feature-extraction-technique-based-character-recognition-using-ann-for-Sinhala-Characters

Used technologies

run the create_target_metrics.m, this will create the y matrics. one character have 25 different sampels.so 25 * 34 = 850 samples in the training_set folder. you can change this according to your program.
run the myFeatureScript.m, this will extract the feature and store in the x matrics.
then you need to train the neural network using train.m.
give the image path in image = imread('image path'); and run the extract_word_from_image.m

output will be printed to output.txt.

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
test_images		test_images
training_set		training_set
About.fig		About.fig
About.m		About.m
Home.fig		Home.fig
Home.m		Home.m
README.md		README.md
create_target_metrics.m		create_target_metrics.m
crop_image.m		crop_image.m
extract_word_from_image.m		extract_word_from_image.m
feature_extract.m		feature_extract.m
featureout.mat		featureout.mat
input_target.fig		input_target.fig
input_target.m		input_target.m
line_classifier.m		line_classifier.m
line_separator.m		line_separator.m
myFeatureScript.m		myFeatureScript.m
net.mat		net.mat
output.txt		output.txt
test.m		test.m
train.m		train.m
training_interface.fig		training_interface.fig
training_interface.m		training_interface.m
x.mat		x.mat
y.mat		y.mat