A circle-based vectorization algorithm for drawings with shadows

Abstract

This work is funded by the University of Malta, under the research grant SCERP02-03.Vectorization algorithms described in the literature assume that the drawings being vectorized are either binary images or have a clear white background. Sketches of artistic objects however also contain shadows which help the artist to portray intent, particularly in potentially ambiguous sketches. Such sketches are difficult to binarise since the shading strokes make these sketches non bimodal. For this reason, we describe a circle-based vectorization algorithm that uses signatures obtained from sample points on the line strokes to identify and vectorize the line strokes in the sketch. We show that the proposed algorithm performs as well as other vectorization techniques described in the literature, despite the shadows present in the sketch.peer-reviewe

Vectorisation of Sketched Drawings Using Co-occurring Sample Circles

“…To avoid these distortions, other methods can be employed to obtain the medial axis, for example, through the use of contours [22] and Delauny triangles [16] among others. Since these methods assess whether each pixel is a candidate pixel on the medial axis, they incur large computational costs which may be reduced by adapting sparse pixel tracking approaches such as that described in [18,17,1] among others. Here, lines are extracted by sampling the image using rectangles [18], squares [17] and circles [1], starting with a strategically located seed pixel which is propagated along the lines of the drawing.…”

“…Since these methods assess whether each pixel is a candidate pixel on the medial axis, they incur large computational costs which may be reduced by adapting sparse pixel tracking approaches such as that described in [18,17,1] among others. Here, lines are extracted by sampling the image using rectangles [18], squares [17] and circles [1], starting with a strategically located seed pixel which is propagated along the lines of the drawing. Since lines have a rectilinear shape, rectangular samplers have the advantage of having a similar shape as the lines being sampled [12].…”

“…However, this will also require rotating the sampler so that it is aligned with the lines in the image [12]. This alignment requirement can be avoided by taking the grey-level profile of the boundary pixels [17,1]. In such cases, circle samplers, being isotropic, are more suited samplers.…”

“…x n (r) = J P + rk θn (1) where x n (r) is the pixel coordinates of a point on the n th line segment forming the junction, r is the displacement from the junction point and…”

SketchSoup: Exploratory Ideation Using Design Sketches

“…Our sketches are dominated by many imperfect and often incomplete strokes—an artefact of the efficiency with which they are executed—and further convey the early and unfinished nature of the design. Vectorizing and lifting these sketches into 3D for the purpose of constructing a 2D design space is thus both difficult [OK11, BC13] and unnecessary. While crystallized product design concepts can be represented by a cleaned‐up network of sketched strokes with coherent geometric properties [XCS*14], this is not an assumption we can make of arbitrary drawings at the ideation stage.…”