Brain tumor detection and classification is the most difficult and tedious task in the area of medicinal image preparing. MRI (Magnetic Resonance Imaging) is a medicinal procedure, generally adopted by the radiologist for representation of inner structure of the human body with no surgery. MRI gives abundant data about the human delicate tissue, which helps in the conclusion of brain tumor. Precise segmentation of MRI image is basic for the conclusion of brain tumor by computer supported clinical device. This paper is focused towards the design of an optimal and more accurate way for the detection of tumor from brain MRI scans and if it confirms the presence of tumor then it is focused on evaluating its stage, ie, benign or malignant. We have experimentally shown that our proposed methodology has a greater accuracy than other existent methods for classifying tumor type to be either as Malignant or Benign since the maximum accuracy for detection of malignant tumor is 99.02% and for Benign tumor is 99.67%.