Using the Mumford-Shah segmentation for Text Detection in images

From LRDE

The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

Abstract

We will propose a method that has the objective of improving text detection in natural images. First of allwe will present the Mumford-Shah segmentation methodbefore explaining how to integrate this method into the laboratory's current toolchain (TextCatcher). Finally, we will evaluate the new text localisation error rate given by this segmentation.