Face recognition using cloud Hopfield neural network

An approach for the recognition of low resolution grey scale facial images using cloud Hopfield neural network (CHNN) is presented. This approach consists of three steps: first we transform the grey scale facial images into binary facial images using Otsu's method, second Hebb rule is employed to store binary faces in the weight matrix of the network and finally correct face is retrieved from distorted face using CHNN retrieval algorithm. CHNN consists of clouds each with r number of unique neurons, these clouds are updated asynchronously unlike conventional asynchronous retrieval in which only one neuron is updated at a time. We compare our results with that of conventional HNN and other face recognition methods. Our results show that even when the distortion in the presented face is 45%, the CHNN is able to give at least 82.8% successful retrieval as compared to only 63% by conventional HNN for the same amount of distortion. This is much greater than previous reported claims to the best of our knowledge.