Conditional random fields for image region labeling with global observation

Abstract

Region (pixel) labeling has attracted increasing attentions from both research and industry communities. In this paper, we present a new approach based on Conditional Random Fields (CRF) to assign the semantic labels to the corresponding regions of images. Different from previous work, our model incorporates the global observation into the region labeling framework with the harness of spatial context modeling of CRF model. The experimental results with two commonly used datasets demonstrate that our method achieves significant improvement on region labeling tasks compared with the strong baselines. © Springer International Publishing Switzerland 2013.