Random walks for image segmentation containing translucent overlapped objects
Image segmentation methods usually separate an image into non-overlapping regions. However, in some data translucent regions partially overlap. Having overlapped regions will cause methods not designed for overlapped objects to perform poorly or not work at all. The random walker (RW) image segmentation algorithm shows interesting results and is widely used in many applications. It creates a graph from the image and segments the non-overlapped regions based on the similarity of nodes in the graph while keeping the results smooth over neighboring nodes. In this paper, we propose a method based on RW that can segment translucent overlapped regions. We create a multi-layer graph from a 2D translucent overlapped input image and generate the Laplacian matrix based on this multilayer graph. Having this new Laplacian matrix and providing a small number of user-defined labels (seeds), the algorithm segments the input image into regions. Experimental results on synthetic images show the strength of our proposed method with more than 98% segmentation accuracy.
|Keywords||Image segmentation, random walks, translucent overlapped images|
|Conference||5th IEEE Global Conference on Signal and Information Processing, GlobalSIP 2017|
Mahyari, T.L. (T. Lotfi), & Dansereau, R. (2018). Random walks for image segmentation containing translucent overlapped objects. In 2017 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2017 - Proceedings (pp. 46–50). doi:10.1109/GlobalSIP.2017.8308601