Natural Image Statistics for Natural Image Segmentation

Analysis of orientation and scale in smoothly varying textures

“…The left column illustrates the scalar segmentation described in [9] which is based solely on intensity values. The middle column is obtained from the algorithm described in [7]. In contrast to our method which exploits joint dependencies, this method uses all outputs of the filter bank assuming they are statistically independent.…”

“…In a discriminative model, a robust texture representation can aid in segmentation to distinguish different objects (e.g. [7]). Here, we present a texture model subject to a deformation field that when combined with information measures is useful for a variety of common computer vision tasks.…”

TouchCut: Fast image and video segmentation using single-touch interaction

“…Chan and Vese [14] propose the approach of active contours also harness Gaussian distributions, partitioning objects via intensity distributions modelled with different variances. Heiler et al [32] also adopt Laplacan distributions for natural image segmentation.…”

Feature Quarrels: The Dempster-Shafer Evidence Theory for Image Segmentation Using a Variational Framework

“…This Model was chosen since compared to multivariate Gaussian Mixture Models (GMM) it leads to comparable results without estimating so many parameters. Other possibilities to model the probability densities given the image cues are, e.g., a Gaussian density with fixed standard deviation [2] or a generalized Laplacian [19]. For the proposed method it is necessary to extend this segmentation framework from gray scale images to feature vector images I = (I 1 , .…”