HOME GUIDE OPERATIONS DOCS ERRORS FORMATS INSTALL NEW TIPS WEB SITES

IP - InterPolate

(04/18/95)

PURPOSE

Interpolates a 2-D picture using bilinear interpolation or a 3-D picture using trilinear interpolation.

SEE ALSO

IP T [interpolation, triangular]
FP [Fourier interpolation]

USAGE

.OPERATION: IP

.INPUT1 FILE: H_PIC001
[Enter the name of the file to be interpolated.]

.OUTPUT FILE: H_IMG024
[Enter the file where the interpolated image is to be kept.]

.ENTER DIMS NSAM,NROW: 128,128
[Enter the dimensions of the interpolated image.]

For 3-D image:

.NSLICE: 79

[PIC001 will then be interpolated to a picture of size 128x128x79, and the interpolated image will be put into INT001.]

NOTES

  1. This operation can be used for either interpolation or extrapolation.

  2. The output image does not have to have the same proportions (ratio NSAM/NROW) as the input picture.

  3. For a rectangular input image, only the new side length (NSAM) of the output image need be entered. The other side length (NROW) will then be computed assuming the same proportions as the input image. For example, let the input image have the size 60x40. Then the sequence .ENTER DIMS NSAM, NROW: 120 will have the same effect as .ENTER DIMS NSAM, NROW: 120,80

  4. For an output picture with dimensions exactly two times smaller than those of an input picture the adding of four neighbouring pixels is done. The resulting picture contains frequencies up to 0.25 (roughly) and has improved SNR (two times). For an output picture with dimensions exactly four times smaller than those of an input picture the adding of sixteen neighbouring pixels is done. Notes 3, and 4 deal with 2-D pictures only.

  5. The memory needed is (NSAM*NROW*NSLICE+NSAM) in MAXMEM.

  6. The program works by throwing a new grid onto the image whose boundaries coincide with those of the old grid. Each pixel of the new grid is computed from its old neighbors by bilinear interpolation or triangular interpolation.

  7. Note that for raw data, a reduction in scale by IP results in a loss of s/n ratio, since only a subset of the original points may be used for interpolation. The proper procedure maintaining the original information as much as possible is by preceding 'IP' with 'BC' (box convolution). The box size used should be comparable with the reduction factor.

SUBROUTINES

INTERP, INTERP3, IRP3

CALLER

UTIL2

© Copyright Notice /       Enquiries: spider@wadsworth.org