One of the major limitations of HMM-based models is the inability to cope with topology: When applied to a visible observation (VO) sequence, HMM-based techniques have difficulty ...
—In this paper, we propose a circular blurred shape model descriptor to deal with the problem of symbol detection and classification as a particular case of object recognition. ...
Abstract. We present a novel method for the segmentation of volumetric images, which is especially suitable for highly variable soft tissue structures. Core of the algorithm is a s...
Abstract. We present a machine learning approach called shape regression machine (SRM) to segmenting in real time an anatomic structure that manifests a deformable shape in a medic...
Symbol spotting problem requires feature extraction strategies able to generalize from training samples and to localize the target object while discarding most part of the image. ...