Steganalisis

Steganalysis is the field concerned with detecting the presence of steganography, the practice of embedding hidden information within ordinary media such as digital images, audio, video, or network traffic. The goal is to determine whether a payload exists, identify the type of steganography used when possible, and, in some cases, recover or extract the hidden data without damaging the cover medium.

Steganalysis distinguishes between targeted (or dedicated) analysis, which looks for specific known embedding methods, and blind or universal steganalysis, which aims to detect a wide range of techniques without prior knowledge of the method. It can also be classified as passive, focusing on observation and evidence collection, or active in contexts where adjustments to the medium may be attempted to impede concealment (though this latter use is controversial in forensics).

Techniques rely on statistical, structural, or perceptual inconsistencies introduced by embedding. Common approaches include feature-based analyses, statistical tests on color channels or transform-domain coefficients, and machine learning classifiers trained to separate cover from stego samples. Transform-domain methods (e.g., DCT, wavelet) are often more robust against simple edits, while spatial-domain methods examine pixel-level patterns. Rich models, SPAM features, and other high-dimensional descriptors are used to capture subtle irregularities. Some workflows include attempts to extract or reconstruct the hidden payload.

Practical steganalysis follows a workflow: collect suspect media, preprocess, extract features, apply classifiers, and validate decisions. Proven effectiveness varies by medium, embedding method, and payload size. The field supports digital forensics, incident response, and security research but must balance sensitivity with the risk of false positives and privacy considerations.