Multi-scale level set based curve evolution for real-time non-rigid object contour tracking

Abstract

In this paper, an efficient multi-scale level set method is proposed to track a non-rigid object contour in real time. The proposed algorithm consists of two cycles which are carried out in different scale domains. In the first cycle on the coarse scale, the algorithm evolves the contour depending on the observed data. In the other cycle on the fine scale, the smoothness regularization is imposed on the curve. In order to maintain the curve data during the scale transition, we propose a simple pattern to interpolate empty grids effectively. In addition, we propose a position prediction scheme so as to accelerate the convergence rate of the algorithm. Experimental results show that dramatic speedup is achieved on video tracking experiments without the performance degradation.

Multi-scale local region based level set method for image segmentation in the presence of intensity inhomogeneity

“…Keeping the level set function ϕ fixed, we minimize the total energy functional in (25) with respect to m 1 ðϕÞ and m 2 ðϕÞ. By calculus of variations, the constant functions m 1 ðϕÞ and m 2 ðϕÞ can be computed as follows:…”

“…Recently, inspired by the multi-scale idea in wavelet transform, some multi-scale level set methods are proposed [24][25][26]. Lin et al [24] applied the coarse scale to extract image boundary and then utilized the extracted coarse boundaries as initial contour to drive contour evolution at fine scale.…”

“…Lin et al [24] applied the coarse scale to extract image boundary and then utilized the extracted coarse boundaries as initial contour to drive contour evolution at fine scale. Kim et al [25] proposed using two evolving curves for level set evolution where one curve at the coarse scale tracked the object boundary and the other evolving curve at fine scale was used to smooth the object boundary. Sui et al [26] decomposed the original image into several sub-images at different scale and performed level set evolution in sub-image with finer scale where the segmentation result of sub-image with coarse scale was regarded as the initial contour.…”