An Experimental Comparison of Discrete and Continuous Shape Optimization Methods

摘要: Shape optimization is a problem which arises in numerous computer vision problems such as image segmentation and multiview reconstruction. In this paper, we focus on certain class of binary labeling can be globally optimized both spatially discrete setting continuous setting. The main contribution paper to present quantitative comparison the reconstruction accuracy computation times allows assess some strengths limitations approaches. We also novel method approximate length regularity graph cut based framework: Instead using pairwise terms introduce higher order terms. These allow represent more accurate discretization L 2 -norm term.