Cryptography using Artificial Neural Network
Satish Kumar Singh · International Journal for Research in Applied Science and Engineering Technology · 2019
The main purpose of this work is to explore the problem the use of artificial neural network for retransmission large satellite image encoding.Central Accreditation uses fixed, arbitrary keys in the learning process such as classical symmetric and asymmetric coding.The network used is NxMxN neurons, hidden and exit levels.The network is being trained weight regulation, and bias is given a fixed value 0 to 1 after normalization.It is biased is determined.Bidability between the input layer and the hidden layer Layer acts as the first key (K1), whereas bias is partial the hidden layer and the outer layer represent the second key (K2).The course method uses K1, K2, or both, and is made through use small sized images to improve speed.Then the network is used to encode and solve images of ordinary satellites.Many tests prepared various satellite optical and SAR pictures and so on content between decoding (decryption quality) good quality images and decoding were at least 98% images that the network has not previously been trained to decode.The also found that the network does not affect geometry image distortion, such as translation, size and rotation.