Master thesis "Exploitation de l'intelligence artificiell dans le control non destructif par ultrasons"
The folder data contains 180 images of ultrasonic scans each images is 128x128 px size the images are extracted from a comparative articles between diffrent classication methods here we use the neuro fuzzy approach to establish a classification of the images between 2 sets and model a fis structure to be able to predict if a an ultrasonic scan corresponds to a sample that has a flaw in it or not the data set contains 90 images with a flaw and 90 images without flaw
- images with name 0_image_x.jpg have flaws
- images with name 1_image_x.jpg don't have flaws the code classicaiton .m is not completed most of the experimental work was done using the neurofuzzy designer as the data was being explored and final parameters were not yet found
hog and lbp codes are for features extraction from the images and hog cell size 4 to 64 files contain the extarcted features with different cell siz also there is a fis structure that was created using the full data set no testing set