A Spectral Clustering based Text Image Segmentation Algorithm

Abstract

An innovative novel approach for solving the text segmentation problem in natural scene images is proposed in the research. The new algorithm uses the normalized vectors as the indicator of the spectral clustering. The weighted matrices used in evaluating the graph cuts are based on the gray levels of an image. In the experiment, we found that the proposed algorithm requires much smaller spatial costs and much lower computation complexity. Experiments show the superior performance of the proposed method compared to the traditional algorithms.