Human image segmentation
S. R. Kharabe, P. S. Hanwate, K. P. Kaliyamurthie, Dhananjay S. Gaikwad · 2017 International Conference on Algorithms, Methodology, Models and Applications in Emerging Technologies (ICAMMAET) · 2017
Instance based human segmentation works on various parameters like labeling of an image at pixel level, partitioning it into distinct instances and the background of the image. The localization, identification and extraction of human image with reliable appearance in a surveillance video are a widely used applications now days. Due to the strong changes in foreground and background and irregularly occurring foreground humans make this problem challenging. The advances in object detection, scene understanding and image co-segmentation, this paper focuses on label and segment human objects. An efficient human instance detector is used to detect exact human objects in the combination with an extended color line model with a poselet-based human detector. It process in three folds one initialization by the human detector, two further enhancement through a generalized geodesic distance transform, and three final refinement with a joint bilateral filter. The integration of the images takes place of the high-level cues from the detector and shape with the low-level. The high-level object cues from the detector and the shape are then integrated with the low-level pixel cues. The mid-level contour cues into a principle CRF framework, which can be efficiently solved by using fast graph cut algorithms.