Abstract:
Digital documents and presentations often contain sketches and diagrams and drawing these diagrams is a time consuming activity despite the availability of a number of drawing tools. In general, diagrams and sketches can be more easily drawn on paper with hand as compared to digital drawings. These hand-drawn sketches, however, are not very pleasing if directly added to digital documents and need to be converted into digital drawings. This automatic conversion of hand-drawn sketches to digital form is the subject of our study. This project presents an Android application that takes images of hand-drawn sketches, applies image analysis and pattern classification techniques and converts the hand-drawn diagram to a digital diagram. The hand-drawn image is provided to the application which binarizes it and extracts the connected components in the image. Hough transform is then applied to identify each shape as a circle, ellipse, rectangle or line. The recognized shapes are then digitally drawn and saved as an image. The system evaluated on varying hand-drawn drawings realized promising recognition rates.
