Smart Pixels: Harnessing Deep Learning and Fibonacci Decomposition for Image Ciphering

Abstract

Due to the rapid and continuous growth of images in the digital environment, There are still concerns about the security and confidentiality of visual data. In this study, we will present a new, developed approach to securing digital images and maintaining their security by taking advantage of deep learning technology and Fibonacci decomposition. We used a deep neural network to generate random numbers, predict the optimal pixel position in the encrypted image, and apply Fibonacci decomposition to alter the pixel value (contrast). We trained the model on multiple images from a standard dataset to enhance its adaptability. The hidden layers in the neural network contributed to increasing the complexity of the generated numbers’ randomness. The encryption key stores the method and functions of the encrypted image, enabling the recipient to decrypt it. We evaluated the research using the standard criteria from previous studies to compare the results. Deep learning helped prevent attacks, and Fibonacci analysis helped increase the image’s security. This makes the proposed method a promising solution for digital image security. This study provides a promising solution that is adaptable to various types of images in the age of digital images.

Keywords

• Image encryption
• deep learning
• image security
• Fibonacci decomposition
• neural network