Intelligent Pixel Selection Steganography: An AI-Enhanced Adaptive LSB Approach with AES Encryption

Image steganography conceals secret information within digital images to prevent unauthorized detection of hidden data. This paper presents a lightweight, AI-enhanced adaptive LSB steganography approach using OpenCV to identify visually complex regions for secure data embedding. The secret message is encrypted using AES prior to embedding. The proposed method achieves a PSNR of 47.74 dB and SSIM of 0.9990, outperforming traditional LSB and rivaling deep learning-based methods — without requiring model training or GPU acceleration.