API Reference Overview

Welcome to the API reference docs for Iris. This page introduces the major structures, traits, and modules exported by the library.

Main Types

Image

The primary image representation type. It wraps a multi-channel Burn tensor of shape [Channels, Height, Width].

pub struct Image<B: Backend> {
    pub tensor: Tensor<B, 3>,
}

Point

Represents a coordinate pair (x, y) in 2D space.

pub struct Point<T> {
    pub x: T,
    pub y: T,
}

Rect

Defines a 2D bounding rectangle.

pub struct Rect<T> {
    pub x: T,
    pub y: T,
    pub width: T,
    pub height: T,
}

Scalar

A 4-element floating-point array, typically representing color values (e.g. RGBA).

pub struct Scalar(pub [f64; 4]);

Size

A structure defining width and height dimensions.

pub struct Size<T> {
    pub width: T,
    pub height: T,
}

Module Index

Module	Description
core	Point, Rect, Size, Scalar, Mat, Rng, image math ops, bitwise ops, normalization.
image	Image struct, open, save, resize, crop, flip, rotate, geometric transforms.
filters	Box, Gaussian, median, bilateral blur, separable filter, distance transform, filter2D.
color	RGB/HSV/HLS/XYZ/Lab/YUV/YCrCb/CMYK conversions, split/merge channels.
edges	Sobel, Canny, Scharr, Laplacian, Hough lines, Hough circles.
morphology	Dilate, erode, open, close, morphology_ex, custom kernels, hit-or-miss, thinning, skeleton.
threshold	Binary, Otsu, triangle, adaptive thresholding.
histogram	Histogram computation, equalization, CLAHE, LUT, comparison, 2D histogram.
drawing	Lines, rectangles, circles, ellipses, text, polylines, arrows, markers, fill poly.
noise	Gaussian, salt-and-pepper, and speckle noise generation.
contours	Contour detection, convex hull, moments, shape analysis, hierarchy, convexity defects.
features	ORB feature detection, keypoints, BFMatcher, FLANN matcher, template matching.
tracking	MOSSE tracker, MeanShift tracker, background subtraction.
dnn	ONNX model loading, weight loaders, blob preprocessing, NMS.
video	Video capture, reading, writing, frame iteration, metadata.
inpaint	Telea Fast Marching Method inpainting for damaged image regions.
stereo	Stereo block matching for disparity map computation.
kalman	Discrete Kalman filter for state estimation and prediction.
hog	Histogram of Oriented Gradients descriptor for feature extraction.
photo	Non-Local Means denoising and Mertens exposure fusion.
stitching	Image stitching for panorama creation.

Importing the Prelude

use iris::prelude::*;
use burn::backend::wgpu::Wgpu;

type Backend = Wgpu;

The prelude re-exports all commonly used types: Image, Point, Rect, Scalar, Size, Mat, Tensor, Backend, Result, and all module-specific types.