We have recently developed an original tracking framework allowing to track textured templates in real time [3]. This framework is based on the use of difference images, the difference between the target template and the template included in predicted area of interest. For efficiency purposes, the difference image was limited to a few points belonging to the area of interest. The measurements were therefore very punctual. In this article a wavelet representation of the area of interest is used as a substitute for this punctual measurement. In this case, the difference image is a difference of wavelet parameters (difference between the wavelet representation of the target and the wavelet representation of the template included in the current position of the region of interest). This algorithm is a part of a real-time system aiming at automatically detecting and tracking vehicles in video sequences. © Springer-Verlag Berlin Heidelberg 2002.
CITATION STYLE
Chateau, T., Jurie, F., Dhome, M., & Clady, X. (2002). Real-time tracking using wavelet representation. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2449, 523–530. https://doi.org/10.1007/3-540-45783-6_63
Mendeley helps you to discover research relevant for your work.