This project is based on the paper Multi-orientation scene text detection with adaptive clustering from PAMI by Xu-Cheng et al.

This is an implementation of a character detection part of the text detection algorithm for natural scenes where the texts are arbitrary put in any orientation.

This implementation includes the writing the MSER algorithm, which is the crux of the character detection, from a component tree. To generate this tree, a quasi-linear time algorithm based on the paper Quasi-linear algorithm for the component tree by L. Najman and M. Couprie

An example run: