Loay & Maha Stamps Dataset

Published: 26 February 2020| Version 3 | DOI: 10.17632/ktr99fc826.3
Contributors:
Maha Abd,

Description

Recently, with the continuous evolution of new technology the world has faced unusual growth of multimedia data especially in document images. Huge quantities of documents are created and used in many important official organizations. The stamps assure the authenticity of the documents contents. The general objective of this dataset is enabling researchers in the field of pattern recognition to analyze, detect, localize and recognize different types of stamps. The size of the dataset is 48.9 MB. It contains a total of 1557 color stamps samples which are stored as BMP image format. The images of stamps were acquired from documents using a scanner and mobile camera. The cropping process was applied to capture only the stamp area of the whole document, using three applications which are paint, snip and sketch software. For each cropped stamp, eight stamp images are produced from rotating the original one at different angles clockwise. Rotation starts from 5 degrees and increases by 5 degrees until it reaches to 40 degrees. Note the orientation of the original stamp image is with the horizon line, which means that it is at an angle of zero. The stamps have different size, shape, complexity, position, directions and colors. For example, there are six categories of stamp shapes arranged as circle, oval, square, rectangular, triangular and some other irregular shapes. Its worth to mention that the collected dataset may be degraded in quality and resolution and the stamp can be located on a relatively complex background. The dataset was collected from various sources, including official documents from Iraqi educational institutions, documents from syndicates such as teachers, engineers, doctors, pharmacists, Iraqi hospitals, provincial councils, and most Iraqi institutions. The dataset also contains some stamps from other Arab countries.

Files

Institutions

University of Babylon

Categories

Clustering, Pattern Recognition Clustering Process, Image Color Processing

Licence