Brain tumour detection is an evergreen topic to attract attention in the examination field of Information Technology innovation with biomedical designing, in view of the gigantic need of proficient and viable strategy for assessment of enormous measure of information. Image segmentation is considered as one of the most vital systems for visualizing tissues in an individual. To robotize image segmentation, we have proposed a calculation to get global optimal thresholding esteem for a specific brain MRI image, utilizing OTSU+Sauvola binarization strategy. The fundamental reason for feature collection is to diminish the quantity of structures utilized in classification while keeping up satisfactory classification exactness. One of the most extra-customary procedures applied for feature extraction is Discrete Wavelet Transform (DWT). Adequately it anticipates the estimation space on a plane to such an extent that the fluctuation of the information is ideally protected. We propose a justifiable model for brain tumours discovery and classification i.e., to classify whether the tumour is benign or malignant, utilizing SVM classification. SVM utilized here deals with basic hazard minimization to group the images for the tumour extraction, and a Graphical User Interface is created for the tumour classification operation, using the MATLAB platform.