TY - GEN
T1 - Recognition of Arabic (Indian) check digits using Spatial Gabor filters
AU - Mahmoud, Sabri A.
PY - 2009
Y1 - 2009
N2 - Arabic (Indian) check digits recognition is useful in a variety of banking applications. The same technique is applicable to postal zip code reading, and handwritten forms processing. In this paper we present a technique for the automatic recognition of Arabic (Indian) check digits using Spatial Gabor filters. A database consisting of 7390 samples of Arabic (Indian) digits for training and 3035 samples for testing extracted from real bank checks is used. The data was tested with and without preprocessing. When preprocessing is used the digits are normalized to a height of 64 pixels and maintaining the aspect ratio. Spatial Gabor filters with several scales and orientations are used to extract the Spatial Gabor-based features. Average recognition rates of 97.99%, 97.37%, and 92.76% using 1-Nearest Neighbor, 3-Nearest Neighbor, and Nearest Mean classifiers, respectively are obtained. These results are comparable and better than published work using the same database. The experimental results indicate the effectiveness of the Spatial Gabor filters for practical Arabic (Indian) bank checks digits' recognition.
AB - Arabic (Indian) check digits recognition is useful in a variety of banking applications. The same technique is applicable to postal zip code reading, and handwritten forms processing. In this paper we present a technique for the automatic recognition of Arabic (Indian) check digits using Spatial Gabor filters. A database consisting of 7390 samples of Arabic (Indian) digits for training and 3035 samples for testing extracted from real bank checks is used. The data was tested with and without preprocessing. When preprocessing is used the digits are normalized to a height of 64 pixels and maintaining the aspect ratio. Spatial Gabor filters with several scales and orientations are used to extract the Spatial Gabor-based features. Average recognition rates of 97.99%, 97.37%, and 92.76% using 1-Nearest Neighbor, 3-Nearest Neighbor, and Nearest Mean classifiers, respectively are obtained. These results are comparable and better than published work using the same database. The experimental results indicate the effectiveness of the Spatial Gabor filters for practical Arabic (Indian) bank checks digits' recognition.
KW - Arabic (Indian) digits
KW - Feature extraction
KW - Recognition of Arabic bank check digits
KW - Spatial Gabor filters
UR - https://www.scopus.com/pages/publications/79953899962
U2 - 10.1109/IEEEGCC.2009.5734258
DO - 10.1109/IEEEGCC.2009.5734258
M3 - Conference contribution
AN - SCOPUS:79953899962
SN - 9781424438853
T3 - 2009 5th IEEE GCC Conference and Exhibition, GCC 2009
BT - 2009 5th IEEE GCC Conference and Exhibition, GCC 2009
ER -