Object recognition and segmentation by non-rigid quasi-dense matching

In this paper, we present a non-rigid quasi-dense match-ing method and its application to object recognition and segmentation. The matching method is based on the match propagation algorithm which is here extended by using lo-cal image gradients for adapting the propagation to smooth non-rigid deformations of the imaged surfaces. The adap-tation is based entirely on the local properties of the im-ages and the method can be hence used in non-rigid im-age registration where global geometric constraints are not available. Our approach for object recognition and seg-mentation is directly built on the quasi-dense matching. The quasi-dense pixel matches between the model and test im-ages are grouped into geometrically consistent groups us-ing a method which utilizes the local afne transforma-tion estimates obtained during the propagation. The num-ber and quality of geometrically consistent matches is used as a recognition criterion and the location of the matching pixels directly provides the segmentation. The experiments demonstrate that our approach is able to deal with exten-sive background clutter, partial occlusion, large scale and viewpoint changes, and notable geometric deformations. 1