Blind Dynamic Quantization-Based DWT-QIM Image Steganography Using Block Complexity Analysis

Abstract

Image steganography is widely used to hide information by embedding secret data within digital images while preserving their visual quality. Among various techniques, Quantization Index Modulation (QIM) provides reliable data extraction; however, conventional fixed QIM methods often introduce unnecessary distortion, as a single quantization factor is applied to all image regions regardless of their local characteristics. To address this limitation, this paper proposes a blind dynamic quantization-based image steganography method in the Discrete Wavelet Transform (DWT) domain. The proposed approach utilizes the LH and HL sub-bands of a grayscale image and performs block-based complexity analysis using energy and variance measurements. A raw quantization value is generated according to the calculated complexity score and mapped to stable quantization levels for QIM embedding. Experiments are conducted on five 512 × 512 grayscale benchmark images, with payloads ranging from 24,000 to 120,000 bits. For a payload of 120,000 bits, the proposed method achieves an average PSNR of 44.94 dB, an average SSIM of 0.9897, and an average BER of 1.7770%. The experimental results show that the proposed framework is able to achieve good visual quality and blind data extraction in the DWT domain.