Localization of scores and average in Algerian baccalaureate transcripts

This work is part of archiving and electronic document management. We are interested here in the extraction of scores and average from the transcripts of Algerian baccalaureate. We started this work by a physical study of the baccalaureate transcripts in order to know their structure. The proposed system extracts the student's scores and average based on the localization of tables containing this information. This localization is reached after several steps of preprocessing aiming to improve the quality of the transcript, and segmentation to extract the connected components of the image and eliminate useless information. After that, the localized tables are segmented into columns and the columns into rows in order to extract the desired information. The proposed system has been applied to 43 baccalaureate transcript images of our test corpus, and the obtained results are encouraging.