1 % MEX_RLOF is a wrapper to the RLOF library that contains a set of sparse
2 % motion estimation methods based on the local optical flow paradigm.
6 % POINTLIST2 = mex_RLOF(IMAGE1, IMAGE1, POINTLIST1, PARAMSTRUCT);
8 % Arguments IMAGE1 and IMAGE1 could be gray scaled images (2-dimensional
9 % arrays) or rgb images (3-dimensional arrays)
10 % They are required to have the same dimension and they must be doubles.
12 % POINTLIST1 is the list of points to track (
double 2xn array).
13 % With n the number of points in the point list.
15 % POINTLIST2 is the list of tracked points corresponding to
"pointlist"
18 % PARAMSTRUCT specifies one or more of the following name/value pairs. Only
19 % the
'method' is mandatory.
22 %
'method' A
string specifying the applied motion estimation
23 % method. The value can be one of the following:
24 %
'PLK' - Pyramidal Lucas Kanade provided by
26 %
'BEPLK' - Pyramidal Lucas Kanade
using
27 % bilinear equations, see
29 %
'CB_PLK' - Pyramidal Lucas Kanade with
30 % adaptive support region, see
32 %
'CB_BEPLK' - Pyramidal Lucas Kanade
using
33 % bilinear equations and an adaptive
37 %
'RLOF' - Robust Local Optical Flow, see
38 % Senst et al. 2011/2012.
39 %
'BERLOF' - Robust Local Optical Flow
using
40 % bilinear equations, see
42 %
'CB_RLOF' - Robust Local Optical Flow with
43 % adaptive support region, see
45 %
'CB_BERLOF' - Robust Local Optical Flow
using
46 % bilinear equations and an adaptive
48 % Senst et al. 2014. Full paper
49 % references are given below.
51 %
'noLevels' A positive number specifying the number of
52 % levels used
for the gaussian image pyramid.
54 %
'noIterations' A positive number specifying the number of
55 % levels used
for the gaussian image pyramid.
57 %
'winSize' A positive number specifying the size of the
58 % support region. Using RLOF or CB derivates
59 % winSize specify the upper bound of the support
62 %
'smallWinSize' Used only
for RLOF and CB derivates. A positive
63 % number specifying the lower bound of the support
66 %
'CBThreshold' Used only
for CB derivates. A positive number
67 % specifying the pixel value threshold \tau, see
70 %
'norm' A vector [V1 V2] specifying the shrinked Hampel
71 % norm \sigma_1 and \sigma_2, see
74 %
'options' A
string specifying the debug options. The value
75 % can be one of the following:
76 %
'PerformTest' - A accuracy check with pre-
77 % estimated motion vectors
for
78 % the ErnstReuter image pair
80 %
'PrintParameter' - Displays the parameter of the
81 % respective algorithm.
88 % This file is the property of the author and Communication Systems Group,
89 % Technische Universität Berlin. All rights reserved.
91 % It may not be publicly disclosed, distributed, used, copied or modified
92 % without prior written authorization by a representative of
93 % Communication Systems Group, Technische Universität Berlin or the author.
94 % Any modified version of
this document needs to contain
this header.
97 % PERSONAL, NON-COMMERCIAL or ACADEMIC USAGE:
98 % You are free to use
this software
for whatever you like. If you use
this
99 % algorithm
for a scientific publication, please cite the one of the
102 % @INPROCEEDINGS{ICIPSenst2014,
103 % AUTHOR = {Tobias Senst and Thilo Borgmann and Ivo Keller and Thomas
105 % TITLE = {Cross based Robust Local Optical Flow},
106 % BOOKTITLE = {21th IEEE International Conference on Image Processing},
109 % PAGES = {1967--1971},
110 % ADDRESS = {Paris, France},
113 % @INPROCEEDINGS{ICIPSenst2013,
114 % AUTHOR = {Tobias Senst and Jonas Geistert and Ivo Keller and Thomas
116 % TITLE = {Robust Local Optical Flow Estimation
using Bilinear Equations
117 %
for Sparse Motion Estimation},
118 % BOOKTITLE = {20th IEEE International Conference on Image Processing},
121 % PAGES = {2499--2503},
122 % ADDRESS = {Melbourne, Australia},
123 % DOI = {10.1109/ICIP.2013.6738515},
126 % @ARTICLE{TCSVTSenst2012,
127 % AUTHOR = {Tobias Senst and Volker Eiselein and Thomas Sikora},
128 % TITLE = {Robust Local Optical Flow
for Feature Tracking},
129 % JOURNAL = {IEEE Transactions on Circuits and Systems
for Video
133 % PAGES = {1377--1387},
136 % DOI = {10.1109/TCSVT.2012.2202070}
139 % @INPROCEEDINGS{WACVSenst2011,
140 % AUTHOR = {Tobias Senst and Volker Eiselein and Rubén Heras Evangelio and
141 % and Thomas Sikora},
142 % TITLE = {Robust Modified L2 Local Optical Flow Estimation and Feature
144 % BOOKTITLE = {IEEE Workshop on Motion and Video Computing},
147 % EDITOR = {Eric Mortensen},
148 % PAGES = {685--690},
149 % ADDRESS = {Kona, USA},
150 % DOI = {10.1109/WACV.2011.5711571},
154 % It is not allowed to use any content of
this package for any commercial
155 % use or any advertisement for upcoming commercial products. If you want to
156 % use any content for such a purpose please contact:
157 % Prof. Dr.-Ing. Thomas Sikora <sikora@nue.tu-berlin.de>.
161 % Software provided by Technische Universität Berlin with this document is
162 % provided "AS IS" and any express of implied warranties including, but
163 % not limited to, the implied warranties of merchantability and fitness
164 % for a particular purpose are disclaimed.
165 % In no event shall the author or contributors be liable for any direct,
166 % indirect, incidental, special, exemplary, or consequential damages
167 % (including, but not limited to, procurement of substitute goods or
168 % services, loss of use, data, or profits or business interruption) caused
169 % in any way out of the use of this software, even if advised of the
170 % possibility of such damage.