PERBANDINGAN STRATEGI PELABELAN OBJEK PADA CITRA DIGITAL DENGAN METODE FLOOD FILLING

Abstract

One type of digital image processing is labeling the object to determine the number of objects in an image to be reviewed. Labeling the object in the image is the act to give a sign of the number of different objects on each object in the image. Labeling process can use the method of flooding the object region (flood filling) with an object number. The output of this research is a software application that can be used to perform simulations labeling objects in digital images by using (1) Recursive techniques; (2) approach the Depth First Search (DFS) and (3) the approach Breadth First Search (BFS). Waterfall model used for the development of such software. Based on tests performed on the entire software application can be concluded that the recursive method is the best method to label objects in a digital image, both in terms of speed and accuracy. Labeling objects using Recursive technique has an accuracy of counting the number of objects by 99%, while 68% DFS and BFS 83%. All methods have a level of accuracy that is better at recognizing all areas / regions of the object, but to the image of the complex and in large numbers, algorithms Stack (approach DFS) or Queue (approach BFS) has a weakness, namely (1) to repeat the numbers the same label for different objects and (2) the number of labels sometimes inconsistent for the same object.Keywords: image processing, recursive techniques, depth first search, breadth first search