vision-calibration-core 0.2.0

Core math types, camera models, and RANSAC primitives for vision-calibration
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209

//! Coordinate transformation utilities for camera projection.
//!
//! This module provides functions for converting between pixel coordinates,
//! normalized camera coordinates, and for applying/reversing distortion models.

use crate::{DistortionModel, Mat3, Pt2, Real, Vec3};

/// Convert pixel coordinates to normalized coordinates using intrinsics.
///
/// Applies K^{-1} to transform pixel coordinates to the normalized image plane
/// (Z=1 in camera frame).
///
/// # Arguments
/// * `pixel` - Pixel coordinates (u, v)
/// * `intrinsics` - Camera intrinsics matrix K
///
/// # Returns
/// Normalized coordinates (x, y) on the Z=1 plane.
///
/// # Panics
/// Panics if the intrinsics matrix is singular (non-invertible).
///
/// # Example
/// ```no_run
/// use vision_calibration_core::{Mat3, Pt2};
/// use vision_calibration_core::pixel_to_normalized;
///
/// let k = Mat3::new(800.0, 0.0, 640.0, 0.0, 800.0, 480.0, 0.0, 0.0, 1.0);
/// let pixel = Pt2::new(640.0, 480.0);
/// let normalized = pixel_to_normalized(pixel, &k);
/// // normalized ≈ (0.0, 0.0) for a pixel at the principal point
/// ```
pub fn pixel_to_normalized(pixel: Pt2, intrinsics: &Mat3) -> Pt2 {
    let k_inv = intrinsics
        .try_inverse()
        .expect("intrinsics matrix should be invertible");
    let v = k_inv * Vec3::new(pixel.x, pixel.y, 1.0);
    Pt2::new(v.x / v.z, v.y / v.z)
}

/// Convert normalized coordinates to pixel coordinates using intrinsics.
///
/// Applies K to transform normalized coordinates (Z=1 plane) to pixel coordinates.
///
/// # Arguments
/// * `normalized` - Normalized coordinates (x, y)
/// * `intrinsics` - Camera intrinsics matrix K
///
/// # Returns
/// Pixel coordinates computed as K * [x, y, 1]^T with homogeneous division.
///
/// # Example
/// ```no_run
/// use vision_calibration_core::{Mat3, Pt2};
/// use vision_calibration_core::normalized_to_pixel;
///
/// let k = Mat3::new(800.0, 0.0, 640.0, 0.0, 800.0, 480.0, 0.0, 0.0, 1.0);
/// let normalized = Pt2::new(0.0, 0.0);
/// let pixel = normalized_to_pixel(normalized, &k);
/// // pixel ≈ (640.0, 480.0) at the principal point
/// ```
pub fn normalized_to_pixel(normalized: Pt2, intrinsics: &Mat3) -> Pt2 {
    let v = intrinsics * Vec3::new(normalized.x, normalized.y, 1.0);
    Pt2::new(v.x / v.z, v.y / v.z)
}

/// Undistort pixel coordinates to normalized coordinates.
///
/// Combines pixel-to-normalized conversion with distortion model undistortion.
/// This is a common operation when preprocessing image points for algorithms
/// that assume ideal pinhole projection.
///
/// # Arguments
/// * `pixel` - Distorted pixel coordinates
/// * `intrinsics` - Camera intrinsics matrix K
/// * `distortion` - Distortion model implementing [`DistortionModel`]
///
/// # Returns
/// Undistorted normalized coordinates on the Z=1 plane.
///
/// # Algorithm
/// 1. Convert pixel to normalized coordinates: `n = K^{-1} * [x, y, 1]^T`
/// 2. Apply iterative undistortion: `n_undist = distortion.undistort(n)`
///
/// # Example
/// ```no_run
/// use vision_calibration_core::{BrownConrady5, Mat3, Pt2};
/// use vision_calibration_core::undistort_pixel;
///
/// let k = Mat3::new(800.0, 0.0, 640.0, 0.0, 800.0, 480.0, 0.0, 0.0, 1.0);
/// let distortion = BrownConrady5 { k1: -0.3, k2: 0.1, k3: 0.0, p1: 0.0, p2: 0.0, iters: 5 };
/// let pixel = Pt2::new(500.0, 400.0);
/// let undist_norm = undistort_pixel(pixel, &k, &distortion);
/// ```
pub fn undistort_pixel<D: DistortionModel<Real>>(
    pixel: Pt2,
    intrinsics: &Mat3,
    distortion: &D,
) -> Pt2 {
    let normalized = pixel_to_normalized(pixel, intrinsics);
    distortion.undistort(&normalized)
}

/// Apply distortion and convert to pixel coordinates.
///
/// Combines distortion model application with normalized-to-pixel conversion.
/// This is the inverse operation of [`undistort_pixel`] (approximately, depending
/// on distortion model convergence).
///
/// # Arguments
/// * `normalized` - Undistorted normalized coordinates
/// * `intrinsics` - Camera intrinsics matrix K
/// * `distortion` - Distortion model implementing [`DistortionModel`]
///
/// # Returns
/// Distorted pixel coordinates.
///
/// # Algorithm
/// 1. Apply forward distortion: `n_dist = distortion.distort(n)`
/// 2. Convert to pixels: `pixel = K * [n_dist.x, n_dist.y, 1]^T`
///
/// # Example
/// ```no_run
/// use vision_calibration_core::{BrownConrady5, Mat3, Pt2};
/// use vision_calibration_core::distort_to_pixel;
///
/// let k = Mat3::new(800.0, 0.0, 640.0, 0.0, 800.0, 480.0, 0.0, 0.0, 1.0);
/// let distortion = BrownConrady5 { k1: -0.3, k2: 0.1, k3: 0.0, p1: 0.0, p2: 0.0, iters: 5 };
/// let normalized = Pt2::new(-0.1, 0.05);
/// let pixel = distort_to_pixel(normalized, &k, &distortion);
/// ```
pub fn distort_to_pixel<D: DistortionModel<Real>>(
    normalized: Pt2,
    intrinsics: &Mat3,
    distortion: &D,
) -> Pt2 {
    let distorted = distortion.distort(&normalized);
    normalized_to_pixel(distorted, intrinsics)
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::BrownConrady5;

    #[test]
    fn pixel_normalized_roundtrip() {
        let k = Mat3::new(800.0, 0.0, 640.0, 0.0, 780.0, 360.0, 0.0, 0.0, 1.0);
        let pixel_orig = Pt2::new(700.0, 400.0);

        let normalized = pixel_to_normalized(pixel_orig, &k);
        let pixel_back = normalized_to_pixel(normalized, &k);

        assert!((pixel_back.x - pixel_orig.x).abs() < 1e-10);
        assert!((pixel_back.y - pixel_orig.y).abs() < 1e-10);
    }

    #[test]
    fn principal_point_maps_to_origin() {
        let k = Mat3::new(800.0, 0.0, 640.0, 0.0, 800.0, 480.0, 0.0, 0.0, 1.0);
        let pixel = Pt2::new(640.0, 480.0);

        let normalized = pixel_to_normalized(pixel, &k);

        assert!(normalized.x.abs() < 1e-10);
        assert!(normalized.y.abs() < 1e-10);
    }

    #[test]
    fn undistort_pixel_no_distortion() {
        let k = Mat3::new(800.0, 0.0, 640.0, 0.0, 800.0, 480.0, 0.0, 0.0, 1.0);
        let distortion = BrownConrady5 {
            k1: 0.0,
            k2: 0.0,
            k3: 0.0,
            p1: 0.0,
            p2: 0.0,
            iters: 5,
        };
        let pixel = Pt2::new(700.0, 500.0);

        let undist = undistort_pixel(pixel, &k, &distortion);
        let expected = pixel_to_normalized(pixel, &k);

        assert!((undist.x - expected.x).abs() < 1e-10);
        assert!((undist.y - expected.y).abs() < 1e-10);
    }

    #[test]
    fn distort_undistort_approximate_roundtrip() {
        let k = Mat3::new(800.0, 0.0, 640.0, 0.0, 800.0, 480.0, 0.0, 0.0, 1.0);
        let distortion = BrownConrady5 {
            k1: -0.3,
            k2: 0.1,
            k3: 0.0,
            p1: 0.001,
            p2: -0.001,
            iters: 5,
        };

        let normalized_orig = Pt2::new(-0.1, 0.05);
        let pixel_dist = distort_to_pixel(normalized_orig, &k, &distortion);
        let normalized_undist = undistort_pixel(pixel_dist, &k, &distortion);

        // Should be close but not exact due to iterative undistortion
        let diff = (normalized_undist - normalized_orig).norm();
        assert!(diff < 1e-6, "roundtrip error too large: {}", diff);
    }
}