| Title |
Vaizdo kontūrų nustatymo būdų analizė / |
| Translation of Title |
Analysis of methods for image edge detection. |
| Authors |
Laskauskas, Ramūnas ; Vyšniauskas, Vytautas |
| Full Text |
|
| Is Part of |
Jaunųjų mokslininkų darbai. 2008, Nr. 5, p. 21-26.. ISSN 1648-8776 |
| Keywords [eng] |
image processing ; contour ; edge ; threshold ; Cany ; Sobel ; Roberts ; Perwitt ; Zerocross ; Laplasian ; LoG and Marr–Hildreth |
| Abstract [eng] |
Edge detection is used in image processing when color, texture or other small components of image are not significant but location and collocation of essential objects in the image are important. Usage of edge against to full image processing saves processing time and requires less computation and hardware recourses. The results of analysis of 8 edge detection methods (Cany, Sobel, Roberts, Perwitt, Zerocross, Laplasian, LoG and Marr–Hildreth) that are used in Matlab are introduced here. 100 images with different scenes and size of objects are used to find a usable range of threshold and best threshold value statistically. The threshold for best image edge is near 100% of image brightness and occupies a few percent or a part of percent. The methods’ usability depends on image quality. There are no relations between image scene and edge detection method used. Research results are usable to save time and recourses of optimal threshold for edge detection when these methods are used. |
| Type |
Journal article |
| Language |
Lithuanian |
| Publication date |
2008 |
