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

use num_traits::clamp;

use crate::{ImageBuffer, Pixel, Primitive};

/// Approximation of Gaussian blur after
/// Kovesi, P.:  Fast Almost-Gaussian Filtering The Australian Pattern
/// Recognition Society Conference: DICTA 2010. December 2010. Sydney.
pub fn fast_blur<P: Pixel>(
    image_buffer: &ImageBuffer<P, Vec<P::Subpixel>>,
    sigma: f32,
) -> ImageBuffer<P, Vec<P::Subpixel>> {
    let (width, height) = image_buffer.dimensions();

    if width == 0 || height == 0 {
        return image_buffer.clone();
    }
    let mut samples = image_buffer.as_flat_samples().samples.to_vec();
    let num_passes = 3;

    let boxes = boxes_for_gauss(sigma, num_passes);

    for radius in boxes.iter().take(num_passes) {
        let horizontally_blurred_transposed = horizontal_fast_blur_half::<P::Subpixel>(
            &samples,
            width as usize,
            height as usize,
            (*radius - 1) / 2,
            P::CHANNEL_COUNT as usize,
        );
        samples = horizontal_fast_blur_half::<P::Subpixel>(
            &horizontally_blurred_transposed,
            height as usize,
            width as usize,
            (*radius - 1) / 2,
            P::CHANNEL_COUNT as usize,
        );
    }
    ImageBuffer::from_raw(width, height, samples).unwrap()
}

fn boxes_for_gauss(sigma: f32, n: usize) -> Vec<usize> {
    let w_ideal = f32::sqrt((12.0 * sigma.powi(2) / (n as f32)) + 1.0);
    let mut w_l = w_ideal.floor();
    if w_l % 2.0 == 0.0 {
        w_l -= 1.0
    };
    let w_u = w_l + 2.0;

    let m_ideal = 0.25 * (n as f32) * (w_l + 3.0) - 3.0 * sigma.powi(2) * (w_l + 1.0).recip();

    let m = f32::round(m_ideal) as usize;

    (0..n)
        .map(|i| if i < m { w_l as usize } else { w_u as usize })
        .collect::<Vec<_>>()
}

fn channel_idx(channel: usize, idx: usize, channel_num: usize) -> usize {
    channel_num * idx + channel
}

fn horizontal_fast_blur_half<P: Primitive>(
    samples: &[P],
    width: usize,
    height: usize,
    r: usize,
    channel_num: usize,
) -> Vec<P> {
    let channel_size = width * height;

    let mut out_samples = vec![P::from(0).unwrap(); channel_size * channel_num];
    let mut vals = vec![0.0; channel_num];

    let min_value = P::DEFAULT_MIN_VALUE.to_f32().unwrap();
    let max_value = P::DEFAULT_MAX_VALUE.to_f32().unwrap();

    for row in 0..height {
        for (channel, value) in vals.iter_mut().enumerate().take(channel_num) {
            *value = ((-(r as isize))..(r + 1) as isize)
                .map(|x| {
                    extended_f(
                        samples,
                        width,
                        height,
                        x,
                        row as isize,
                        channel,
                        channel_num,
                    )
                    .to_f32()
                    .unwrap_or(0.0)
                })
                .sum()
        }

        for column in 0..width {
            for (channel, channel_val) in vals.iter_mut().enumerate() {
                let val = *channel_val / (2.0 * r as f32 + 1.0);
                let val = clamp(val, min_value, max_value);
                let val = P::from(val).unwrap();

                let destination_row = column;
                let destination_column = row;
                let destination_sample_index = channel_idx(
                    channel,
                    destination_column + destination_row * height,
                    channel_num,
                );
                out_samples[destination_sample_index] = val;
                *channel_val = *channel_val
                    - extended_f(
                        samples,
                        width,
                        height,
                        column as isize - r as isize,
                        row as isize,
                        channel,
                        channel_num,
                    )
                    .to_f32()
                    .unwrap_or(0.0)
                    + extended_f(
                        samples,
                        width,
                        height,
                        { column + r + 1 } as isize,
                        row as isize,
                        channel,
                        channel_num,
                    )
                    .to_f32()
                    .unwrap_or(0.0)
            }
        }
    }

    out_samples
}

fn extended_f<P: Primitive>(
    samples: &[P],
    width: usize,
    height: usize,
    x: isize,
    y: isize,
    channel: usize,
    channel_num: usize,
) -> P {
    let x = clamp(x, 0, width as isize - 1) as usize;
    let y = clamp(y, 0, height as isize - 1) as usize;
    samples[channel_idx(channel, y * width + x, channel_num)]
}