Module pixels 

Pixel-domain conversion for the Y (luminance) channel.

Provides IDCT (coefficient → pixel) and forward DCT (pixel → coefficient) transforms for geometry-resilient steganography. Only operates on the Y channel — no RGB conversion needed since the DFT template is embedded in luminance.

Functions

dct_block

8×8 forward DCT + quantize → 64 DCT coefficients.

dct_block_unquantized

8×8 forward DCT + divide by QT, but do NOT round.

idct_block

Dequantize + 8×8 IDCT → 64 spatial-domain pixel values.

jpeg_to_luma_f64

Convert entire Y-channel DctGrid → row-major f64 pixel array.

luma_f64_to_jpeg

Write f64 pixel array back into Y-channel DctGrid.

rgb_to_luma_blocks

Convert RGB pixel buffer to Y (luminance) 8×8 blocks.