12<small> <a href="index.html#libs">Back to list of all libraries</a> | <a href="modules.html" >See all modules</a> </small>
13<br>
14
15<h2>Library <code>mrpt-base</code></h2>
16<hr>
17
18This is the most fundamental library in MRPT, since it provides a vast amount of utilities and OS-abstraction classes upon which
19the rest of MRPT is built. Here resides critical functionality such as mathematics, linear algebra, serialization, smart pointers
20and multi-threading.
21l
22This library comprises classes in a number of namespaces, briefly described below (click on the namespaces names to see the
23complete list of its classes):
24
25
26<h3>mrpt::poses</h3>
27
28A comprehensive collection of geometry-related classes to represent all kind of 2D and 3D geomtry transformations in different formats
29(Euler angles, rotation matrices, quaternions), as well as networks of pose constrains (as used typically in SLAM problems).
30
31There are also implemented representations for probability distributions over all of these transformations, in a generic way that
32allow mono and multi-modal Gaussians and particle-based representations.
33
34See mrpt::poses for the complete list of classes here.
35
36
37<h3>mrpt::utils</h3>
38
39<ul>
40<li><b>RTTI (RunTime Type Information):</b> A cross-platform, compiler-independent RTTI system is built around the base class
41mrpt::utils::CObject.</li>
42
43<li><b>Smart pointers:</b> Based on the std::shared_ptr<>, any class CFoo inheriting from CObject, automatically has an associated smart
44pointer class CFooPtr. MRPT implements advanced smart pointers capable of multi-thread safe usage and smart pointer typecasting with
45runtime check for correct castings (<a href="http://www.mrpt.org/Smart_pointers" >tutorial</a>).</li>
46
47<li><b>Image handling:</b> The class mrpt::utils::CImage represents a wrapper around OpenCV IplImage's, plus extra functionality
48such as on-the-fly loading of images stored in disk upon first usage. The internal IplImage is always available so OpenCV's functions can
49be still used to operate on MRPT images. </li>
50
51<li><b>Serialization/Persistence:</b> Object serialization in a simple but powerful (including versioning) format
52is supported by dozens of MRPT classes, all based on mrpt::utils::CSerializable. </li>
53
54<li><b>Streams:</b> Stream classes (see the base mrpt::utils::CStream) allow serialization of MRPT objects. There are classes
55for tranparent GZ-compressed files, sockets, serial ports, etc. </li>
56
57<li><b>XML-based databases:</b> Simple databases can be mantained, loaded and saved to files with mrpt::utils::CSimpleDatabase. </li>
58
59<li><b>Name-based argument passing:</b> See the structure mrpt::utils::TParameters </li>
60
61<li><b>Configuration files:</b> There is one base virtual class (mrpt::utils::CConfigFileBase) which can be used to read/write configuration
62files (including basic types, vectors, matrices,...) from any "configuration source" transparently (an actual configuration file, a text block
63created on the fly, etc.). </li>
64
65
66</ul>
67
68<h3>mrpt::math</h3>
69
70MRPT defines a number of generic <i> math containers</i>, which are:
71
72<ul>
73<li><b>Matrices:</b> Dynamic-size matrices (see mrpt::math::CMatrixDouble) and compile-time fixed-size matrices (see mrpt::math::CMatrixFixedNumeric, mrpt::math::CMatrixDouble33, etc.). </li>
74<li><b>Arrays:</b> Fixed-size vectors, just like plain C arrays but with support for STL-like iterators and much more. See mrpt::math::CArrayDouble<>. </li>
75</ul>
76
77For a more in-depth description of these types, and their relation to the base Eigen classes,