LinedCut: Image segmentation using single line interaction

In this paper, we propose LinedCut: a novel method for interactive image segmentation which requires only a single line drawing to identify the object of interest in the image. The handy interaction mode can address the problem of object scale very well. Our approach consists of the following three steps: first, a given image is over-segmented into superpixels using superpixel algorithm; secondly, a merging technique based on bag-of-color feature and k-means is applied to merge similar adjacent superpixel pairs; finally, a graph-cut based approach which exploits user interaction information is introduced to get the final segmentation result. Despite its simplicity, we show that the LinedCut method is able to achieve a performance comparable to the state-of-the-art. The method can be easily developed by replacing any one method among the three steps with other methods.