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
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* NVIDIA Corporation and its licensors retain all intellectual
* property and proprietary rights in and to this software and
* related documentation and any modifications thereto.
* Any use, reproduction, disclosure, or distribution of this
* software and related documentation without an express license
* agreement from NVIDIA Corporation is strictly prohibited.
*/
#include <math.h>
#include "TestIntegralImage.h"
template <class T_in, class T_out>
TestIntegralImage<T_in, T_out>::TestIntegralImage(std::string testName_, NCVTestSourceProvider<T_in> &src_,
Ncv32u width_, Ncv32u height_)
:
NCVTestProvider(testName_),
src(src_),
width(width_),
height(height_)
{
}
template <class T_in, class T_out>
bool TestIntegralImage<T_in, T_out>::toString(std::ofstream &strOut)
{
strOut << "sizeof(T_in)=" << sizeof(T_in) << std::endl;
strOut << "sizeof(T_out)=" << sizeof(T_out) << std::endl;
strOut << "width=" << width << std::endl;
strOut << "height=" << height << std::endl;
return true;
}
template <class T_in, class T_out>
bool TestIntegralImage<T_in, T_out>::init()
{
return true;
}
template <class T_in, class T_out>
bool TestIntegralImage<T_in, T_out>::process()
{
NCVStatus ncvStat;
bool rcode = false;
Ncv32u widthII = this->width + 1;
Ncv32u heightII = this->height + 1;
NCVMatrixAlloc<T_in> d_img(*this->allocatorGPU.get(), this->width, this->height);
ncvAssertReturn(d_img.isMemAllocated(), false);
NCVMatrixAlloc<T_in> h_img(*this->allocatorCPU.get(), this->width, this->height);
ncvAssertReturn(h_img.isMemAllocated(), false);
NCVMatrixAlloc<T_out> d_imgII(*this->allocatorGPU.get(), widthII, heightII);
ncvAssertReturn(d_imgII.isMemAllocated(), false);
NCVMatrixAlloc<T_out> h_imgII(*this->allocatorCPU.get(), widthII, heightII);
ncvAssertReturn(h_imgII.isMemAllocated(), false);
NCVMatrixAlloc<T_out> h_imgII_d(*this->allocatorCPU.get(), widthII, heightII);
ncvAssertReturn(h_imgII_d.isMemAllocated(), false);
Ncv32u bufSize;
if (sizeof(T_in) == sizeof(Ncv8u))
{
ncvStat = nppiStIntegralGetSize_8u32u(NcvSize32u(this->width, this->height), &bufSize, this->devProp);
ncvAssertReturn(NPPST_SUCCESS == ncvStat, false);
}
else if (sizeof(T_in) == sizeof(Ncv32f))
{
ncvStat = nppiStIntegralGetSize_32f32f(NcvSize32u(this->width, this->height), &bufSize, this->devProp);
ncvAssertReturn(NPPST_SUCCESS == ncvStat, false);
}
else
{
ncvAssertPrintReturn(false, "Incorrect integral image test instance", false);
}
NCVVectorAlloc<Ncv8u> d_tmpBuf(*this->allocatorGPU.get(), bufSize);
ncvAssertReturn(d_tmpBuf.isMemAllocated(), false);
NCV_SET_SKIP_COND(this->allocatorGPU.get()->isCounting());
NCV_SKIP_COND_BEGIN
ncvAssertReturn(this->src.fill(h_img), false);
ncvStat = h_img.copySolid(d_img, 0);
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
if (sizeof(T_in) == sizeof(Ncv8u))
{
ncvStat = nppiStIntegral_8u32u_C1R((Ncv8u *)d_img.ptr(), d_img.pitch(),
(Ncv32u *)d_imgII.ptr(), d_imgII.pitch(),
NcvSize32u(this->width, this->height),
d_tmpBuf.ptr(), bufSize, this->devProp);
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
}
else if (sizeof(T_in) == sizeof(Ncv32f))
{
ncvStat = nppiStIntegral_32f32f_C1R((Ncv32f *)d_img.ptr(), d_img.pitch(),
(Ncv32f *)d_imgII.ptr(), d_imgII.pitch(),
NcvSize32u(this->width, this->height),
d_tmpBuf.ptr(), bufSize, this->devProp);
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
}
else
{
ncvAssertPrintReturn(false, "Incorrect integral image test instance", false);
}
ncvStat = d_imgII.copySolid(h_imgII_d, 0);
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
if (sizeof(T_in) == sizeof(Ncv8u))
{
ncvStat = nppiStIntegral_8u32u_C1R_host((Ncv8u *)h_img.ptr(), h_img.pitch(),
(Ncv32u *)h_imgII.ptr(), h_imgII.pitch(),
NcvSize32u(this->width, this->height));
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
}
else if (sizeof(T_in) == sizeof(Ncv32f))
{
ncvStat = nppiStIntegral_32f32f_C1R_host((Ncv32f *)h_img.ptr(), h_img.pitch(),
(Ncv32f *)h_imgII.ptr(), h_imgII.pitch(),
NcvSize32u(this->width, this->height));
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
}
else
{
ncvAssertPrintReturn(false, "Incorrect integral image test instance", false);
}
NCV_SKIP_COND_END
//bit-to-bit check
bool bLoopVirgin = true;
NCV_SKIP_COND_BEGIN
for (Ncv32u i=0; bLoopVirgin && i < h_img.height() + 1; i++)
{
for (Ncv32u j=0; bLoopVirgin && j < h_img.width() + 1; j++)
{
if (sizeof(T_in) == sizeof(Ncv8u))
{
if (h_imgII.ptr()[h_imgII.stride()*i+j] != h_imgII_d.ptr()[h_imgII_d.stride()*i+j])
{
bLoopVirgin = false;
}
}
else if (sizeof(T_in) == sizeof(Ncv32f))
{
if (fabsf((float)h_imgII.ptr()[h_imgII.stride()*i+j] - (float)h_imgII_d.ptr()[h_imgII_d.stride()*i+j]) > 0.01f)
{
bLoopVirgin = false;
}
}
else
{
ncvAssertPrintReturn(false, "Incorrect integral image test instance", false);
}
}
}
NCV_SKIP_COND_END
if (bLoopVirgin)
{
rcode = true;
}
return rcode;
}
template <class T_in, class T_out>
bool TestIntegralImage<T_in, T_out>::deinit()
{
return true;
}
template class TestIntegralImage<Ncv8u, Ncv32u>;
template class TestIntegralImage<Ncv32f, Ncv32f>;