laplace input output
input
List of images to be convolved.
output
List of output images. The number of output images must equal the number of
input images. If the input image name equals the output image name the
convolved image will replace the input image.
laplace = xycentral
The Laplacian filters are a set of four three by three kernels which
approximate the Laplacian operator, where a Laplacian operator is defined
as the sum of the partial second derivatives in x and y.
The elements of the four Laplacian kernels are shown in detail below.
xycentral
The elements of the central column and row of a 3 by 3 image subraster are
combined to estimate the Laplacian at the position of the central pixel.
diagonals
The elements of the two diagonals of a 3 by 3 image subraster are combined
to estimate the Laplacian at the position of the central pixel.
xyall
The three columns and rows of a three by three image subraster are averaged
to estimate the Laplacian at the position of the central pixel.
xydiagonals
The central row and column and the two diagonals of a three by three image
subraster are combined to estimate the Laplacian at the position of the
central pixel.
boundary = nearest
The algorithm used to compute the values of the out of bounds pixels.
The options are:
nearest
Use the value of the nearest boundary pixel.
constant
Use a constant value.
reflect
Generate a value by reflecting around the boundary.
wrap
Generate a value by wrapping around to the opposite side of the image.
constant = 0.
The constant for constant-valued boundary extension.
LAPLACE convolves the list of images specified by input with one of four 3 by 3 Laplacian kernels specified by laplace and places the convolved images in output. If the image names in output equal the image names in input the Laplacian operation is performed in place and the original images are overwritten. Out of bounds pixels are computed using the algorithm specified by boundary.
The Laplacian filters are high-pass filters which act as a local edge detector. A characteristic of the Laplacian is that it is zero at points where the gradient is a maximum or a minimum. Therefore points detected as gradient edges would generally not be detected as edge points with the Laplacian filter. Another characteristic of Laplacian operators is that a single grey level transition may produce two distinct peaks one positive and one negative in the Laplacian which may be offset from the gradient location.
The four Laplacian filters are listed below. The I[*,*] are the elements of the input image and the O[*,*] are the elements of the output image.
xycenter
0*I[-1,1] + 1*I[0,1] + 0*I[1,1] +
O[0,0] = 1*I[-1,0] - 4*I[0,0] + 1*I[1,0] +
0*I[-1,-1] + 1*I[0,-1] + 0*I[1,-1]
diagonals
I[-1,1]/sqrt(2) + I[0,1]*0 + I[1,1]/sqrt(2) +
O[0,0] = I[-1,0]*0 - I[0,0]*4/sqrt(2) + I[1,0]*0 +
I[-1,-1]/sqrt(2) + I[0,-1]*0 + I[1,-1]/sqrt(2)
xyall
2/3*I[-1,1] - 1/3*I[0,1] + 2/3*I[1,1] +
O[0,0] = - 1/3*I[-1,0] - 4/3*I[0,0] - 1/3*I[1,0] +
2/3*I[-1,-1] - 1/3*I[0,-1] + 2/3*I[1,-1]
xydiagonals
I[-1,1]/sqrt(2)/2 + I[0,1]/2 + I[1,1]/sqrt(2)/2 +
O[0,0] = I[-1,0]/2 - I[0,0]*(2-sqrt(2)) + I[1,0]/2 +
I[-1,-1]/sqrt(2)/2 + I[0,-1]/2 + I[1,-1]/sqrt(2)
1. Convolve an image with the Laplacian filter xyall using nearest neighbour boundary extension.
cl> laplace m83 m83.lap xyall
LAPLACE requires approximately 1.7 cpu seconds to convolve a 512 square real image with a 3 by 3 Laplacian kernel on a Sparc Station 1.
convolve, gauss, gradient, boxcar,