Roberts Cross

From Wikipedia, the free encyclopedia

A picture of a steam engine.

The Roberts operator applied to that image.

In computer vision, the Roberts' Cross operator is one of the earliest edge detection algorithms, which works by computing the sum of the squares of the differences between diagonally adjacent pixels. This can be accomplished by convolving the image with two 2x2 kernels:

+1	0
0	-1

0	+1
-1	0

In other words, each pixel in the output image is calculated as follows:

tmp1 = input_image(x,y) - input_image(x+1,y+1)
tmp2 = input_image(x+1,y) - input_image(x,y+1)
output_image(x,y) = absolute_value(tmp1) + absolute_value(tmp2)

Roberts' Cross is still in use due to the speed of computation, but performance compared to the alternatives is poor, with noise sensitivity a significant problem. It produces finer lines than other edge detection methods. It is sometimes called 'Roberts Filter'.