vedist − euclidean distance transform

vedist [if=] [of=] [-2d] [wx=] [wy=] [wz=] [-f] [-i]

vedist vedist performs weighted euclidean distance transform on a binary image. Each pixel in the output image is encoded with the euclidean distance to the closest non-zero pixel of the input image. The weighted euclidean distance between (x1,y1,z1) and (x2,y2,z2) is defined as: sqrt((wx(x1-x2))^2 +(wy(y1-y2))^2 +(wz(z1-z2))^2), where wx, wy and wz are the coordinate weights.

By default, 3D distance transform is performed. When -2d flag is used, 2D distance transform is performed in individual slice. The default coordinate weights are derived by resolution information of the input image (same as setting wx=1, wy=y_resolution/x_resolution, and wz=z_resolution/x_resolaution). If any coordinate weight is specified (wx = , wy=, or wz=) or -i flag is used, the resolution of the input image will be ignored and unspecified weights will be set to 1.

The algorithm is based on " MAURER, C., QI, R., AND RAGHAVAN, V. 2003. A linear time algorithm for computing the Euclidean distance transform in arbitrary dimensions. IEEE Trans. Patt. Anal. Mach. Intell. 25, 2 (February), 265-270."

if=

input binary image of byte format

of=

short image(by default) or a floating point image (if -f flag is used)

-2d

perform 2D distance transform for each slice

wx=

weight for x coordinate (default 1)

wy=

weight for y coordinate (default 1)

wz=

weight for z coordinate (default 1)

-f

output floating point image instead of short image

-i

ignore resolution information of the image, use wx, wy and wz if they are specified, otherwise use 1 for the weights.

vdist

Shuang Liu, November 2013