Evaluating The Effectiveness Of Region Growing And Edge Detection Segmentation Algorithms

One of the important problems that ever exist in performance evaluation of any segmentation algorithm is that, when we ingrain the obtained results in a specific application, these results may not be expandable to any other application. So, it is very difficult to appraise whether one algorithm produces more precise segmentation than the other one. This paper, presents a novel technique through which the evaluation of the effectiveness of Region Growing and Edge Detection segmentation algorithms is carried out. The proposed evaluation metric is based on the EXOR measure approach, which was originally proposed for the evaluation of skin tumor borders [1]. This performance measure is then extended to a condition where the evaluation of these two image segmentation algorithms can be compared in a suitable and appropriate manner. In order to validate the proposed performance measure, we used 300 images from the publicly available Berkley Segmentation Dataset. These images are classified into seven groups of images, according to the dominant image. The evaluation and comparison results shows that the effectiveness of edge detection segmentation algorithm is better than region growing segmentation algorithm in many applications. [Journal of American Science 2010;6(10):580-587]. (ISSN: 1545-1003).