IPedge

Synopsis

enhance edges in an ip_Image

Input Port

&in_image	ip_Image
&in_roi	ip_Roi

Parameters

MethodsRB	UIradioBox	select the edge detection technique
Filter Width	slider	set width of convolution kernels
Filter Height	slider	set height of convolution kernels

Output Port

dst	ip_Image_Out
TmpField	IPfld
obj	output renderable object

Description

IPedge performs an edge enhancement operation using the algorithm specified.

The algorithms use convolution kernels to sharpen an image in the horizontal direction and then in the vertical direction. The algorithms then perform a quadratic add on the resulting images. Images are converted to floating point before performing the convolutions.

Input

&in_image

The input is a reference to an ip_Image.

&in_roi

ip_Roi. An optional region of interest. The data must be of type byte.

Parameters

MethodsRB

UIradioBox. Selects which edge detection technique to perform. The choices are:

Prewiit

Roberts

Compass

Frei Chan

Marr Hildreth (variable kernel)

Nevatia Babu

Robinson 3

Robinson 5

Macleod (variable kernel)

Argyle (variable kernel)

Kirsh

Boxcar

Sobel

Derivative of Gaussian (variable kernel)

Weighted Line

Unweighted Line

Filter Width
Filter Height

float sliders. Some of the algorithms, such as the Argyle, Macleod, Marr Hildreth, and the Derivative of Gaussian use variable width kernels. Filter Width and Filter Height set the width and height of these variable width kernels. The sliders only appear for the selected method as appropriate.

The default for both is 3.0. Their range is 0.1 to 10.0.

The variables specify the functional size of the kernel, not the actual size. A particular algorithm generates the actual kernel size from these values. A variable width kernel is useful because you can make the width smaller to detect smaller detail, or larger to ignore noisy edges in an image.

Be aware that you can supply widths that will produce large kernels, which will require large amounts of processing time. In these cases, you may find that you can perform an edge enhancement operation faster if you first perform a fourier transform on the image.

Output Port

dst

The output is a new ip_Image of the same type and number of bands as the input ip_Image.

TmpField

This output is the ip_Image converted back into an AVS/Express field.

obj

This is a renderable version of the output.

Example

Libraries.Examples.Image_Processing.IPedge

File

v/ip.v

See also

Not applicable.