Autopano-SIFT is a helper program for panorama image generation. It automatically generates control point information for a number of image files for use with the hugin panorama generation software. To do this, it utilizes the SIFT (Scale-Invariant Feature Transform) algorithm, a sophisticated image feature matching algorithm.
libperceptronnetwork is a library that models a multilayer feedforward perceptron network, a well-known and understood type of neural network. It can be used for a large number of recognition and categorization tasks. Besides providing the standard propagation and backpropagation algorithms, the library also provides variations to the backpropagation algorithm, such as optional weight-decay, momentum-term and optimal-tolerance methods. It comes with a handy programmers reference manual and two examples.
The scale-invariant feature transform is an algorithm to identify and locate interesting points within an image. For all such points, a descriptive signature is extracted. The signatures can be stored and matched among multiple images, allowing for a large number of interesting applications, such as aligning overlapping images and identifying objects or motion within image sequences. libsift is used by the autopano-sift program to create panorama images.
Tuwo is a C++ library for solving high-level computer vision tasks such as image classification and object recognition. Computer vision is a subfield of computer science and deals with systems that can make sense of image information. Building systems suited for high-level computer vision tasks such as object recognition and image classification requires the use of robust image statistics in the form of image features and machine learning algorithms to separate discriminative information from noise. This C++ library provides code to extract image features and learn decision functions from training data. It is suitable for high-level computer vision tasks.
Waveblend is a focus enhancing tool. It implements an image fusion algorithm to combine many images to one, using the images most in focus at each point; the resulting image is in focus everywhere. It uses the complex dual-tree two dimensional discrete wavelet transform algorithm, with a number of pre- and post-processing steps.