blob: 104bb30ed7a620477e0ff47c642c9c91c017b86d (
plain)
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
|
#!/usr/bin/env python
""" """
__author__ = "Théo de la Hogue"
__credits__ = []
__copyright__ = "Copyright 2023, Ecole Nationale de l'Aviation Civile (ENAC)"
__license__ = "BSD"
from typing import TypeVar, Tuple, Any
from dataclasses import dataclass, field
from argaze import GazeFeatures
import numpy
from scipy.spatial.distance import cdist
@dataclass
class ScanPathAnalyzer(GazeFeatures.ScanPathAnalyzer):
"""Implementation of Nearest Neighbor Index (NNI) as described in Di Nocera et al., 2006
"""
def __post_init__(self):
pass
def analyze(self, scan_path: GazeFeatures.ScanPathType, screen_dimension: tuple[float, float]) -> float:
"""Analyze scan path."""
assert(len(scan_path) > 1)
# Gather fixations focus points
fixations_focus = []
for step in scan_path:
fixations_focus.append(step.first_fixation.focus)
# Compute inter fixation distances
distances = cdist(fixations_focus, fixations_focus)
# Find minimal distances between each fixations
minimums = numpy.apply_along_axis(lambda row: numpy.min(row[numpy.nonzero(row)]), 1, distances)
# Average of minimun distances
dNN = numpy.sum(minimums / len(fixations_focus))
# Mean random distance
dran = 0.5 * numpy.sqrt(screen_dimension[0] * screen_dimension[1] / len(fixations_focus))
return dNN / dran
|
