Document Image Segmentation into Text, Continuous-Tone and Screened-Halftone Region by the Neural Networks

In this paper we propose the image segmentation incthod by the neural networks (NN) to extract text, continuous-tone and screened-halftone region in the document image. Each feature extractor for text, continuons-tone and screened-halftone is composed of the 3-layer NN which is upper part of 5-layer lio~irglass model NN; each hourglass model NN is trained by different learning samples which specify its own characteristics. We call these extractors as the Feature Neural Networks (FNNs). The classification stage is composed of another 3layer NN whose inputs are applied from feature extractors. We call this as the Classification Neural Network (CNN). Experimental results show this method successful to segment text, continuous-tone and screenedhalftone region.