Relative Width and Height of Handwritten Letter

Abstract

This is an exploratory study that analyzes the width and the height of letters in two texts written by each of the 21 writers analyzed. After detrending the linear, text, and allograph trends, we proceeded to comparing the sizes obtained in different texts. The different detrended series were compared by means of correlation and t-test. According to the results regarding the width of letters, the texts of 19 of 21 writers correlated strongly, whereas the texts of two writers did not correlate with the limits of the threshold. With regard to the height of letters, texts written by between 18 and 21 writers of 21 writers correlated strongly, whereas texts that did not correlate were within the threshold value. Regarding both the width and the height of letters, of 21 writers, texts written by between 19 and 21 individuals were found to correlate strongly.KEYWORDS: forensic science, forensic handwritten examination, questioned document examination, handwriting processing, document analysis, letters width and height This study is based on a concern about the possible individuality of the width and the height of handwritten letters written by a specific writer. As Figs 1 and 2 show, in the measurements of the width and the height of several letters, it may be seen that allographs "a" and "e" tend to be narrower than the allograph "t." If this trend were to be proven in this particular writer, it could help us distinguish this writer from others who write the very same allographs inversely when it comes to width or height.This embryonic idea led on to attempts to systematize these possible differences and distinguish the writer according to the width or height profile of different allographs.Regarding the width of letters, if there are two states for each of the 52 allographs in lower case written by an individual -"wide allograph" and "narrow allograph," the possible combinations of two states for letters in lower case raised to the 26th power are 67,108,864. Hence, more than 67 million might potentially be characterized. If there were five states ("very wide," "wide," "neither wide nor narrow," "narrow," and "very narrow"), the possible combinations of five states raised to the 26th power are 1, 490,116,119,384,765,625; that is to say a number which is 19 digits long. Having said that, taking into account that the width of letters is a continuous magnitude, the possible states of 52 allographs are much larger. It is at this point where we need the correlation and test of statistical significance of correlation coefficient (t-test) techniques, which are capable of comparing objects to a vast number of possible states. However, we should not forget that we do not know the distribution of the frequency of the possible states.Taken into account the specific features of the texts, common measurable features that adjust to every allograph and text were selected.We must point out that this is an exploratory study; thus, it must be emphasized that the results obtained are to be applied exclusively to the 42 texts ...