PCA Scanner

Principal Component Analysis (PCA) transforms the three correlated RGB color channels into three uncorrelated components ordered by how much image variation they capture. The first component (PC1) typically represents overall brightness and resembles a grayscale conversion. The second (PC2) captures the main chromatic contrast — often warm-vs-cool tones. The third (PC3) holds the residual variation, which in natural photos is mostly sensor noise. In manipulated images, PC3 often reveals structured artifacts invisible in normal viewing. For tonal distribution analysis, pair PCA with the Histogram Analyzer. For overall quality scoring, see the Quality Analyzer.

PCA is a powerful forensic technique because it separates meaningful image content from noise and artifacts. When an image has been composited from multiple sources, PC3 often shows sharp boundaries between regions with different noise characteristics. AI-generated images may produce unusually clean or uniform PC3 components compared to photographs captured by real sensors. Color-graded or heavily filtered images show distinctive eigenvector orientations that differ from natural lighting. For dedicated AI detection, use the AI Detector. For noise pattern analysis, try the Noise Analysis Scanner.

The eigenvectors describe the direction of each principal component in RGB color space — they tell you which mix of red, green, and blue makes up each component. The eigenvalues measure how much variance each direction captures. In a typical outdoor photo, PC1 might explain 90% of variance with an eigenvector close to equal RGB weights (luminance), while PC3 explains under 3%. Unusual variance distribution or eigenvector orientations can indicate non-standard color processing. For per-channel statistical breakdown, see the Histogram Analyzer. For bit-level channel inspection, try the Bit Plane Analyzer.

Steganographic techniques that embed data across color channels often disturb the natural correlation between RGB values. PCA is particularly effective at detecting these disturbances because the embedded data appears as structured noise in the minor components. When combined with contrast enhancement, even subtle steganographic modifications become visible as patterns in the PC3 view. For dedicated steganography detection, use the Stego Scanner. For LSB-level analysis, see the Bit Plane Analyzer. For compression-based detection, try the ELA Scanner.