Struct PIAPixelizer 

pub struct PIAPixelizer { /* private fields */ }

A Pixelizer that relies on Pixelated Image Abstraction (PIA).

Implementations

impl PIAPixelizer

pub fn new( temperature_init: Option<f32>, temperature_final: f32, m: f32, alpha: f32, verbose: bool, ) -> Self

Creates a new PIA Pixelizer with custom main parameters.

Arguments

• temperature_init - Optional starting temp. Pass None to auto-calculate based on image content.
• temperature_final - Stop threshold (usually 1.0).
• m - Compactness factor. High = Square pixels, Low = Adaptive shapes.
• alpha - Cooling rate (0.0 - 1.0).
• verbose - Enable logging.

pub fn set_verbose(&mut self, verbose: bool)

Enables or disables printing iteration progress to stdout.

Useful for debugging the simulated annealing process or tracking convergence speed.

Trait Implementations

impl Default for PIAPixelizer

fn default() -> Self

Creates a pixelizer with the parameters recommended by the original Gerstner et al. (2012) paper.

Defaults

• m: 45.0 (High compactness)
• alpha: 0.7 (Moderate cooling rate)
• temperature_final: 1.0
• perturbation: 0.8
• post_process_saturation: 1.1

impl Pixelizer for PIAPixelizer

fn pixelize( &self, img: &DynamicImage, width: u32, height: u32, num_colors: usize, ) -> Result<DynamicImage, PixelizationError>

Applies the PIA algorithm to the input image.

Performance Note

Unlike K-Means, this algorithm runs an iterative simulated annealing process. It is significantly slower. For large target dimensions (>128px), execution time may be high.

Constraints

• num_colors must be between 2 and 256.

Process

1. Converts image to CIELAB color space.
2. Initializes superpixels on a grid.
3. Iteratively refines superpixel shapes and the global color palette using Simulated Annealing.
4. Renders the final image with the reduced palette.

Errors

Returns PixelizationError::ColorError if num_colors is not between 2 and 256.