| Summary: | 碩士 === 國立中興大學 === 土木工程學系所 === 97 === In practical applications using satellite images or aerial images, we are concerned about the amount of edge information preserved in the images for following-up processes and studies. It has been well developed in using images in manual, semi-automatic, or automatic matching, registration, moasicking, reconstruction, and database search, etc. With respect to the edge extraction from color images, however, it is common in current model that the color images were transformed into grayscale images followed by applying an edge detector. But it is not clear what kind of color transformation is good for preserving the edge information of the original color images.
In this study, the experimental images were transformed into different color models, followed by detecting corresponding edges in space domain and frequency domain, respectively for comparison with the edges extracted from the original color images. Experimental results of this study show that Lu’v’ color transformation preserves the most edge information in space domain while there is no dominant color transformation in frequency domain. It is recommended that in frequency domain the color images may transformed into every color model for finding an optimal representation of the image in follow-up edge extraction.
|
|---|
