method: HUST_MCLAB_VER2.02015-04-02
Authors: Cong Yao, Zheng Zhang, Chengquan Zhang,Xiang Bai,Baoguang Shi
Description: We combine with the text line proposals which are generated by MSER, SWT and Symmetry-based method, then we adopt two layer filter(character level and text line level) to divide text line proposals into words and remove the false positive.