stegoeggo 0.2.3

Rights-reservation metadata and AI-training restriction notices for images, with optional steganographic markers
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
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277

#![allow(dead_code)]

use image::RgbaImage;

/// Compute embedding cost for each pixel in an RGBA image.
///
/// Returns a `Vec<f32>` of length `width * height`, one cost per pixel.
/// Lower cost = cheaper to modify (prefer embedding here).
/// Higher cost = more expensive to modify (avoid embedding here).
///
/// Uses a Laplacian-based local complexity metric:
/// - Smooth regions (low local energy) → high cost (avoid)
/// - Textured regions (high local energy) → low cost (prefer)
///
/// The cost is normalized so the minimum is 1.0 and the maximum is
/// approximately `max_cost` (default 10.0).
pub fn compute_pixel_costs(img: &RgbaImage, max_cost: f32) -> Vec<f32> {
    let (w, h) = img.dimensions();
    let total = (w as usize) * (h as usize);
    let mut costs = vec![0.0f32; total];

    // Convert to grayscale for cost computation
    let gray: Vec<f32> = img
        .pixels()
        .map(|p| {
            let r = p[0] as f32;
            let g = p[1] as f32;
            let b = p[2] as f32;
            0.299 * r + 0.587 * g + 0.114 * b
        })
        .collect();

    for y in 0..h {
        for x in 0..w {
            let idx = (y as usize) * w as usize + x as usize;

            // Laplacian energy: sum of squared differences with 4-neighbors
            let center = gray[idx];
            let mut energy = 0.0f32;

            if x > 0 {
                let diff = center - gray[idx - 1];
                energy += diff * diff;
            }
            if x + 1 < w {
                let diff = center - gray[idx + 1];
                energy += diff * diff;
            }
            if y > 0 {
                let diff = center - gray[idx - w as usize];
                energy += diff * diff;
            }
            if y + 1 < h {
                let diff = center - gray[idx + w as usize];
                energy += diff * diff;
            }

            // Also compute local variance in 3x3 window for better texture detection
            let mut sum = 0.0f32;
            let mut count = 0u32;
            for dy in -1i32..=1 {
                for dx in -1i32..=1 {
                    let nx = x as i32 + dx;
                    let ny = y as i32 + dy;
                    if nx >= 0 && nx < w as i32 && ny >= 0 && ny < h as i32 {
                        let ni = ny as usize * w as usize + nx as usize;
                        sum += gray[ni];
                        count += 1;
                    }
                }
            }
            let mean = sum / count as f32;
            let mut variance = 0.0f32;
            for dy in -1i32..=1 {
                for dx in -1i32..=1 {
                    let nx = x as i32 + dx;
                    let ny = y as i32 + dy;
                    if nx >= 0 && nx < w as i32 && ny >= 0 && ny < h as i32 {
                        let ni = ny as usize * w as usize + nx as usize;
                        let diff = gray[ni] - mean;
                        variance += diff * diff;
                    }
                }
            }
            variance /= count as f32;

            // Combined complexity: Laplacian energy + local variance
            let complexity = energy + variance;

            // Cost is inversely proportional to complexity
            // Smooth pixels (complexity ≈ 0) get high cost
            // Textured pixels (complexity high) get low cost
            let cost = if complexity < 1.0 {
                max_cost
            } else {
                (1.0 / complexity.sqrt()) * max_cost
            };

            costs[idx] = cost;
        }
    }

    costs
}

/// Compute embedding costs for a grayscale slice of an RGBA image.
///
/// This is used for tiled embedding where only a sub-image is available.
pub fn compute_pixel_costs_sub(
    img: &RgbaImage,
    x0: u32,
    y0: u32,
    w: u32,
    h: u32,
    max_cost: f32,
) -> Vec<f32> {
    let total = (w as usize) * (h as usize);
    let mut costs = vec![0.0f32; total];

    // Extract grayscale sub-image
    let gray: Vec<f32> = (0..h)
        .flat_map(|dy| {
            (0..w).map(move |dx| {
                let p = img.get_pixel(x0 + dx, y0 + dy);
                let r = p[0] as f32;
                let g = p[1] as f32;
                let b = p[2] as f32;
                0.299 * r + 0.587 * g + 0.114 * b
            })
        })
        .collect();

    for y in 0..h {
        for x in 0..w {
            let idx = (y as usize) * w as usize + x as usize;
            let center = gray[idx];
            let mut energy = 0.0f32;

            if x > 0 {
                let diff = center - gray[idx - 1];
                energy += diff * diff;
            }
            if x + 1 < w {
                let diff = center - gray[idx + 1];
                energy += diff * diff;
            }
            if y > 0 {
                let diff = center - gray[idx - w as usize];
                energy += diff * diff;
            }
            if y + 1 < h {
                let diff = center - gray[idx + w as usize];
                energy += diff * diff;
            }

            let mut sum = 0.0f32;
            let mut count = 0u32;
            for dy in -1i32..=1 {
                for dx in -1i32..=1 {
                    let nx = x as i32 + dx;
                    let ny = y as i32 + dy;
                    if nx >= 0 && nx < w as i32 && ny >= 0 && ny < h as i32 {
                        let ni = ny as usize * w as usize + nx as usize;
                        sum += gray[ni];
                        count += 1;
                    }
                }
            }
            let mean = sum / count as f32;
            let mut variance = 0.0f32;
            for dy in -1i32..=1 {
                for dx in -1i32..=1 {
                    let nx = x as i32 + dx;
                    let ny = y as i32 + dy;
                    if nx >= 0 && nx < w as i32 && ny >= 0 && ny < h as i32 {
                        let ni = ny as usize * w as usize + nx as usize;
                        let diff = gray[ni] - mean;
                        variance += diff * diff;
                    }
                }
            }
            variance /= count as f32;

            let complexity = energy + variance;
            let cost = if complexity < 1.0 {
                max_cost
            } else {
                (1.0 / complexity.sqrt()) * max_cost
            };
            costs[idx] = cost;
        }
    }

    costs
}

#[cfg(test)]
mod tests {
    use super::*;
    use image::{ImageBuffer, Rgba, RgbaImage};

    fn make_smooth_image(w: u32, h: u32) -> RgbaImage {
        ImageBuffer::from_pixel(w, h, Rgba([128, 128, 128, 255]))
    }

    fn make_textured_image(w: u32, h: u32) -> RgbaImage {
        ImageBuffer::from_fn(w, h, |x, y| {
            Rgba([
                ((x * 73 + y * 151) % 256) as u8,
                ((x * 53 + y * 97) % 256) as u8,
                ((x * 29 + y * 43) % 256) as u8,
                255,
            ])
        })
    }

    #[test]
    fn smooth_image_has_high_costs() {
        let img = make_smooth_image(16, 16);
        let costs = compute_pixel_costs(&img, 10.0);
        // Smooth image should have high costs (expensive to modify)
        let avg: f32 = costs.iter().sum::<f32>() / costs.len() as f32;
        assert!(
            avg > 8.0,
            "smooth image avg cost should be > 8.0, got {avg}"
        );
    }

    #[test]
    fn textured_image_has_lower_costs() {
        let img = make_textured_image(16, 16);
        let costs = compute_pixel_costs(&img, 10.0);
        let avg: f32 = costs.iter().sum::<f32>() / costs.len() as f32;
        assert!(
            avg < 8.0,
            "textured image avg cost should be < 8.0, got {avg}"
        );
    }

    #[test]
    fn costs_are_non_negative() {
        let img = make_textured_image(32, 32);
        let costs = compute_pixel_costs(&img, 10.0);
        for &c in &costs {
            assert!(c >= 0.0, "cost should be non-negative, got {c}");
        }
    }

    #[test]
    fn costs_length_matches_image() {
        let img = make_textured_image(10, 8);
        let costs = compute_pixel_costs(&img, 10.0);
        assert_eq!(costs.len(), 80);
    }

    #[test]
    fn sub_cost_matches_full_cost_at_interior() {
        let img = make_textured_image(32, 32);
        let full_costs = compute_pixel_costs(&img, 10.0);
        let sub_costs = compute_pixel_costs_sub(&img, 0, 0, 16, 16, 10.0);

        // Compare at interior positions (away from boundaries where neighbor
        // handling differs between full-image and sub-image computation)
        for y in 2..14 {
            for x in 2..14 {
                let full_idx = y * 32 + x;
                let sub_idx = y * 16 + x;
                assert!(
                    (full_costs[full_idx] - sub_costs[sub_idx]).abs() < 0.001,
                    "cost mismatch at ({x}, {y}): full={} sub={}",
                    full_costs[full_idx],
                    sub_costs[sub_idx]
                );
            }
        }
    }
}