Aplikasi Deteksi Nilai Uang pada Mata Uang Indonesia dengan Metode Feature Matching

Giovinna Khoharja (Program Studi Teknik Informatika)
Liliana Liliana (Program Studi Teknik Informatika)
Anita Nathania Purbowo (Program Studi Teknik Informatika)

Jurnal Infra
Vol 5, No 1 (2017)

Banknote can’t be detected and recognized by blind people.Application for banknote recognition, which can run on Androidsmartphone with camera, is made in this paper to help blind people.Indonesian Rupiah(IDR) is used as a working example. This doesnot require any communication with remote server, and all thenecessary computations take place on the phone itself. In thisapplication, user can tak picture of the banknote with double tapon the screen, and the results will be announced by voice. Thesystem relies on computer vision algorithm, such as featuredetection, feature description, and matching. Each application ofORB, SURF, and SIFT is applied in matching captured banknoteimages to template images. To improve the confidence of thebanknote recognition, homography is used to filter the featurematching results.The systems evaluate the performance on 700 images and report anaccuracy of 93.14% using SURF, 92.57% using SIFT, and 89.17%using ORB