Development of an Offline Hausa Handwritten Word Recognition System Using Support Vector Machine

ABSTRACT
Handwritten word recognition is a critical component in the advancement of optical character
recognition (OCR) systems, particularly for languages with unique scripts such as Hausa. This
project focuses on developing an offline Hausa handwritten word recognition system utilizing
Support Vector Machine (SVM) classifiers. The proposed system comprises several stages:
preprocessing, feature extraction, and classification. Initially, preprocessing techniques are
applied to enhance the quality of the handwritten input, addressing issues like noise and varying
stroke widths. Subsequently, robust feature extraction methods are employed to capture the
distinctive characteristics of Hausa script. Finally, SVM classifiers are trained to accurately
recognize and categorize handwritten words. The efficacy of the system is evaluated using a
dataset of handwritten Hausa words, demonstrating promising recognition rates and
highlighting the potential of SVMs in handling the complexities of Hausa handwriting. The
system shows 97% recognition rate and 3% rejection rate for the tested Hausa characters, and
16.7% recognition rate and 83.3% rejection rate for the tested Hausa words. This research
contributes to the broader field of OCR by providing insights into the application of machine
learning techniques for underrepresented languages.