Word Based Recognition of Cursive Script

We describe a system for on line or off line recognition of cursive script. Our earlier work established that cursive script can be synthesised out of individual characters by using polynomial merging functions which satisfy boundary conditions of continuity of the displacement functions x(t) and y(t) for each character and their first and second derivatives. We showed that even individual characters could be synthesised out of more primitive elements by using the same merging functions. The elements we choose are straight lines: not the usual line segments but a much smaller number of directed lines which we call shape vectors, ranging from only three vectors for simple characters such as e, 1 and o and a maximum of seven for m. We use slopes of the shape vectors and relative locations of points of maximum curvature (both highly quantised) as parameters for recognition. The system extracts parameters for individual characters from single specimens written in isolation and uses these to construct feature matrices for words in the vocabulary. During recognition, these are matched with the feature matrices of test words. The system achieves recognition scores of 94% for vocabulary sizes in the range of 100 words.