demosaic 0.3.0

Demosaicing algorithms for Bayer and X-Trans CFA sensors
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
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234

use crate::bayer;
use crate::cfa::CfaPattern;
use crate::error::DemosaicError;
use crate::BayerAlgorithm;

/// 2x2-bin a single-channel image by averaging each 2x2 block of pixels.
///
/// Reduces dimensions by half in each axis.
///
/// # Arguments
/// - `input`: row-major single-channel data, length = `width * height`
/// - `width`, `height`: input dimensions, must both be even
/// - `output`: pre-allocated buffer, length = `(width / 2) * (height / 2)`
pub fn bin2x2(
    input: &[f32],
    width: usize,
    height: usize,
    output: &mut [f32],
) -> Result<(), DemosaicError> {
    if width % 2 != 0 || height % 2 != 0 {
        return Err(DemosaicError::DimensionsNotEven { width, height });
    }
    let npix = width * height;
    if input.len() != npix {
        return Err(DemosaicError::InputSizeMismatch { expected: npix, got: input.len() });
    }
    let out_w = width / 2;
    let out_h = height / 2;
    let out_npix = out_w * out_h;
    if output.len() != out_npix {
        return Err(DemosaicError::OutputSizeMismatch { expected: out_npix, got: output.len() });
    }

    for by in 0..out_h {
        for bx in 0..out_w {
            let r = 2 * by;
            let c = 2 * bx;
            let sum = input[r * width + c]
                + input[r * width + c + 1]
                + input[(r + 1) * width + c]
                + input[(r + 1) * width + c + 1];
            output[by * out_w + bx] = sum * 0.25;
        }
    }

    Ok(())
}

/// Demosaic a Quad Bayer CFA image via 2x2 binning + standard Bayer demosaic.
///
/// Produces a full-color image at half the input resolution in each dimension.
///
/// # Arguments
/// - `input`: Quad Bayer CFA data, row-major, length = `width * height`
/// - `width`, `height`: input dimensions (must be even)
/// - `cfa`: a 4x4 Quad Bayer CFA pattern (e.g. `CfaPattern::quad_bayer_rggb()`)
/// - `algorithm`: which Bayer demosaic to apply after binning
/// - `output`: pre-allocated buffer, planar CHW, length = `3 * (width/2) * (height/2)`
pub fn demosaic_quad_binned(
    input: &[f32],
    width: usize,
    height: usize,
    cfa: &CfaPattern,
    algorithm: BayerAlgorithm,
    output: &mut [f32],
) -> Result<(), DemosaicError> {
    if !cfa.is_quad_bayer() {
        return Err(DemosaicError::UnsupportedAlgorithm {
            algorithm: "quad_binned",
            cfa: "non-Quad-Bayer",
        });
    }

    if width % 2 != 0 || height % 2 != 0 {
        return Err(DemosaicError::DimensionsNotEven { width, height });
    }

    let npix = width * height;
    if input.len() != npix {
        return Err(DemosaicError::InputSizeMismatch { expected: npix, got: input.len() });
    }

    let out_w = width / 2;
    let out_h = height / 2;
    let out_npix = out_w * out_h;
    if output.len() != 3 * out_npix {
        return Err(DemosaicError::OutputSizeMismatch {
            expected: 3 * out_npix,
            got: output.len(),
        });
    }

    // Step 1: Bin the quad Bayer input into a standard Bayer image.
    let mut binned = alloc::vec![0.0f32; out_npix];
    bin2x2(input, width, height, &mut binned)?;

    // Step 2: Derive the standard Bayer CFA pattern.
    let bayer_cfa = cfa.quad_to_bayer()
        .expect("is_quad_bayer() already validated");

    // Step 3: Dispatch to existing Bayer demosaic.
    match algorithm {
        BayerAlgorithm::Bilinear => bayer::bilinear(&binned, out_w, out_h, &bayer_cfa, output),
        BayerAlgorithm::Mhc => bayer::mhc(&binned, out_w, out_h, &bayer_cfa, output),
        BayerAlgorithm::Ppg => bayer::ppg(&binned, out_w, out_h, &bayer_cfa, output),
        BayerAlgorithm::Ahd => bayer::ahd(&binned, out_w, out_h, &bayer_cfa, output),
        BayerAlgorithm::Vng => bayer::vng(&binned, out_w, out_h, &bayer_cfa, output),
    }

    Ok(())
}

/// Demosaic a Quad Bayer CFA image via binning to interleaved RGB output.
///
/// Same as [`demosaic_quad_binned`], but outputs interleaved HWC layout.
pub fn demosaic_quad_binned_interleaved(
    input: &[f32],
    width: usize,
    height: usize,
    cfa: &CfaPattern,
    algorithm: BayerAlgorithm,
    output: &mut [f32],
) -> Result<(), DemosaicError> {
    let out_w = width / 2;
    let out_h = height / 2;
    let out_npix = out_w * out_h;
    let mut planar = alloc::vec![0.0f32; 3 * out_npix];
    demosaic_quad_binned(input, width, height, cfa, algorithm, &mut planar)?;
    crate::planar_to_interleaved(&planar, output);
    Ok(())
}

#[cfg(test)]
mod tests {
    use alloc::vec;
    use super::*;
    use crate::cfa::Channel;

    #[test]
    fn bin2x2_basic() {
        let input = [1.0, 2.0, 3.0, 4.0,
                     5.0, 6.0, 7.0, 8.0,
                     9.0, 10.0, 11.0, 12.0,
                     13.0, 14.0, 15.0, 16.0f32];
        let mut output = [0.0f32; 4];
        bin2x2(&input, 4, 4, &mut output).unwrap();
        assert_eq!(output[0], (1.0 + 2.0 + 5.0 + 6.0) / 4.0);
        assert_eq!(output[1], (3.0 + 4.0 + 7.0 + 8.0) / 4.0);
        assert_eq!(output[2], (9.0 + 10.0 + 13.0 + 14.0) / 4.0);
        assert_eq!(output[3], (11.0 + 12.0 + 15.0 + 16.0) / 4.0);
    }

    #[test]
    fn bin2x2_odd_dimensions() {
        let input = [0.0f32; 15];
        let mut output = [0.0f32; 4];
        assert!(matches!(
            bin2x2(&input, 5, 3, &mut output),
            Err(DemosaicError::DimensionsNotEven { .. })
        ));
    }

    #[test]
    fn quad_binned_solid_color() {
        for cfa in &[
            CfaPattern::quad_bayer_rggb(),
            CfaPattern::quad_bayer_bggr(),
            CfaPattern::quad_bayer_grbg(),
            CfaPattern::quad_bayer_gbrg(),
        ] {
            let (w, h) = (64, 64);
            let mut input = vec![0.0f32; w * h];
            for y in 0..h {
                for x in 0..w {
                    input[y * w + x] = match cfa.color_at(y, x) {
                        Channel::Red => 0.8,
                        Channel::Green => 0.5,
                        Channel::Blue => 0.3,
                    };
                }
            }

            let ow = w / 2;
            let oh = h / 2;
            let out_npix = ow * oh;
            let mut output = vec![0.0f32; 3 * out_npix];

            for alg in &[
                BayerAlgorithm::Bilinear,
                BayerAlgorithm::Mhc,
                BayerAlgorithm::Ppg,
                BayerAlgorithm::Ahd,
            ] {
                demosaic_quad_binned(&input, w, h, cfa, *alg, &mut output).unwrap();

                let border = 4;
                for y in border..oh - border {
                    for x in border..ow - border {
                        let idx = y * ow + x;
                        let r = output[idx];
                        let g = output[out_npix + idx];
                        let b = output[2 * out_npix + idx];
                        assert!((r - 0.8).abs() < 0.05,
                            "{alg:?}: R at ({y},{x}) = {r}");
                        assert!((g - 0.5).abs() < 0.05,
                            "{alg:?}: G at ({y},{x}) = {g}");
                        assert!((b - 0.3).abs() < 0.05,
                            "{alg:?}: B at ({y},{x}) = {b}");
                    }
                }
            }
        }
    }

    #[test]
    fn quad_binned_output_dimensions() {
        let cfa = CfaPattern::quad_bayer_rggb();
        let (w, h) = (64, 64);
        let input = vec![0.5f32; w * h];
        let ow = w / 2;
        let oh = h / 2;

        // Correct size should succeed
        let mut output = vec![0.0f32; 3 * ow * oh];
        assert!(demosaic_quad_binned(&input, w, h, &cfa, BayerAlgorithm::Bilinear, &mut output).is_ok());

        // Wrong size should fail
        let mut wrong = vec![0.0f32; 3 * w * h];
        assert!(matches!(
            demosaic_quad_binned(&input, w, h, &cfa, BayerAlgorithm::Bilinear, &mut wrong),
            Err(DemosaicError::OutputSizeMismatch { .. })
        ));
    }
}