Visual object detection by an active region model
Youssef Chahir, Youssef Zinbi, Abderrahim Elmoataz · 2005
Abstract: In this paper, we propose an unsupervised image segmentation which integrates perceptual informations of regions in an image from colour and texture properties in order to perform image segmentation. This proposed segmentation algorithm is based on a active region model. To correctly take semantic object (region and boundary information) into account, a mixture of Gaussians is used to model pixels of the background image and those of the semantic object. The model minimizes a modified Chan–Vese functional over each component of the colour image, and improves appreciably the chan-vese algorithm and is fast and robust with respect to noise. The results for segmenting objects in an sequence of images are demonstrated.