UC Santa Cruz

We propose a new technique for the accurate segmenta- tion of text strokes from an image. The algorithm takes in a cropped image containing a word. It first performs a coarse segmentation using a Fully Convolutional Network (FCN). While not accurate, this initial segmentation can usually identify most of the text stroke content even in difficult situ- ations, with uneven lighting and non-uniform background. The segmentation is then refined using a fully connected Conditional Random Field (CRF) with a novel kernel defini- tion that includes stroke width information. In order to train the network, we created a new synthetic data set with 100K text images. Tested against standard benchmarks with pixel- level annotation (ICDAR 2003, ICDAR 2011, and SVT) our algorithm outperforms the state of the art by a noticeable margin.