Soft computing-based edge-enhanced dominant peak and discrete Tchebichef extraction for image segmentation and classification using DCML-IC

Texture analysis is a very predominant scope in the area of computer vision and associated fields. In this work, edge-enhanced dominant valley and discrete Tchebichef (EDV-DT) method is presented to eradicate noise and segment image into number of partitions with higher accuracy and lesser time. In EDV-DT method, an edge-enhancing anisotropic diffusion filtering technique is used to perform the preprocessing for MRI, CT and texture features. The adaptive anisotropic diffusion creates scale space and reduces the image noise without removing the texture image content (i.e., edges, lines) that is found to be essential for texture image segmentation. Followed by preprocessing, histogram dominant peak valley segmentation technique is applied to segment the localization of region of interest. Valleys in histogram for the preprocessed images help in segmenting the texture image into equal-sized texture regions. Finally, with the segmented images, discrete Tchebichef moment feature extraction is applied to extract relevant features from the segmented texture image for reducing the dimensionality. This in turn helps in improving the feature extraction rate. Further a deep convolution multinomial logarithmic-based image classification (DCML-IC) model is presented for predicting results via positive and negative fact classification. The proposed system provides the better prediction of accuracy and the prediction of time to compare the other existing methods.