DEEP LEARNING TECHNIQUES FOR EXOTICISM MINING FROM VISUAL CONTENT BASED IMAGE RETRIEVAL

Abstract

Early diagnosis has been aided by electronic restorative imaging and examination approaches employing several modalities.
The advancement of computer-aided retrieval systems in recent years has made them a nondestructive and popular tool for
disease identification in medical photographs. In this paper, an adaptable Gabor wavelet filter bank and a feature descriptor
based on Texton are created for medical picture retrieval. The suggested descriptor basis's architecture allows for flexibility in
extracting the dominating directional characteristics from medical pictures. In addition, we provide an unique end-to-end
integrated deep learning model that employs the Convolutional Neural Network (CNN) and the Long Short-Term Memory
cell (LSTM). Using datasets such as New Caltech, Corel-1000, Oliva, and Corel-10,000, the proposed integrate deep learning
descriptor is compared to existing descriptors such as CCM, CHD, MTH, and MSD.