Stereo Vision Reference

Provides stereo block matching for disparity map computation from rectified stereo image pairs.

::: note This module is under active development. API signatures may change between versions. :::

StereoBlockMatcher

pub struct StereoBlockMatcher {
    block_size: i32,
    num_disparities: i32,
    min_disparity: i32,
}

impl StereoBlockMatcher {
    pub fn new(block_size: i32, num_disparities: i32) -> Result<Self>;
    pub fn with_min_disparity(self, min_disparity: i32) -> Self;
    pub fn compute<B: Backend>(
        &self,
        left: &Image<B>,
        right: &Image<B>,
    ) -> Result<Tensor<B, 2>>;
}

Parameters

Parameter	Type	Description
block_size	i32	Size of the matching block (odd number, e.g., 3, 5, 7).
num_disparities	i32	Maximum disparity range (must be divisible by 16).
min_disparity	i32	Minimum disparity to search (default 0).

Example

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

type Backend = Wgpu;
let device = Default::default();

let left = Image::<Backend>::open("left.png", &device)?;
let right = Image::<Backend>::open("right.png", &device)?;

let matcher = StereoBlockMatcher::new(5, 64)?;
let disparity = matcher.compute(&left, &right)?;
println!("Disparity map shape: {:?}", disparity.dims());