IVR MATLAB CODE

This task does a Bayesian classifier identification of flat parts for the IVR review. An example of the execution is below.

Relevant files are:

adapt.m	does adaptive thresholding
bgremove.m	uses the background intensity to remove illumination gradients
blocks.mat	the stored statistical model file
buildmodel.m	computes the statistical model given the labeled feature vectors
complexmoment.m	calculates the complex central moments
classify.m	finds the best classification
cleanup.m	uses image morphology to remove holes and spurs
conv1d.m	1D convolution
doall.m	top level training and classification process
dohist.m	computes the hsitogram of pixel values
esthomog.m	estimates the homography between 2 planar views
extractprops.m	finds a feature vector for the largest region
findthresh.m	finds the threshold for binarization
getlargest.m	selects the largest of the connected image regions
getproperties.m	gets compactness and moment invariants
liveimagejpg.m	acquires a live captured image
mahalanobis.m	computes the Mahalanobis distance
multivariate.m	computes the probability of a multivariate Gaussian distributed variable
mybwarea.m	computes binary region area w/o using the image proc. toolkit (bwarea)
mybwlabel.m	labels binary image into connected regions w/o using the image proc. toolkit (bwlabel)
mybwperim.m	computes binary region perimeter w/o using the image proc. toolkit (bwperim)
mygausswin.m	creates a gaussian smoothing window w/o using the sig. proc. toolkit (gausswin)
myim2bw.m	thresholds a grey image w/o using the image proc. toolkit (im2bw)
myimdilate.m	does image dilation w/o using the image proc. toolkit (imdilate)
myimerode.m	does image erosion w/o using the image proc. toolkit (imerode)
myjpgload.m	loads a prestored JPG file
myrgb2gray.m	converts a RGB image to grey w/o using the image proc. toolkit (rgb2gray)
testdata1/f1.jpg	Part 1 image 1
testdata1/f10.jpg	Part 1 image 2
testdata1/f11.jpg	Part 1 image 3
testdata1/f12.jpg	Part 1 image 4
testdata1/f2.jpg	Part 2 image 1
testdata1/f3.jpg	Part 2 image 2
testdata1/f4.jpg	Part 2 image 3
testdata1/f5.jpg	Part 2 image 4
testdata1/f6.jpg	Part 3 image 1
testdata1/f7.jpg	Part 3 image 2
testdata1/f8.jpg	Part 3 image 3
testdata1/f9.jpg	Part 3 image 4
partgradgr.jpg	test image
remap.m	projective transfer example
hzinput.jpg	projective transfer test image
lightgradgr.jpg	test image

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

        
doall(2)
Model file name (filename)
?blocks
Number of classes (int)
?3
Want to acquire training data (0,1)
?1
Training image file stem (filestem)
?testdata1/f
Use live training data (0,1)
?0
Number of training images (int)
?12
Train image 1 true class (1..3)
?1
Train image 2 true class (1..3)
?1
Train image 3 true class (1..3)
?1
Train image 4 true class (1..3)
?1
Train image 5 true class (1..3)
?2
Train image 6 true class (1..3)
?2
Train image 7 true class (1..3)
?2
Train image 8 true class (1..3)
?2
Train image 9 true class (1..3)
?3
Train image 10 true class (1..3)
?3
Train image 11 true class (1..3)
?3
Train image 12 true class (1..3)
?3

Want to use live test data (0,1)
?0
Test image file stem (filestem)
?testdata1/f
true class (1..3)
?1
evaluations =    61.2911         0         0
class = 1

Want to process another image 2 (0,1)
?1
true class (1..3)
?1
evaluations =    61.2911         0         0
class = 1

Want to process another image 3 (0,1)
?1
true class (1..3)
?1
evaluations =    61.2911         0         0
class =      1

Want to process another image 4 (0,1)
?1
true class (1..3)
?1
evaluations =    61.2911         0         0
class =      1
           
Want to process another image 5 (0,1)
?1
true class (1..3)
?2
evaluations =    
class =            2

Want to process another image 6 (0,1)
?1
true class (1..3)
?2
evaluations =    0    1.7356         0
class =      2

Want to process another image 7 (0,1)
?1
true class (1..3)
?2
evaluations =    0    1.7356         0
class =      2


Want to process another image 8 (0,1)
?1
true class (1..3)
?2
evaluations =    0    1.7356         0
class =      2

Want to process another image 9 (0,1)
?1
true class (1..3)
?3
evaluations =        0.0000     0.0000   37.2140
class =             3

Want to process another image 10 (0,1)
?1
true class (1..3)
?3
evaluations =    0.0000     0.0000   37.2140
class =      3

Want to process another image 11 (0,1)
?1
true class (1..3)
?3
evaluations =   0.0000     0.0000   37.2140
class =      3

Want to process another image 12 (0,1)
?1
true class (1..3)
?3
evaluations =   0.0000     0.0000   37.2140
class =      3

Want to process another image 13 (0,1)
?0
ans =
scatter matrix (rows are true, cols are classified)

Scatter =

     4     0     0
     0     4     0
     0     0     4

>>

Back to IVR index page

This page is maintained by the course lecturer, Bob Fisher, r.b.fisher@ed.ac.uk, room 2107D JCMB, ext 513441

Home : Teaching : Courses : Ivr