E. Martínez-Martín and A.P. del Pobil (Spain, Korea)
Motion detection; background maintenance; object tracking; image segmentation; video processing.
Motion detection plays a fundamental role in any object tracking or visual surveillance algorithm and many different approaches exist for motion segmentation. However, recent comparisons show that often basic methods are equivalent to more complex ones and the few existing hybrid approaches that combine frame differencing and background subtraction are surprisingly effective and very fast. We claim, and show quantitatively, that a new hybrid algorithm based on a proper combination of basic methods can outperform many of the most commonly-used methods. It is based on frame differencing and background subtraction, along with a single-Gaussian background model and a maintenance mechanism. We evaluate its performance using existing public benchmark datasets. It is simple, efficient, robust and effective.
Important Links:
Go Back
