Skip to main content

estimate_noise_sigma_u8

Function estimate_noise_sigma_u8 

Source
pub fn estimate_noise_sigma_u8(
    img: &[u8],
    width: u32,
    height: u32,
) -> Result<Option<f32>, ChessError>
Expand description

Estimate the standard deviation of an 8-bit grayscale image’s pixel noise, in gray levels.

This is the estimate Detector uses to express position covariances in px² under NoiseLevel::Estimate; calling it on the same image returns the same value. Strong edges, corners, smooth shading and areas near black or white are excluded, so it suits images dominated by a calibration board. It is not reliable on demosaiced or JPEG images; supply NoiseLevel::Known for those. Cost is independent of the image size (well under a millisecond).

img is width * height bytes, row-major. Returns Ok(None) when the image is smaller than 4×4 or has too few structure-free areas for an estimate. The full accuracy statement is on [chess_corners_core::estimate_noise_sigma_u8].

// A flat gray image with no noise: the estimate is the 8-bit
// rounding floor, near zero.
let img = vec![128u8; 64 * 64];
let sigma = chess_corners::estimate_noise_sigma_u8(&img, 64, 64)?;
assert!(sigma.is_none_or(|s| s < 0.5));

§Errors

Returns ChessError::DimensionMismatch if img.len() != width * height.