Improved Technique in Arabic Handwriting Recognition

Arabic handwriting recognition has significant applications in fields like postal sorting, handwritten text identification, and cheque processing. The process involves several steps: preprocessing, feature extraction, and classification. Preprocessing enhances image quality through noise removal, normalisation, and binarisation, which are essential for accurate segmentation. Feature extraction captures key information such as stroke direction and spatial relationships, which are crucial for distinguishing between different characters. Hybrid methods, statistical features, and structural features are typical feature extraction strategies. Next, classification methods such as K-nearest neighbour and Support Vector Machines are employed to categorise the extracted features into predefined classes. The effectiveness of Arabic handwriting recognition systems depends heavily on the quality of feature extraction, which directly impacts recognition accuracy. Researchers have explored various techniques, including structural and statistical feature extraction, to optimise these systems. Exceptional accuracy rates are achieved through the utilisation of the proposed SVM linear kernel and KNN classifier with 99.64% and 97%, respectively.