Bagging Stochastic Watershed on Natural Color Image Segmentation

• Main Authors: Franchi, Gianni, Angulo, Jesus
• Other Authors: Centre de Morphologie Mathématique (CMM), MINES ParisTech - École nationale supérieure des mines de Paris, Université Paris sciences et lettres (PSL)-Université Paris sciences et lettres (PSL)
• Format: Conference Object
• Language: English
• Published: HAL CCSD 2015
• Subjects: Berkeley segmentation database; unsupervised image segmentation; stochastic watershed; [INFO.INFO-TS]Computer Science [cs]/Signal and Image Processing; [INFO.INFO-CV]Computer Science [cs]/Computer Vision and Pattern Recognition [cs.CV]; Iceland
• Online Access: https://hal.archives-ouvertes.fr/hal-01104256; https://hal.archives-ouvertes.fr/hal-01104256/document; https://hal.archives-ouvertes.fr/hal-01104256/file/Bagging_Stochastic_Watershed_FRANCHI_ANGULO_V2.pdf; https://doi.org/10.1007/978-3-319-18720-4_36

International audience The stochastic watershed is a probabilistic segmentation ap-proach which estimates the probability density of contours of the image from a given gradient. In complex images, the stochastic watershed can enhance insignificant contours. To partially address this drawback, we introduce here a fully unsupervised multi-scale approach including bag-ging. Re-sampling and bagging is a classical stochastic approach to im-prove the estimation. We have assessed the performance, and compared to other version of stochastic watershed, using the Berkeley segmentation database.