; AND 
hes from 
ction has 
tructures, 
been low 
compared 
ns on the 
hen from 
of edges. 
age taken 
m is that 
c. exhibit 
lilatation. 
enn et al. 
; obtained 
les of the 
els. In an 
change in 
e can be 
sitivity of 
observed 
1s shapes. 
Ie. 
imple for 
rtically. 
image of 
in 5 parts; 
methods, 
ise, edge 
il to the 
the most 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
  
productive methods of finding final edges is to designate the 
immediate changes of grey level. For f (i, j) grey level function, 
the first derivative of the image creates local maximum in edge 
areas. Consequently, edge image is obtained by taking the 
gradient of image and limiting it to an available threshold value. 
Another method is to determine zero possings by taking second 
derivative. 
Derivative methods are "gradient method" that uses first 
derivative and “Laplasyan method” that uses second derivative. 
(Erdon, 1992) 
In this research; Sobel, Robert and Prewitt which are scalar 
edge operators among edge detection systems based on gradient 
methods involved in derivative methods are used. Scalar edge 
operators are not adjusted for any directions and supply 
outcomes that have equal spaces for all directions. For diagonal 
images Robert mask is more effective, but in the overall 
evaluation of Sobel mask forms more effective edge images. 
In this study, broadcasting stratigraphic construction is targeted 
by the use of various edge detection operators of this kind. 
Similarly, the effects of derivative methods on determining 
stratigraphic construction which is one of the edge detection 
techniques are examined. The image used in such applications 
is taken by a non-metric ordinary camera and scanned by 
Scanjet 3c scanner with 300 dpi definition. On the obtained 
digital image; Sobel, Prewitt, Robert and LoG edge detection 
algorithms are applied successively to determine stratigraphic 
construction and the results are interpreted. 
3. STUDY AREA 
The study area is a part of Zonguldak city, located in Western 
Black Sea region of Turkey. It is famous with being one of the 
main hard coal mining field in the world. Area has a rolling 
topography and our stratigraphic construction area is shown in 
Figure 1. 
Figure 1. Study area 
  
127 
4. GEOLOGIC APPLICATIONS USING EDGE 
DETECTION OPERATORS 
4.1 Threshold value 
As a result of the analysis of the sample grey tone values that 
are applied to the image, the value that shows the required 
sharpness as detail below the threshold value are eliminated 
automatically. 
4.2 Dilatation 
The purpose of this process is to clarify by filling the gaps on 
the main lines that set the stratigraphic construction. The reason 
for this process is that the high contrast lines that binary 
gradient pattern demonstrates, do not expose the main lines of 
stratigraphic construction totally. 
That’s why the lines are exaggerated and made visible through 
the process of dilatation. 
4.3 Fill interior 
On the extended stratigraphic image; although the lines become 
as clear as the image allows, some spaces exist in the line. This 
process enables the filling of these spaces. 
These three main process steps are successively applied with 
Sobel, Prewitt, Robert edge detection operators and LoG filter 
on the image. The obtained images do not display the 
stratigraphic construction of the original image could not be 
revealed exactly, however stratification part lines of the specific 
parts of the picture have been determined. When these parts are 
examined, the length differences and contrast changes of the 
original construction prove to be effective in the clarification of 
this line. 
When the original image is examined, stratigraphic line is 
observed to form a horizontal slope of approximately 30°. If the 
construction was horizontal, how would the process of 
revealing this line be with a scalar scanning at the horizontal 
direction at this image? To achieve this aim; the line is brought 
to an approximate horizontal position by spinning the image 
30° clockwise and operators are applied horizontally. 
By the result of horizontal detail removing made on that image, 
vertical details have disappeared; however, a clear image could 
not be obtained because of rock surfaces' being rough on the 
general image. In the direction of these results, as lines' not 
being able to become definite sharply, Medyan filtering method 
was applied to 30? spinned original image by taking the noise 
effect into consideration. 
In the Medyan filtering process that has been done, filter 
dimension was selected as 25*25 and its effect to stratigraphic 
construction was tried to be designated by means of Sobel 
operators. 0,1 grey toning value was used in applying Medyan 
filtering on the image by the result of the investigation that was 
made. 
Edge detection techniques applied in this application was 
utilized in determining the west wing of the North Anatolian 
Fault Zone from TK350 satellite image as a different research.