Money Reader Application - Zhuorui Yang & Lili Wan

Abstract : In this project, the original name for it is money reader, but we would like to name it currency identifier for professional consideration. As we known, visual‐impaired people will face a lot of trouble in real life, especially for item recognition, in this project we plan to solve one of the most important problems, that is, let them identify denomination of bills to make their life easier and more convenient by using Android Smartphone. We try to build an app that can instantly recognize currency and speak the demonization by pointing device's camera at a bill and hearing the description of the denomination in real time by the application. We will try to make it recognize as instant as possible to obtain the real time performance without using the Internet by capturing the picture of the object and run the image processing on the Android phone.

We implemented such a system on the Android Smart phone, the bill captured by camera was first pass through a SURF (Speeded Up Robust Features) detector and implemented SURF algorithm for bills matching [2] in the real time in the Android phone, the bill used for matching are stored in internal storage and Text To Speech API is used to speak the denomination of the bill. The benefit of a working system on an Android Smart phone image‐capturing device is that real‐time results can be obtained, allowing for quick feedback for the user, which can be important in helping them to recognize currency. Moreover, the SURF algorithm which is written in C/C++ can match the image in a short time.