Texture Analysis for MRI Brain Images Using Automatic Segmentation Algorithm

摘要

The regular scales of the image represent the structure of the texture. Texture can be determined from the feature extraction. The raw image is represented in its reduced form by applying feature extraction. The MRI brain images are classified by an accurate and robust classifier which gives an easy way to identify the variations in the images. This paper proposes the method of feature extraction, selection and classification of MRI brain images. The various levels such as textures, structural features classify the tumor as normal and abnormal. PCA (Principle Component Analysis) and LDA (Linear Discriminant Analysis) are applied to the images. The SVM (Support Vector Machine) classifier compares linear and non linear techniques. The MRI brain images of Astrocytoma, Glioblastoma, Lymphoma, Meningioma are detected by applying the proposed technique with high classification accuracy.