Alexandra Bartolo

Scribbles to vectors: preparation of scribble drawings for CAD interpretation

This paper describes the work carried out on off-line paper based scribbles such that they can be incorporated into a sketch-based interface without forcing designers to change their natural drawing habits. In this work, the scribbled drawings are converted into a vectorial format which can be recognized by a CAD system. This is achieved by using pattern analysis techniques, namely the Gabor filter to simplify the scribbled drawing. Vector line are then extracted from the resulting drawing by means of Kalman filtering.

Line tracking algorithm for scribbled drawings

This paper describes a line tracking algorithm that may be used to extract lines from paper based scribbles. The proposed algorithm improves the performance of existing sparse-pixel line tracking techniques that are used in vectorization by introducing perceptual saliency and Kalman filtering concepts to the line tracking. Furthermore, an adaptive sampling size is used such that it is possible to adjust the size of the tracking step to reflect the stroke curvature.

Adaptation of Brensen's thresholding algorithm for sketched line drawings

Image binarisation is one of the first image processing techniques within a sketched line drawing interpretation system. In order to achieve an automated system, it is necessary to have a single algorithmwhich may be used for all image types. This is difficult to obtain if the parameters required by an algorithm are set manually, however, the adaptive evaluation of parameters form image properties is a step towards this goal. This paper discusses the methods by which the parameters required for Brensens Algorithm may be evaluated adaptively form the image properties.

A new sketch based interface using the gray-level co-occurrence matrix for perceptual simplification of paper based scribbles

The sketching activity has an important role in conceptual design and a variety of tools exist which help designers to facilitate the generation of 3D models form sketched drawings. This paper describes a new sketch-to-3D tool, which uses annotations to aid the interpretation of the drawing. Over-traced lines present in the designers scribbles provide an interpretation challenge, which must be resolved in order to obtain 3D models from these sketches. Perceptual grouping techniques used to interpret such images require that the drawing is represented as vectors. These are generally obtained through thinning or edge detection. However, we show that processing scribbles using these techniques result in a large number of vectors which do not provide a faithful representation of the drawing. This paper investigates the use of the co-occurrence matrix to perceptually simplify these drawings, thus obtaining a smaller number of vectors which describe the drawing more faithfully.

Paper-based scribble simplification: where do we stand?

This paper presents a formal evaluation of the paper-based scribble simplification algorithm described in [BCFB07] and [BCFB08]. A comparative analysis of different aspects of the algorithm with other algorithms described in the literature such as Sparse Pixel Vectorization, spatial moving average filtering and Principal Component Analysis is performed, hence establishing the qualities of this paper-based scribble simplification algorithm. To quantify the performance of the algorithm, performance measures established in the literature, such as the Pixel Recovery Index are used when suitable. However, since there exists no quantitative measure which measures scribble simplification, this paper proposes a new methodology with which scribble simplification may be quantitatively assessed. Through the evaluation described in this paper, we will be able to determine remaining difficulties in the interpretation of paper-based scribbles and hence identify future research areas.

Image binarisation using the extended Kalman filter

Form design is frequently carried out through paper sketches of the designer’s mental model of an object. To improve the time it takes from solution concept to production it would therefore be beneficial if paperbased sketches can be automatically interpreted for importation into three-dimensional geometric computer aided design (CAD) systems. This however requires image pre-processing before initiating the automated interpretation of the drawing. This paper proposes a novel application of the Extended Kalman Filter to guide the binarisation process, thus achieving suitable and automatic classification between image foreground and background.