The LTI-Lib is an object oriented library with algorithms and data structures frequently used in image processing and computer vision.
LTI-Lib has been developed at the Chair of Technical Computer Science (Lehrstuhl fuer Technische Informatik) LTI at the Aachen University of Technology, as part of many research projects in computer vision dealing with robotics, object recognition and sin
The main goal of the LTI-Lib is to provide an object oriented library in C++, which simplifies the code sharing and maintenance, but still providing fast algorithms that can be used in real applications.
It has been developed using GCC under Linux, and Visual C++ under Windows NT. We have not tested it under other platforms.
Many classes encapsulate Windows/Linux functionality in order to simplify dealing with system or hardware specific code (for example classes for multi-threading and synchronization, time measurement and serial port access).
The rest of the more than 300 classes deal mainly with one of following fields:
Matrices, Vectors, Tensors, and functors to extract eigenvalues, eigenvectors, linear equations solutions, statistics, etc. are provided.
Classification and Clustering
Radial Basis Function classifiers, Support Vector Machines, k-Means, Fuzzy C-Means, classification statistics are just some examples of what you can do with the LTI-Lib.
The most classes deal with image processing problems. Different segmentation approaches, linear filters, wavelets, steerable filters, und much more are already available.
Visualization and Drawing Tools
The most difficult part when developing image processing algorithms in C++ is showing temporary images while debugging. Due to the object oriented architecture of the LTI-Lib, you just need to create a viewer object and give it the image you need to show. That's it. An if you additionally need to draw some extra information on that image (some text, ellipses, boxes, lines or points) you can use one of the drawing objects. This will help you to save lots of time!
· GCC 2.95.3
What's New in This Release:
· examples/shapeRecognition now saves the network anyway and returns also the message from the network if training failed. That helps in finding the reasons for failure.
· dilation/erosion failed on channels outside norm interval [0;1] when using Gray mode.
· the vector/matrix/array classes had a problem with resize(). Now they return to the old functionality that leaves the data alone if the size is not actually changed. The documentation was updated accordingly. The previous version which was compliant with the old documentation fails for a quite common usage case of matrix::getRow() See threads for details: http://sourceforge.net/mailarchive/ forum.php?thread_id=9028254&forum_id=46984 http://sourceforge.net/ mailarchive/forum.php?thread_id=9047176&forum_id=46984
· Marco Wilka has contributed patches for the system dependent classes that let the LTI-Lib compile under Mac OS X, xcode2.2 (gcc4.0.1). Great job, thanks.