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
|
#ifndef NORMALIZE_PASSBAND_H
#define NORMALIZE_PASSBAND_H
#include "../algorithms/thresholdtools.h"
#include <aocommon/uvector.h>
#include <utility>
#include <vector>
#ifdef __SSE__
#define USE_INTRINSICS
#endif
#ifdef USE_INTRINSICS
#include <xmmintrin.h>
#endif
namespace algorithms {
class NormalizeBandpass {
public:
static void NormalizeStepwise(TimeFrequencyData& data, size_t steps) {
const size_t height = data.ImageHeight();
aocommon::UVector<num_t> stddev(steps);
for (size_t step = 0; step != steps; ++step) {
const size_t startY = step * height / steps,
endY = (step + 1) * height / steps;
std::vector<num_t> dataVector((1 + endY - startY) * data.ImageWidth() *
data.ImageCount());
std::vector<num_t>::iterator vecIter = dataVector.begin();
const Mask2DCPtr mask = data.GetSingleMask();
for (size_t i = 0; i != data.ImageCount(); ++i) {
const Image2D& image = *data.GetImage(i);
for (size_t y = startY; y != endY; ++y) {
const num_t* inputPtr = image.ValuePtr(0, y);
const bool* maskPtr = mask->ValuePtr(0, y);
for (size_t x = 0; x != image.Width(); ++x) {
if (!*maskPtr && std::isfinite(*inputPtr)) {
*vecIter = *inputPtr;
++vecIter;
}
++inputPtr;
++maskPtr;
}
}
}
dataVector.resize(vecIter - dataVector.begin());
num_t mean;
ThresholdTools::WinsorizedMeanAndStdDev<num_t>(dataVector, mean,
stddev[step]);
}
for (size_t i = 0; i != data.ImageCount(); ++i) {
const Image2D& image = *data.GetImage(i);
Image2DPtr destImage =
Image2D::CreateUnsetImagePtr(image.Width(), image.Height());
for (size_t step = 0; step != steps; ++step) {
const size_t startY = step * height / steps,
endY = (step + 1) * height / steps;
float correctionFactor;
if (stddev[step] == 0.0)
correctionFactor = 0.0;
else
correctionFactor = 1.0 / stddev[step];
#ifdef USE_INTRINSICS
const __m128 corrFact4 = _mm_set_ps(correctionFactor, correctionFactor,
correctionFactor, correctionFactor);
#endif
for (size_t y = startY; y != endY; ++y) {
const float* inputPtr = image.ValuePtr(0, y);
float* destPtr = destImage->ValuePtr(0, y);
#ifdef USE_INTRINSICS
for (size_t x = 0; x < image.Width(); x += 4) {
_mm_store_ps(destPtr, _mm_mul_ps(corrFact4, _mm_load_ps(inputPtr)));
inputPtr += 4;
destPtr += 4;
}
#else
for (size_t x = 0; x < image.Width(); x++) {
*destPtr = correctionFactor * *inputPtr;
inputPtr++;
destPtr++;
}
#endif
}
}
data.SetImage(i, std::move(destImage));
}
}
static void NormalizeSmooth(TimeFrequencyData& data) {
TimeFrequencyData real = data.Make(TimeFrequencyData::RealPart),
imag = data.Make(TimeFrequencyData::ImaginaryPart);
const size_t height = data.ImageHeight();
aocommon::UVector<double> rms(height);
std::vector<std::complex<num_t>> dataVector;
for (size_t y = 0; y != height; ++y) {
dataVector.resize(data.ImageWidth() * data.ImageCount());
auto vecIter = dataVector.begin();
const Mask2DCPtr mask = data.GetSingleMask();
for (size_t i = 0; i != real.ImageCount(); ++i) {
const Image2D& realImage = *real.GetImage(i);
const Image2D& imagImage = *imag.GetImage(i);
const num_t* realPtr = realImage.ValuePtr(0, y);
const num_t* imagPtr = imagImage.ValuePtr(0, y);
const bool* maskPtr = mask->ValuePtr(0, y);
for (size_t x = 0; x != realImage.Width(); ++x) {
if (!*maskPtr && std::isfinite(*realPtr) && std::isfinite(*imagPtr)) {
*vecIter = std::complex<num_t>(*realPtr, *imagPtr);
++vecIter;
}
++realPtr;
++imagPtr;
++maskPtr;
}
}
dataVector.resize(vecIter - dataVector.begin());
rms[y] = ThresholdTools::WinsorizedRMS<num_t>(dataVector);
}
for (size_t i = 0; i != data.ImageCount(); ++i) {
const Image2D& image = *data.GetImage(i);
Image2DPtr destImage =
Image2D::CreateUnsetImagePtr(image.Width(), image.Height());
for (size_t y = 0; y != height; ++y) {
num_t correctionFactor;
if (rms[y] == 0.0)
correctionFactor = 0.0;
else
correctionFactor = 1.0 / rms[y];
const num_t* inputPtr = image.ValuePtr(0, y);
num_t* destPtr = destImage->ValuePtr(0, y);
for (size_t x = 0; x < image.Width(); x++) {
*destPtr = correctionFactor * *inputPtr;
++inputPtr;
++destPtr;
}
}
data.SetImage(i, std::move(destImage));
}
}
};
} // namespace algorithms
#endif
|
