Intelligent Adaptive Anisotropic Diffusion Filtered Deep Neural Network With Gaussian Activation For Image Classification

This paper presents a novel adaptive anisotropic diffusion filtered deep neural network (AADF-DNN) model for achieving effective image classification with increase the accuracy and reduces the running time, false-positive ratio. The proposed AADF-DNN model uses deep learning and Gaussian activation function to reduce the false-positive ratio. First, a number of input images are given to the input layer get pre-processed by adaptive anisotropic diffusion filtered reducing the noise. Then, the input layer sends the input images into hidden layers. The hidden layer is used to extract significant features such as shape, color, texture, and size for reducing the running time. Next, the Gaussian activation function is used to classify the images into corresponding classes based on the measurement value between the extracted features and pre-stored features. Finally, the classification results of input images are obtained. Experimental results illustrate that the AADF-DNN model enhances the classification of image performance with higher accuracy at the minimal running time than compared to the PCGRBM.