Exploring touch interaction methods for image segmentation on mobile devices

Abstract

Photo taking is more and more moving to mobile devices. With this transition, we observe that consumers are also directly editing and sharing photos. Editing of the whole image became very popular, for example with the filters Instagram offers. Some effects, however, require to edit a specific object of an image and for this an interactive segmentation is needed. This process is still tedious, as existing segmentation approaches focus on the algorithms rather than on the input methods. This paper explores different interaction techniques for image segmentation on mobile devices. We conducted two studies. The first to identify intuitive input methods and the second to evaluate the findings of the first study. As a result, we identified preferred gestures for the initial step to select a foreground object in an interactive image segmentation process. The results give first insights which input methods users intuitively use, how the users behave and what are the problems of the methods. This insights should be taken in consideration by the development of new interactive segmentation algorithms for mobile touch-based devices.