Image Segmentation Hyeun-gu Choi

---

Abstract     Image segmentation is a very important step in image processing. Extracting  useful  information from an image is the goal of image segmentation.
    In this paper, two main segmentation methods (Region growing and Contour following) are tested in order to research the advantages, disadvantages and characteristics. Tested algorithms are four neighbor region growing algorithm, gradient operators, histogram based thresholding,  thresholding based on boundary characteristics and Hough transform.
    The performance of each algorithm was evaluated by using real MRI images, remote sensing images and synthetic images. The synthetic images were used to validate the algorithms and to illustrate difficulties that are encountered with real images. The test environment was constructed as an IDL(5) widget (graphic user interface).
 

 

Table of Contents