earticle

영어

Automatic handwritten digits recognition is useful in a large variety of applications such as cheque verification and mail sorting. However, the selection of the technique for feature extraction remains the big challenge step for achieving high recognition accuracy. This paper presents a technique based on DWT and DCT to capture the discriminative features of handwritten digits. DCT coefficients are extracted from low-frequency sub-band (LL) of DWT image. These coefficients are fed into the ANN in the classification stage. This work has been tested with ADBase database containing 70,000 digits images, and a comparison made against some existing techniques, and promising results have been obtained.