BDM Capstone Project team : HungPD - Supervisor ThanhLN – Leader ManhDC BienVT NinhVH
Introduction Requirements Implementation Results & Conclusions
Idea’s Origin Existing Methods Objective System Members’ Roles & Responsibilities Project Plan
Nowadays, many vending machines are used widely in life but it seems that they are not suitable with the habit of buying in Vietnam. In the real world, this module will be used widely in automation vending machines.
Module is used as personal device Module is used in Banknote counter
ATM Machine WithdrawTransactionDeposit Vending Machine Used with Coin Used with Banknote
Banknote Counter Counts Banknote Banknote Classify Need Banknote Detecting Module ATM Banknote Counter Vending machine
Existing Products
Existing Product
Module is suitable for a vending machine with its small size, low weight, accuracy, and flexibility
ThanhLN – Leader + QA + dev Neural Network Detect object ManhDC – PTL + dev + test Linux GUI development Image process BienVT – Hardware Implementer + test Build ARM9 embedded Linux NinhVH – Dev + test Image process Neural Network
Project Initialization
Project Planning
Project Implementation
Project Release and closing
Functional Requirements Non-functional Requirements Hardware Requirements
Turn on/turn off the module Recognize the input which customers put in is Banknote or not Banknote Recognize type of the Banknote such as: VND, VND etc Recognize quality of the Banknote Display information on the display screen
Response time User interface Maintainability Reliability Low cost
Small size and low weight Touch screen Camera : Minimum configuration : 1.3 Mp Zoom 10x Embedded machine : Minimum configuration : Chip speed : Above 300Mhz RAM : above 64Mb, 32 bit data bus Have touch screen Have a USB gate to connect with Camera
Analysis & Selection of Devices, Tools Design Implemented Technical Problems Testing
Software: Platform: QT SDK IDE : QT creator Document : Microsoft office 2007 Design : Microsoft Visio 2007 OS : Windows, Ubuntu Language : C, C++
Hardware : Camera : Colorvis CVC ND40. Embedded machine : mini 2440 ARM9.
Banknote scanning and collecting
NN data
Slab and mask process
Algorithm uses axis-symmetry mask Program has 50 different masks to calculate slab values. Slab value is calculated by sum of non- masked pixels
Neural Network
Project uses Back propagation NN
Implement completed program from Linux to Embedded computer
Learn how to control a webcam in linux Learn how to use Neural Network Configure Embedded System to use in this module
Hardware ARM9 machine Webcam Graphic User Interfaces Button Output view System Process Banknote detect Non-banknote detect
Test result :
Results Risks and Limitations Conclusions
Over the world, especially in embedded system, the perfect does not exist. Every system has trouble. Around the system as a capstone project, we try best to create and develop an acceptable and stable system. The system cannot quick, easy to use cannot good, but the stable of system can be believable Our module can change function or requirement easily. With a full database and a developer of team, we can add USD or EURO to detect. This is increase value of the module. Module can use in international transaction, and in vending machine, in every where
Risk Hardware may not stable Aspect of environment Camera is not very good Limitation Not include fake money detect
This product is our last project in FPT University as students. So we want to send our appreciation to : Teachers, who gave us knowledge through 4 years in this university. Specially to Dr HungPD, who guide us to this success Thanks to all staff and students, who went with us from 1 st year
Thank you for listening