Automated Postal Sorting System Using Optical Character Recognition and Image Processing

The growing demand for automated postal sorting has increased due to E-commerce and global trade. Traditional manual and barcode-based methods struggle with rising parcel volumes, while Optical Character Recognition (OCR) and image processing offer a label-free alternative. This paper presents an OCR-based postal sorting system that extracts and classifies postal codes from envelope images. The system integrates a high-resolution camera, OCR processing unit, and real-time GUI, employing image acquisition, preprocessing, postal code detection, OCR extraction, and classification. Tests on nine postal codes achieved over $96 \%$ accuracy, with detection times of $0.61-0.84$ seconds. Errors primarily occurred in thin-stroke fonts under 3 mm, such as TH SarabunPSK and Angsana New, whereas larger fonts improved recognition. The system enhances sorting efficiency and reduces manual workload, proving its feasibility in regional logistics centers. Future work will explore deep learning-based OCR enhancements, multilingual postal code support, and integration with hardware-based automation.