Efficient Net with Adaptive Siberian Tiger Optimization Based Image Steganography and Steganalysis for Biometric Image

Abstract

In today’s modern world, the need for secure communication has become more paramount particularly in situations where confidentiality is important. Steganalysis intends to determine whether a digital media file comprises concealed data by detecting and analyzing hidden information within the file. The difficulty is effectually differentiating among stego and non-stego files while adapting to increasingly sophisticated steganographic techniques. In this research, an Adaptive Siberian Tiger Optimization_EfficientNet (ASTO_EfficientNet) is designed for image steganography and steganalysis. Firstly, a bit map image is attained from the input biometric cover image. Next, the message to be hidden in the image and bit map image is subjected to the Exclusive-OR (XOR) operation. Thereafter, the XOR-ed outcome is passed to the embedding processes for the generation of the embedded image. Next, the steganalysis process is done to determine the hidden message after the steganography process. The embedded image is  converted into a bit map image and is applied to EfficientNet for detecting the hidden message. The performance of EfficientNet is fine-tuned by tuning the optimum weights employing ASTO. The experimental outcomes demonstrated that the ASTO_EfficientNet acquired superior performance with a maximal Peak-Signal-to-Noise Ratio (PSNR) of 40.765 dB and minimum Bit Error Ratio (BER) of 1.147x10-5.