction of activity in videos, for tracking, motion segmentation, video classification and other applications. One large category of methods processes data in the spatial domain, and another in the frequency domain. This paper presents a novel approach for frequency-domainmotion estimation, which avoids drawbacks of illumination-based methods, based on the the Continuous Wavelet Transform (CWT). The proposed algorithm processes all video frames at the same time to create a frequency modulated (FM) signal, which is then processed using the CWT. The proposed system is shown to be robust to local measurement noise and occlusions, as it processes the available data in a global, integrated manner. Experiments take place with both synthetic and real video sequences to demonstrate the capabilities of the proposed approach.