Function normalize_chw 

pub fn normalize_chw(
    img: &[f32],
    channels: usize,
    h: usize,
    w: usize,
    mean: &[f32],
    std: &[f32],
) -> VisionResult<Vec<f32>>

Normalize a CHW image channel-wise: output[c, h, w] = (input[c, h, w] - mean[c]) / std[c].

Parameters

• img: flat [channels × h × w] input buffer.
• channels: number of channels; must equal mean.len() and std.len().
• h: image height in pixels.
• w: image width in pixels.
• mean: per-channel mean values; length must equal channels.
• std: per-channel standard deviation values; length must equal channels. Each element must be positive.

Errors

Returns VisionError::InvalidImageSize if any dimension is zero. Returns VisionError::DimensionMismatch if img.len() != channels * h * w. Returns VisionError::ShapeMismatch if mean.len() != channels or std.len() != channels. Returns VisionError::NonFinite if any std[c] <= 0 (would produce NaN/Inf).