This paper presents an automatic image analysis of MR brain classification of stationary wavelet transformation based threshold decomposition and GLCM texture features. The contribution of this work is the Stationary Wavelet Transformation (SWT) based image analysis approach provides better success rate rather than the discrete wavelet approach. This is achieved by the outcomes of decomposed image transform of stationary wavelet that provides clear image details visually for the subsequent steps of image analysis. The performance of this particular SWT based segmentation and classification approach has evaluated against the same sequence process of DWT with use of sensitivity, specificity, and accuracy measures. The validation of these approaches is started with process of pre-processing, wavelet transformation, segmentation, and feature based classification. The preprocessed image is transformed and segmented using the approach of stationary, discrete wavelet transforms and threshold binarization technique individually. From the recognized abnormal image samples, the area of the tumor and a total number of affected cells are computed. It is found that the SW based image segmentation and classification model has superior results rather than the DW-GLCM feature classification techniques.
Volume 12 | Issue 7
Pages: 60-69
DOI: 10.5373/JARDCS/V12I7/20201985