forked from chene77/RobartsICP
-
Notifications
You must be signed in to change notification settings - Fork 0
/
asicp.cpp
293 lines (239 loc) · 7.81 KB
/
asicp.cpp
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
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
/*=========================================================================
Program: Robarts ICP
Module: $RCSfile: asicp.cpp,v $
Creator: Elvis C. S. Chen <chene@robarts.ca>
Language: C++
Author: $Author: Elvis Chen $
Date: $Date: 2014/03/03 13:36:30 $
Version: $Revision: 0.99 $
==========================================================================
This is an open-source copyright license based on BSD 2-Clause License,
see http://opensource.org/licenses/BSD-2-Clause
Copyright (c) 2013, Elvis Chia-Shin Chen
All rights reserved.
Redistribution and use in source and binary forms, with or without modification,
are permitted provided that the following conditions are met:
- Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the documentation and/or
other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
=========================================================================*/
// C++ includes
#include <assert.h>
#include <cmath>
// ANN includes
#include <ANN/ANN.h>
// local includes
#include "matrix.h"
#include "asicp.h"
#include "mathUtils.h"
#include "ASOPP_Major.h"
#include "pointRegistration.h"
void asicp_md( Matrix<double> &points,
Matrix<double> &model,
Matrix<double> &R,
Matrix<double> &A,
Vec<double> &t,
double &FRE, double threshold,
Vec<double> &FREmag )
{
assert( points.num_rows() == 3 );
assert( model.num_rows() == 3 );
int nPoints = points.num_cols();
int nModel = model.num_cols();
// build a KD-tree of the "model"
ANNpointArray data_pts; // data (model) points
ANNpoint query_pt; // query point
ANNidxArray nn_idx; // near neighbour indices
ANNdistArray dists; // near neighbour distances
ANNkd_tree *model_tree=0; // search structure (KD-tree)
int ANNk = 1; // find 1 near neighbor
int ANNdim = 3; // dimension
double ANNeps = 0.0; // exact near neighbor
query_pt = annAllocPt( ANNdim );
data_pts = annAllocPts( nModel, ANNdim );
nn_idx = new ANNidx[ ANNk ];
dists = new ANNdist[ ANNk ];
int nIter = 0;
Matrix<double> r( 3, nPoints ), l( 3, nPoints ), lprime( 3, nPoints );
//
// the ICP loop
//
Matrix<double> residual;
Vec<double> residMag;
double oldFRE = FRE;
int maxIter = 2000;
//
// estimate the scales based on the eigenvalues of the covariance matrices
//
double initScale;
lprime = R * points;
for ( int i = 0; i < points.num_cols(); i++ )
{
for ( int j = 0; j < points.num_rows(); j++ )
{
lprime[j][i] += t[j];
}
}
estimateScalesFromPoints( lprime, model, initScale, R ); // R is given
A = eye( 3, initScale );
std::cerr << "Initial Scales: " << A << std::endl;
Matrix<double> RR(3,3), AA(3,3);
Vec<double> tt(3);
double residuals;
bool changedScales;
lprime = R * A * points;
while ( nIter < maxIter )
{
std::cerr << "i: " << nIter << std::endl;
// loop for a fixed number of times 'cus we may never converge
// in this loop, "l" is the transformed points,
// "r" is the nearest neighbour in the model
l = R * A * points; // scaling followed by rotation
for ( int i = 0; i < nPoints; i++ )
{
for ( int j = 0; j < 3; j++ )
{
l[j][i] += t[j]; // followed by translation
} // j
} // i
// estimate the scales needed to be used in Mahalanobis distance
//
// what we do here is to use the correspondanding fiducials from
// the previous iteration of ICP and register that to the current
// iteration to get an idea of the scaling.
//no
ASMajor_point_register( lprime, l, RR, AA, tt, residuals, threshold, FREmag );
// std::cerr << "i2: " << nIter << std::endl;
lprime = l; // save l
// find the closest point
//
// using the estimates scales from the previous iteration
// for the calculation of Mahalanobis Distance
// closestPoint_with_MahalanobisDistance( l, model, AA, r );
if ( model_tree )
{
delete model_tree;
model_tree = 0;
}
Matrix<double> sqrtAA;
sqrtAA = AA;
for ( int i = 0; i < 3; i++ )
{
sqrtAA[i][i] = sqrt( AA[i][i] );
} // i
//
// copy all the model points into ANN structure
//
// NOTE: since we are interested in Mahalabonis Distance,
// both the data/model points needs to be scaled
for ( int i = 0; i < nModel; i++ )
{
for ( int j = 0; j < 3; j++ )
{
data_pts[i][j] = model[j][i] / sqrtAA[j][j];
} // j
} // i
//
// initialize the KD tree
//
model_tree = new ANNkd_tree( data_pts, nModel, ANNdim );
// find the nearest neighbour of each point in 'l'
for ( int i = 0; i < nPoints; i++ )
{
for ( int j = 0; j < 3; j++ )
{
query_pt[j] = l[j][i] / sqrtAA[j][j];
} // j
model_tree->annkSearch( query_pt, ANNk, nn_idx, dists, ANNeps );
for ( int j = 0; j < 3; j++ )
{
r[j][i] = data_pts[ nn_idx[0] ][j] * sqrtAA[j][j];
}
} // i
// once the correspondances are found, solve for the
// anisotropic-scaled orthogonal procrustes analysis
// using the solution by Dosse and Ten Berge
ASMajor_point_register( points, r,
R, A, t,
FRE, threshold, FREmag );
std::cerr << "Scales: " << A;
// need to regularize scaling factors A here, and
// recompute FRE if necessary.
//
//
// In practice, scaling factors need to be bounded, otherwise
// trivial (and wrong) solution such as scaling~=0 would occure
// and the computed FRE would be minimal.
changedScales = false;
if ( A[0][0] < .9 * initScale )
{
A[0][0] = .9 * initScale;
changedScales = true;
}
if ( A[0][0] > 1.1 * initScale )
{
A[0][0] = 1.1 * initScale;
changedScales = true;
}
if ( A[1][1] < .9 * initScale )
{
A[1][1] = .9 * initScale;
changedScales = true;
}
if ( A[1][1] > 1.1 * initScale )
{
A[1][1] = 1.1 * initScale;
changedScales = true;
}
if ( A[2][2] < .9 * initScale )
{
A[2][2] = .9 * initScale;
changedScales = true;
}
if ( A[2][2] > 1.1 * initScale )
{
A[2][2] = 1.1 * initScale;
changedScales = true;
}
//
// scales has been changed, so re-compute the rotation and FRE
//
if ( changedScales )
{
std::cerr << "Scales changed" << std::endl;
l = A * points;
point_register( l, r, R, t, FRE, threshold, FREmag );
}
if ( oldFRE == FRE )
{
nIter = maxIter + 1;
}
oldFRE = FRE;
if ( FRE <= threshold )
{
nIter = maxIter + 1;
}
nIter++;
} // while
// kd tree cleanup
delete [] nn_idx;
delete [] dists;
if ( model_tree )
{
delete model_tree;
model_tree = 0;
}
annClose();
}