* unit tests
* SIMD support
* more wavelets (CDF 5/3, Haar)
* more data types (int)
* performance measurement
* more example applications
* more effective lifting implementation (take a look at another implementations, e.g. OpenJPEG or JasPer)
* support for full decomposition in 1D
* relicense to GNU LGPL
* OpenCV wrapper (C)
* C++ interface
* drop deprecated interface
* remove duplicate code
* use Par4All tool
* tiling (e.g., as in JPEG 2000)
* remove "lib" prefix pro source file names
* group lifting steps into one for-loop (suitable for SIMD implementation and UTIA EdkDSP platform)
* 2D DWT: single-loop approach (iteration of horizontal part is immediately fed into iteration of vertical part)
* memory access through incremented pointers instead of indexed arrays
* cache optimizations (stride should be prime?)
* use size_t, ssize_t, etc. instead of unsigned int, int, ...
