2019/01/08 by Fabrício Breve, Breve, Fabricio Aparecido
Computer Science · #Advanced Image and Video Retrieval Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Medical Image Segmentation Techniques #Visual Attention and Saliency Detection
paper · pdf · doi:10.48550/arxiv.1901.02573
openalex publication_date 2019/01/08 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Interactive image segmentation is a topic of many studies in image\nprocessing. In a conventional approach, a user marks some pixels of the\nobject(s) of interest and background, and an algorithm propagates these labels\nto the rest of the image. This paper presents a new graph-based method for\ninteractive segmentation with two stages. In the first stage, nodes\nrepresenting pixels are connected to their k-nearest neighbors to build a\ncomplex network with the small-world property to propagate the labels quickly.\nIn the second stage, a regular network in a grid format is used to refine the\nsegmentation on the object borders. Despite its simplicity, the proposed method\ncan perform the task with high accuracy. Computer simulations are performed\nusing some real-world images to show its effectiveness in both two-classes and\nmulti-classes problems. It is also applied to all the images from the Microsoft\nGrabCut dataset for comparison, and the segmentation accuracy is comparable to\nthose achieved by some state-of-the-art methods, while it is faster than them.\nIn particular, it outperforms some recent approaches when the user input is\ncomposed only by a few "scribbles" draw over the objects. Its computational\ncomplexity is only linear on the image size at the best-case scenario and\nlinearithmic in the worst case.\n