Verification of the Document Components from Dual Extraction of MRTD Information

Abstract

This paper proposes a method for character region extraction and picture separation in a passport by adopting a preprocessing phase for the passport recognition system. Character regions required for the recognition make black pixel and remainder of the passport regions makes white pixel in the detected character spaces. This method uses the MRZ sub-region in order to automatically calculate the threshold value of the binary image which is also applied to the other character regions. This method also applies horizontal/ vertical histogram projection in order to remove the picture region from the binary image. After region detection of picture area, the image part of the passport is stored in the database of face images. The remainder of the passport is composed of character part. Recognition on the extracted character is performed on the various different passports. From the obtained information, auto-correlation of extracted characters within a given passport are accomplished after character recognition. A cross-check process of MRZ information and field information of passport on similarity is implemented. For this purpose, this paper uses the auto-correlation between the binarization method based on the color information and an extracted image from a passport to propose a characteristic extraction method to prevent passport forgery.