Projects

Automated Detection of Skin Diseases using Texture Features

In this proposed approach automated detection of Skin Diseases using GLCM texture
features have been proposed, efforts have been made to detect three categories of skin
diseases (Dermatitis, Eczema & Urticaria) using GLCM features with high recognition
accuracy. Texture refers to visual patterns or spatial arrangement of pixels that
regional intensity or color alone cannot sufficiently describe. The main task of the
current work is to identify features which can distinctly diagnose different skin
diseases.
Here, two new GLCM features are applied on the skin disease dataset images. The
dataset is divided into training & testing dataset. At first, the input image is converted
to grayscale to retrieve texture features. Then the gray-level values are converted to
double precision value. Then feature values are calculated along 4 GLCM directions
0°, 45°, 90° & 135° directions. These values are then averaged to get final feature
vectors. These feature values are then used to train three input layer artificial neural
network. After that, the system is tested with test dataset feature values.
Experimental results show that the proposed system gives more than 96% accuracy.



Tags :
3.5
Your rating: None Average: 3.5 (4 votes)