An Intelligent Vision System for Product Label and Banknote Recognition for Visually Impaired Assistance

บทคัดย่อ

Visual impairment significantly limits individuals’ ability to independently perform daily tasks such as identifying banknotes and reading product labels, leading to potential errors in financial transactions and reduced accessibility to essential information. This study proposes an intelligent vision-based assistive system that integrates barcode detection with ORB (Oriented FAST and Rotated BRIEF) feature matching to enable accurate and real-time object recognition on mobile devices. The system captures images via a smartphone camera and provides audio feedback through a text-to-speech mechanism to support visually impaired users. Experimental results demonstrate that the proposed system achieves an overall recognition accuracy of 96.2%, with 94.6% for product labels and 97.8% for banknotes, along with an AUC of 0.97 and an average processing time of approximately 1.80 seconds. The system also exhibits balanced classification performance with high precision, recall, and F1-score values. A user evaluation involving 30 visually impaired participants confirms that the system improves task performance and usability, particularly with minimal training. These findings suggest that the proposed hybrid approach offers a robust, efficient, and practical solution for real-world assistive applications, thereby contributing to the development of intelligent technologies that enhance accessibility, support independent living, and align with the vision of a smart ageing society.