Astrophysics Pyramid Image Processing (APIP) is a set of multiresolution image processing routines and statistical descriptors developed by Dr. Charles H. Anderson at the Washington University Medical School in collaboration with Dr. William D. Langer at the Jet Propulsion Laboratory, California Institute of Technology.
APIP is different from most other multiresolution software packages in its emphasis on overcomplete non-orthogonal representations. While most scientists are more familiar with the orthonormal wavelets, their strength is in data compression and not in image analysis. Orthonormal wavelets provide critically sampled representations with minimal storage, but the constraint of orthogonality leads to filter designs that are not necessarily the best for image analysis (cf. Strang 1989, Simoncelli et al. 1992).
The multiresolution representations with the most utility for image analysis are the Gaussian and Laplacian Pyramids (Burt and Adelson 1983). More selective multiresolution representations provided in the APIP package are oriented pyramids, gradient pyramids, and orthonormal wavelets. APIP also contains a number of basic statistical measures including, a local measure of the power at different spatial scales, space-scale global energy, intermittency, space-scale contrast, and cluster analysis. Statistics on these clusters can be computed based on properties such as, perimeter, area, intensity, and orientation.
The APIP package provides multiresolution image processing functions, along with several basic image processing functions and display utilities that run in a UNIX X-windows environment. 2D clustering utilities and the calculation of the basic statistics are included in APIP. The functions operate on image files in FITS formats and a more generic format utilized in the IAS group at JPL. Limited support is also provided for conversion of 8 bit TIFF and PostScript image files. An IDL implementation is being developed and should be available by Fall 1999.
APIP consists of a set of 80+ UNIX command line functions with MAN pages. It is available in the form of ANSI C source code with makefiles for Sun4, Solaris, SGI, HP, gcc, Borland, and MS Visual C++ compilers. See the README file for more details on features and installation.
Development of APIP was supported in part by a grant from NASA's Astrophysics Data Program (ADP) and the Information Systems Branch of NASA's Office of Space Science.