EUROGRAPHICS Symposium on Sketch-Based Interfaces and Modeling (2009) C. Grimm and J. J. LaViola Jr. (Editors) Sort, Merge, Repeat: An Algorithm for Effectively Finding Corners in Hand-sketched Strokes A. Wolin, B. Paulson, and T. Hammond 1 1 Sketch Recognition Lab Department of Computer Science Texas A&M University College Station, TX 77843-3112 {awolin, bpaulson, hammond}@cse.tamu.edu Abstract Free-sketch recognition systems attempt to recognize freely-drawn sketches without placing stylistic constraints on the users. Such systems often recognize shapes by using geometric primitives that describe the shape’s appearance rather than how it was drawn. A free-sketch recognition system necessarily allows users to draw several primitives using a single stroke. Corner finding, or vertex detection, is used to segment these strokes into their underlying primitives (lines and arcs), which in turn can be passed to the geometric recognizers. In this paper, we present a new multi-pass corner finding algorithm called MergeCF that is based on continually merging smaller stroke segments with similar, larger stroke segments in order to eliminate false positive corners. We compare MergeCF to two benchmark corner finders with substantial improvements in corner detection. Categories and Subject Descriptors (according to ACM CCS): I.4.6 [Computing Methodologies]: Image Processing and Computer Vision—Segmentation - Edge and feature detection 1. Introduction Sketch recognition is an emerging field that utilizes pen- based interfaces in an attempt to make human-computer in- teraction as natural as human-human interaction. In these interfaces, electronic styli replace traditional mice and key- boards and allow users to draw onto a digital screen as if it were intelligent paper. Free-sketch recognition (or natural sketch recognition) al- lows users to draw without developer-placed constraints on the drawing style. Such constraints could force a user to draw primitives in separate strokes, draw strokes in a certain order, or learn a set of prespecified gestures (e.g., [Rub91, ACLLRM00]). Researchers have built free-sketch recogni- tion systems in domains such as circuit diagrams [AD04], figure recognition [SvdP06], and UML diagrams [HD02] (Figure 1). Some important free-sketch recognition systems that avoid constraining the user are geometric recognizers, which define shapes by sets of primitives and geometric rules [HD07] or graphical models [CD04]. A geometric Figure 1: A free-sketch recognition system for UML dia- grams, implemented in LADDER. recognition system works by first breaking down drawn strokes into primitive shapes, and then classifies the group of c  The Eurographics Association 2009.