README

MBED-based labyrinth solver project created for the ACE (Arm Campus Engineering) challenge at the University of Bristol.

License:
========
Everything under GPLv3 unless otherwise specified.

Exceptions:
-----------
-openCV-GUI/QtCvExample.orig/ : from http://www.qtcentre.org/threads/43056-Qt-OpenCV-simple-example?s=c397b1b1ece971320ed7936d428b4a32 , cf openCV-GUI/QtCvExample.orig/README.txt
-serial/ : from http://www.teuniz.net/RS-232/, GPLv2

Quick guide:
============
Qt-independent things:
----------------------
-serial: serial port communication for GNU/Linux
-test0: openCV DisplayImage example
-test1: openCV Canny filter example
-test2a: maze solver 
-test2b: printMaze function
-test3: Contains quite a few useful functions.
-test4: openCV how_to_scan_images example

Qt-openCV GUIs:
---------------
-openCV-GUI/openCV-GUI: main GUI, transforming a picture into a series of numbers representing the maze and allowing sending that string to stdout
-openCV-GUI/QtCvExample.orig: Basic Qt-OpenCV example with Canny edge-detect filter implementation on a streaming video feed or video file.
-openCV-GUI/QtCvExample: Hacked-up version of openCV-GUI/QtCvExample.orig
-openCV-GUI/openCV-integration: Qt-openCV compile test + QVector<QVector<QString> > test

Usage of openCV-GUI:
====================
General usage:
--------------
-"File->Open" or "Open" button to load picture. Should create a corresponding matrix of 0,1,2,3 values in the textbox.
-Click send data to send whatever is in the textbox to stdout

To send the data to the MBED:
-----------------------------
-open the /dev/ttyACM0 port with cutecom or similar, set the baudrate to 115200
-run openCV-GUI with "./openCV-GUI >/dev/ttyACM0"
-uncheck "line endings" before opening a picture
-"File->Open" or "Open" button to load picture. Should create a corresponding matrix of 0,1,2,3 values in the textbox.
-Click send data

Notes:
------
-The nRows/nCols spinboxes allow you to define how many rows and columns should be in the generated matrix.
-The other "Load image" button was/is meant for a dynamically updating processed picture, i.e. should make it easier to test out other processing algorithms.
But there seem to be some problems with the pictures displayed that way, especially for pictures with just a few pixels. Probably an error in the conversion from OpenCV::Mat to OpenCV::IplImage.

Compiling:
==========
For cmake based projects: cmake . && make
For qmake based projects: qmake && make

Dependencies:
=============
-openCV
-Qt4
Note: You can use QtCreator for easy GUI development.

TODO:
=====
-Add MBED code for the maze solver and servo controls.
-Finish maze recognition.