Multiscale conditional random fields for image labeling

TLDR

Image labeling components vary in the information they encode, ranging from image‑label mappings to label‑field patterns, and operate at different scales from fine to global. The authors propose incorporating contextual features into pixel‑wise image labeling. They embed contextual features in a probabilistic framework that fuses multiple components and train it with supervised contrastive divergence on labeled images. The approach achieves competitive performance on two real‑world image databases, outperforming a classifier and a Markov random field.

Abstract

We propose an approach to include contextual features for labeling images, in which each pixel is assigned to one of a finite set of labels. The features are incorporated into a probabilistic framework, which combines the outputs of several components. Components differ in the information they encode. Some focus on the image-label mapping, while others focus solely on patterns within the label field. Components also differ in their scale, as some focus on fine-resolution patterns while others on coarser, more global structure. A supervised version of the contrastive divergence algorithm is applied to learn these features from labeled image data. We demonstrate performance on two real-world image databases and compare it to a classifier and a Markov random field.