Text extraction from scene images through local binary pattern and business features based color image segmentation

Abstract

This article proposes a scheme for automatic extraction of text from scene images. First, we apply a color image segmentation algorithm to a scene image. To improve the color image segmentation performance, we incorporate local binary pattern (LBP) and business features within it. Local Binary Pattern (LBP) operator is a texture descriptor for grayscale images. On the other hand, the business feature describes the variation in intensity. The segmentation procedure separates out certain homogenous connected components from the image. We next inspect these connected components in order to identify possible text components. Here, we define a number of shape based features that distinguish between text and non-text connected components. Our experiments are based on the ICDAR 2011 Born Digital data set. The experimental results are satisfactory.