Edge Detection

This procedure can also be used as an automated edge detection algorithm. This application can be accomplished by comparing a single image with a slightly shifted version of itself. When this is done the coherent structure is illuminated in the magnitude display of the disparity.

 

edge1
Figure 8 The original image and the boundaries found by image disparity with itself.

Figure  shows the boundaries that have been detected using this algorithm for the first Duri image in figure . This method was also tested on a more complicated migrated data set that contains steeply dipping reflectors, from Tang and Clapp (2006). The result is shown in figure .

 

edge2
Figure 9 The original image and boundaries found by image disparity with itself.

It is encouraging even though this method fails to distinguish all of the layer boundaries that are clearly seen at the top of the original image. Although not demonstrated here, it is possible to shift only in the x-direction to highlight the vertical edges or shift in the y-direction to accentuate the horizontal edges. The sum of these two shifts will yield the same result as a single diagonal shift as done for these examples.