POSIT:Part-basedobjectsegmentationwithoutintensivetraining

摘要: Object segmentation is a well-known difficult problem in pattern recognition. Until now, most of the existing object methods need to go through time-consuming training phase prior segmentation. Both robustness and efficiency have room for improvement. In this work, we propose new methodology, called POSIT, without intensive process. We construct part-based shape model substitute framework, sequentially register parts an image so that searching space largely reduced. Another advantage sequential matching that, instead predefining weighting parameters terms evaluation function, can estimate our on fly. Finally, fine-tune previous coarse by localized graph cuts. experiments, POSIT has been tested numerous natural horse cow images obtained results show accuracy, proposed method.