Vectorization of graphical components in sketch-based interfaces

Sketching rapidly communicates ideas through approximate visual pictures with low overhead (pencil and paper), easy to correct and revise without necessity of specialized knowledge or precision. A picture can contain characters, symbols and graphical components. Usually, the recognition of graphical components and their vectorization follows the text and symbols detection phase in pattern recognition algorithms. This paper presents an algorithm, which analyses an hand-drawn sketch and recognizes and vectorizes the graphical components of the sketch. The proposed algorithm make the assumption that the sketched picture is given by an array of pixel and the graphical components are represented by black pixels on a white background. The algorithm uses a representation based on coding horizontal sequences of black pixels identified in the image. The algorithm consists of two steps: the first one identifies the horizontal sequences of black pixels and stores them in a list L, the second step read the list L, recognizes and stores the graphical components in different lists classified by types.