An Efficient Method for Stamps Verification Using Haar Wavelet Sub-bands with Histogram and Moment

摘要

Stamps have been used as visible certificates for documents since long time ago. At present, they are used for certifying official books, memos, agreements those government institutions issued in order to ensure that their contents are not changed. The stamp image suffers from various problems such as (different sizes, different shapes, irregular shapes, ink color difference, geometric deformations, and non-uniform background); which in turn make the recognition and verification tasks more difficult and challenging. In this paper, an efficient stamp recognition and verification system are presented. This paper proposed a method to verify stamps, by using Haar wavelet transform with histogram and moment. The devised method has four stages: (i) stamp extraction; (ii) preprocessing the image; (iii) feature extraction; and (iv) matching. Stamp extraction is done to isolate complex background and removing unwanted data or noise that surrounding the stamp area. The preprocessing stage is necessary to improve the stamp brightness and, also to eliminate the rotation that occurs during the stamping process. In feature extraction, the extracted information will be representing the desirable features vector in order to discriminate between stamps. Finally, each extracted feature vector will be saved in the dedicated system database for matching purpose. The performance of the proposed stamp recognition and verification system is affected by different factors which are the number of blocks, and overlapping ratio. The test results conducted on the public stamps dataset and indicate that the proposed system provides a high recognition rate (i.e., 93.35%), while the verification tests indicated the achieved minimum equal error rate (EER) is 0.106%, and accuracy is 99.98%.