Text lines and words segmentation
I developed under the supervision of Beatriz Marcotegui a software tool for text-lines and words segmentation in handwriting images. This software has been submitted to the ICDAR 2009 Handwriting Segmentation Contest.A segmentation example is shown in the images below :
![]() |
![]() |
![]() |
The proposed method achieves a performance rate of 98.6% in text-lines segmentation and 84.1% in words segmentation in the training dataset (100 images).
In the evaluation dataset (200 images), the obtained performance was 98.3% for text-lines segmentation and 88.8% for words segmentation.
More details about the competition can be found in this ICDAR 2009 article.


