Elastic Net Constraints for Shape Matching

Abstract

We consider a parametrized relaxation of the widely adopted quadratic assignment problem (QAP) formula-tion for minimum distortion correspondence between de-formable shapes. In order to control the accuracy/sparsity trade-off we introduce a weighting parameter on the com-bination of two existing relaxations, namely spectral and game-theoretic. This leads to the introduction of the elastic net penalty function into shape matching problems. In com-bination with an efficient algorithm to project onto the elas-tic net ball, we obtain an approach for deformable shape matching with controllable sparsity. Experiments on a stan-dard benchmark confirm the effectiveness of the approach. 1.