METR4202 2013 Computer Vision Files

METR4202 Lab 2 "Coin finding, and counting"

property of Mitchell Fullelove and Kianoosh Soltani Naveh, 2013


############################PACKAGE INFO#############################################

Capture Image
The package uses Kinext Mex for capturing images. To use these functions sample_nImage_mod.m and
sample_nImage.m copy them to KinectMex Toolbox folder. Copy the content of Config folder to Config
folder in KinectMex toolbox. 

Copy all of the following folders into KinectMex Toolbox

Color Calibration
The package contains the functions that detect color calibration chart in the scene. The code builds
on the CCFind package from  K. Hirakawa, "Color Checker Finder," accessed from 
   http://campus.udayton.edu/~ISSL/software.
Note that a bug associated with the Hirakawa's work was identified and has been fixed in 
this package. 

Camera Calibration
Camera caibration uses a modification of the RADOCC Toolbox which automates the calibration procedure.
"click_clib.m" and "data_calib.m" have to copied in RADOCC Toolbox.

Coin Detection and Localization
This is a package of matlab functions used to detect Australian coins in an image taken by kinect.
It uses the depth map from kinect along with coin size and color properties to find, classify and
locate the coins.
The program also tries to detect the notes in the scene and mention their presence.

Note Detection
This is a set of functions used to identify Australian notes in the images. These functions require
vl_feat toolbox.

There are sample images in this package that can be used to test the package.

###################################Instructions#########################################
1- First to run color calibration use sample_nImage to capture an image. Then use tellMeTheColor.m
to detect color calibration chart and the information about cell 12 and 21. 

2- To carryout camera calibration, use sample_nImage_mode and use a calibration chart to get camera
parameters.

3- Then take an image from  the scene again using sample_nImage and use inspect_scene.m. This function
detects coins in the scene, evaluates their value, localizes them in the scene and returns a list
of all coins and their position.