Malaysian Banknote Reader For Visually Impaired Person

Abstract

Visually impaired person facing difficulty in shopping, running business and activities that dealing with handling banknote. This paper presents a Malaysian banknote reader targeted for visually impaired person to allow them identifying Malaysian banknotes value and dealing with dealers that performing business transaction in Ringgit. This paper proposes a Malaysian banknote reader for visually impaired person using perceptual hashing based image searching and fuzzy logic algorithm. the banknote image captured by the reader will be sent to cloud/internet server for image processing, via Wi-Fi to eliminate the bulky size reader problem. A server in cloud is set up to receive the captured image and return the analysis results after running the image processing algorithm. This will eventually reduce the costs and size of the banknote reader. The banknote reader reached about 95% accuracy by testing 600 different worn, torn and new Malaysian banknotes. The proposed reader managed to recognize banknotes at about 480ms to 560ms for single-sided recognition, after image taken by the reader’s camera. The test of the reader was also run with both visually impaired and visually good participants to recognize 100 pieces of different Malaysian banknote. The result shown that the recognition responses is comparable, only with slight difference of 1.0908 second per banknote reading time.