Advancements in Image Steganalysis: Integrating Deep Learning and Statistical Feature Analysis

With the intriguing concept of Steganography and Steganalysis, this research paper focuses on analyzing images to detect cover images. The system core is the Deep convolutional neural networks (CNNs), capable of automatically recognizing patterns indicative of steganography. Deep convolutional neural networks combined with Markov features-based analysis, spatial pixel distribution and subtractive pixel adjacency matrices to analyze the statistical properties can significantly improve anomaly detection and steganalysis. The Gabor Filter Residual algorithm detects hidden messages by recognizing anomalies in the residuals left after the Gabor Filters application. Ensemble classifiers contribute to creating a robustsystem with better generalizability. A multimodal deep learning framework assimilates various sourced data, adding the ability to detect a broad range of steganography methods. The traditional RS Analysis method can compare/create a baseline for the system's effectiveness. Lastly, Lightweight Steganalysis with block-wise pruning reduces the computational cost and is beneficial in resourceconstrained environments.