The traffic of identification document images through digital media is already a common practice in several countries. A large amount of information can be extracted from these images through computer vision and image processing techniques, as well Machine Learning models. This information can generate data for different purposes such as extraction of text fields and dates for use in OCR systems and handwritten signatures to be used as biometrics, in addition to other characteristics and patterns present in images of identification documents.
Researching different applications in the field, we observed that few works were developed in the treatment of images of identification documents, specially related with image segmentation of ID documents. The importance of this challenge is to stimulate researchers and scientists in the search for new techniques and scientific productions for the treatment of these document images based on image segmentation. The availability of a database is another factor of great importance, given there are few image databases of ID documents available free of charge to the community . As usual, due to privacy constraints all personal text information in the documents was synthesized with fake data. The original signatures were replaced and new ones collected randomly from different sources were stamped in the right places of ID documents. All these changes were performed by well-designed algorithms and post-processed by humans whenever needed to keep the real-world conditions as much as possible.