imgal 0.3.0

A fast and open-source scientific image processing and algorithm library.
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

use std::collections::HashMap;

use ndarray::{Array1, Array2, ArrayBase, AsArray, Axis, Dimension, ViewRepr};
use rayon::prelude::*;

use crate::prelude::*;

/// Create a ROI point cloud map from an n-dimensional label image.
///
/// # Description
///
/// Creates a region of interest (ROI) "cloud" map from an n-dimensional label
/// image. For a given input image each label is converted into a 2D array
/// representing a point cloud with shape `(p, D)`, where `p` and `D` are the
/// number of points and dimensions respectively. Each label's point cloud is
/// stored with it's associated key (*i.e.* label ID) in the output `HashMap`.
///
/// # Arguments
///
/// * `labels`: The n-dimensional label image.
/// * `threads`: The requested number of threads to use for parallel execution.
///   If `None` or `Some(1)` sequential execution is used. If `Some(0)`, then
///   the maximum available parallelism is used. Thread counts are clamped to
///   the systems maximum.
///
/// # Returns
///
/// * `HashMap<u64, Array2<usize>>`: A ROI `HashMap` where the keys are the ROI
///   label IDs and values are the ROI point clouds.
#[inline]
pub fn roi_cloud_map<'a, A, D>(labels: A, threads: Option<usize>) -> HashMap<u64, Array2<usize>>
where
    A: AsArray<'a, u64, D>,
    D: Dimension,
{
    let data: ArrayBase<ViewRepr<&'a u64>, D> = labels.into();
    let vec_to_arr = |k: u64, v: Vec<Vec<usize>>| {
        let arr = Array2::from_shape_vec((v.len(), v[0].len()), v.into_iter().flatten().collect())
            .expect("Failed to reshape ROI point cloud into an Array2<usize>.");
        (k, arr)
    };
    let labels_to_map_seq = || {
        let mut cloud_map: HashMap<u64, Vec<Vec<usize>>> = HashMap::new();
        data.view()
            .into_dyn()
            .indexed_iter()
            .filter(|&(_, &v)| v != 0)
            .for_each(|(p, &v)| {
                cloud_map
                    .entry(v)
                    .or_default()
                    .push(p.as_array_view().to_vec());
            });
        cloud_map
    };
    let labels_to_map_par = || {
        data.view()
            .into_dyn()
            .indexed_iter()
            .par_bridge()
            .filter(|&(_, &v)| v != 0)
            .fold(
                HashMap::new,
                |mut map: HashMap<u64, Vec<Vec<usize>>>, (p, &v)| {
                    map.entry(v).or_default().push(p.as_array_view().to_vec());
                    map
                },
            )
            .reduce(HashMap::new, |mut map_a, map_b| {
                map_b.into_iter().for_each(|(k, mut v)| {
                    map_a.entry(k).or_insert_with(Vec::new).append(&mut v);
                });
                map_a
            })
    };
    let cloud_map = par!(threads,
        seq_exp: labels_to_map_seq(),
        par_exp: labels_to_map_par());
    cloud_map
        .into_par_iter()
        .map(|(k, v)| vec_to_arr(k, v))
        .collect()
}

/// Create a ROI data map from n-dimensional data and a label image.
///
/// # Description
///
/// Creates a region of interest (ROI) "data" map from input n-dimensional data
/// and label images. For a given `data` and `labels` image pair, each
/// coordinate within every label in the label image is used to query the
/// image data. Each label's associated raw data is stored as a 1D array with
/// the label's key (*i.e.* label ID) in the output `HashMap`.
///
/// # Arguments
///
/// * `data`: The input n-dimensional image data.
/// * `labels`: The corresponding n-dimensional label image for `data`.
/// * `threads`: The requested number of threads to use for parallel execution.
///   If `None` or `Some(1)` sequential execution is used. If `Some(0)`, then
///   the maximum available parallelism is used. Thread counts are clamped to
///   the systems maximum.
///
/// # Returns
///
/// * `Ok(HashMap<u64, Array1<T>>)`: A ROI `HashMap` where the keys are the ROI
///   label IDs and the values are 1D arrays containing raw values from the ROI.
/// * `Err(ImgalError)`: If `data.shape() != labels.shape()`.
#[inline]
pub fn roi_data_map<'a, T, A, B, D>(
    data: A,
    labels: B,
    threads: Option<usize>,
) -> Result<HashMap<u64, Array1<T>>, ImgalError>
where
    A: AsArray<'a, T, D>,
    B: AsArray<'a, u64, D>,
    D: Dimension,
    T: 'a + AsNumeric,
{
    let data: ArrayBase<ViewRepr<&'a T>, D> = data.into();
    let labels: ArrayBase<ViewRepr<&'a u64>, D> = labels.into();
    if data.shape() != labels.shape() {
        return Err(ImgalError::MismatchedArrayShapes {
            a_arr_name: "data",
            a_shape: data.shape().to_vec(),
            b_arr_name: "labels",
            b_shape: labels.shape().to_vec(),
        });
    }
    let data = data.into_dyn();
    let rcm = roi_cloud_map(labels, threads);
    let mut rdm: HashMap<u64, Array1<T>> = HashMap::new();
    rcm.iter().for_each(|(&k, c)| {
        let cloud_lns = c.lanes(Axis(1));
        let cloud_data = cloud_lns
            .into_iter()
            .map(|l| match l.as_slice() {
                Some(s) => *data.get(s).unwrap(),
                None => {
                    let coords = l.to_vec();
                    *data.get(coords.as_slice()).unwrap()
                }
            })
            .collect::<Vec<T>>();
        rdm.insert(k, Array1::from_vec(cloud_data));
    });
    Ok(rdm)
}