Image Labeling by Integration of Local Co-Occurrence Histogram and Global Features

Abstract

In this paper, we perform image labeling based on the probabilistic integration of local and global features. Several conventional methods label pixels or regions using features extracted from local regions and local contextual relationships between neighboring regions. However, labeling results tend to depend on local viewpoints. To overcome this problem, we propose an image labeling method that utilizes both local and global features. We compute the posterior probability distributions of the local and global features independently, and they are integrated by the product. To compute the probability of the global region (entire image), Bag-of-Words is used. In contrast, local co-occurrence between color and texture features is used to compute local probability. In the experiments, we use the MSRC21 dataset. The result demonstrates that the use of global viewpoint significantly improves labeling accuracy.