|Year : 2021 | Volume
| Issue : 4 | Page : 152-158
Pilot study on associating pen pressure with pen holding position using three-dimension property of stroke indentation
YingXiu Guo1, Bing Li2
1 Fada Institute of Forensic Medicine & Science; China Everbright Bank, Beijing, China
2 Institute of Evidence Law and Forensic Science, China University of Political Science and Law; Key Laboratory of Evidence Law and Forensic Science; Fada Institute of Forensic Medicine & Science, Beijing, China
|Date of Submission||08-Oct-2021|
|Date of Decision||15-Nov-2021|
|Date of Acceptance||25-Nov-2021|
|Date of Web Publication||30-Dec-2021|
Institute of Evidence Law and Forensic Science, China University of Political Science and Law, Key Laboratory of Evidence Law and Forensic Science, Beijing, China. Fada Institute of Forensic Medicine and Science, No. 26, Houtun South Road, Qinghe Xiaoying Area, Haidian District, Beijing
Source of Support: None, Conflict of Interest: None
Objectives: The author collected several samples with Chinese characters and utilized the 3D property of stroke indentation to explore how pen strokes left on the paper can assist with forensic investigation. Methods: We chose to collect the three-dimension property of handwriting strokes and analyze strokes indentation. In this study, the author extracted stroke indentation through focus variation microscopy as implemented by AB8000. Results and Conclusion: Upon experimenting with different types of pens and different writing surface, the author found that this result can tell with a great degree of certainty whether the sample is formed with a left-hand or right-hand writer and how the pen was being held. Although there are variations among the test results, the discovered pattern generally holds for different types of pens and writing surfaces, except that the result from using felt tip pens or extremely hard surfaces are less satisfactory. These findings are expected to be able to assist in narrowing down who wrote a certain handwriting sample.
Keywords: Left-hand writing, pen pressure, right-hand writing, three-dimension property of stroke indentation
|How to cite this article:|
Guo Y, Li B. Pilot study on associating pen pressure with pen holding position using three-dimension property of stroke indentation. J Forensic Sci Med 2021;7:152-8
|How to cite this URL:|
Guo Y, Li B. Pilot study on associating pen pressure with pen holding position using three-dimension property of stroke indentation. J Forensic Sci Med [serial online] 2021 [cited 2022 Jan 20];7:152-8. Available from: https://www.jfsmonline.com/text.asp?2021/7/4/152/334492
| Introduction|| |
Handwriting examination is an important part of forensic science. Like many other branches of forensic disciplines, it examines the available evidence to reconstruct a past event. Handwriting examination has a wide range of applications, for example, to determine whether a will or a deed is written by the claimed person, to determine the order of handwritings and stamps, to evaluate the age of a document, and so on.
The forensic science of document examination in handwriting and signature is comprised of several subcategories, for instance, signature identification, the order of stamp and handwriting, printed documents, and so on. This study focuses on using the indentation made on the paper to deduce which hand the writer used and which angle the pen was held with.
For many years, determining whether a handwriting sample is genuine has always been carried out by expert experience. In recent years, along with other types of expert opinions, there is a development of analyzing handwritings in a more quantitative way. There have also been an increasing number of novel approaches to handwriting examination. In this study, we explore the indentation of pen strokes left on the paper, which makes a handwriting sample three dimensions (3D) instead of just two. By examining the characters of those indentations, it is safe to claim that there is a great potential to using the indentation of pen strokes on the paper to deduce which hand was used and how the pen was held, hereby determining whether the writing is formed by a writing habit that is consistent with the claimed writer.
Although generally used in a court setting to settle disputes, handwriting examinations have a wide range of applications. For instance, Agius et al. have expanded the use of handwriting examination and proposed that it can be used to obtain additional information for intelligence purposes.
There are many approaches to forensic document examinations, including paper, ink, and the characters. Kumar et al. explored the use of Fourier transform infrared spectroscopy and chemo metrics in discriminating paper samples, and their results showed that this method is completely analytical and can highly in forensic investigations.
It is not a novelty to introduce statistics and quantitative methods in forensic handwriting examinations, and there is a trend to use recently developed techniques in this field. One reason for this is that there have been an increasing number of reports on the unreliability of experts, who as human beings are subjects to bias and irrelevant information. In the field of forensic handwriting examination, Found and Ganas acknowledged the potential of context information affecting decision-making and therefore introduced a context management scheme. There is a trend to move away from relying solely on exert opinions. Saini and Kaur demonstrated that image processing technique can be used to detect computer-based manipulations of written documents. Taroni et al. presented a framework for assessing, articulating, and communicating evidential value of handwriting features, basing on the idea that Bayesian approach would make evidence evaluation more methodological. Chen et al. presented a new methodology for the quantitative feature extraction and statistical analysis of Chinese signature examination.
Upon reviewing the existing literatures on handwriting examinations, the author realized that although there has been an extensive effort in forensic handwriting analysis, none has been found on exploring the indentation of the pen strokes, nor utilizing it to deduce the identity of the writer. Few people have realized that like footprints, handwriting samples often exist in 3D. Therefore, it is always helpful to explore new ways of how handwriting characteristics relate with writing movement.
| Methods|| |
We chose to collect the 3D property of handwriting strokes and analyze strokes indentation. In this study, the author extracted stroke indentation through focus variation microscopy (FVM) as implemented by AB8000. For each axial step move, a high resolution image is captured, and the contrast level of all XY positions in each image is calculated. After all images have been captured, an advanced proprietary algorithm calculates the Z-position with the highest contrast level of all XY-positions in all the images. Based on this information and a comprehensive AI filtering in all 3D, a highly precise image is being rendered. Therefore, FVM is a very suitable method for measurement of stroke indentation in paper.
Among other features, this method can show the details of the cross-section of a stroke with great clarity. One of the important advantages of this method is that it is based on a magnifying principle, which means that it is completely harmless to the sample. Leaving the sample intact is crucial in forensic science because of the high stakes usually involved in the judicial process, as well as the possible needs of further analyses.
During the first stage of the experiments, the writers used a black gel pen and ordinary white paper. To facilitate the formation of the indentation, there are two pages of the same paper under the page when writing each stroke and character. The writing conditions are kept constant throughout the experiments.
Some preliminary pilot tests were designed to better develop the main experiments. First the author analyzed the four most basic strokes in Chinese characters, i.e., “一,” “丨,” “丿,” and “╰.” The author wrote these strokes when the pen was held in 90, 60, 45, and 30 degrees. There were three repetitions in each scenario but for 30° cases to ensure the robustness of the test. We predicted that the cross-section of a horizontal stroke cannot indicate the angle of the pen, as the pen movement is in the same direction of the groove. In addition, if the person writing the paper in question holds the pen in a vertical position, then the indentation on the paper would have a deep furrow in the center of the stroke, and overall the image would appear symmetrical.
The results outputted from AB8000 are as follows in [Table 1], where N denotes that no angle can be discerned and Y denotes the contrary.First of all, we can acknowledge that the pilot tests confirmed our prediction that the horizontal stroke “一” is useless in terms of determining whether the pen was held tilted. It is clear that this method is sufficient to tell when the pen slanted 45°, but we nevertheless tested a couple of strokes that were formed in 30°. As expected, this method yielded all positive results. It is logical that no angle can be observed when the pen was held vertical. Yet, the we found that the indentation from 60° is also symmetrical most times, with only one case that the results suggests the pen was slanted with some degrees. This means that this method is not subtle enough to distinguish vertically held pens and pens that were slanted with a small degree.
It was discovered during the pilot tests that the indentation left on the paper is indicative of which hand the writer is using. Specifically, while writing the strokes other than “一,” the deepest indentation tends to appear on [Figure 1]. Therefore, we form the second hypothesis that one can determine which hand the writer used based on the shape of the indentation formed on the paper.
To summarize, the pilot tests served two purposes. First, it established what the angles of the pen tilting we should measure, i.e., around 60° and around 45°. Second, it proposed the possibility of determining which hand wrote a stroke by analyzing the indentation it left.
Based on the hypothesis formed from the previous section, a series of experiments were designed to test the possibility of determining whether the left hand or the right hand was used to write a single stroke. The author used the left hand to write three strokes, “丨,” “丿,” and “╰,” with three repetitions. The pen was forced to be held with 45° since the pilot tests indicate 45° is the threshold. The other writing conditions are kept same as before.
Instead of experimenting with the right hand again, the author reused the data from the pilot test to avoid unnecessary repetition, which includes the results from writing these three strokes when the pen was held 45° using the right hand.
Pen holding position
Here, we developed that the assumption is that the pen slanting degree is related to where one holds the pen. Because of geometry, the closer to the tip one holds the pen, the more vertical the pen is, and vice versa. Based on this, we asked participants to hold the pen in two different points while writing some samples. This is more natural than forcing the participants to hold the pen in a specific degree.
We choose to analyze the Chinese character “一”as shown below in [Figure 2], as it is composed of the four most fundamental basic Chinese strokes: “一,” “丨,” “丿,” and “╰.” The participants write this character in different pen holding positions but with the same pen and same type of writing conditions to control variables.
There are 32 repetitions in two holding positions. In each scenario, trials 1 to 4 were written by one person, and trials 5 to 8 were written by the second, trials 9 to 12 were written by the third, and trial 13 to 16 were written by the fourth person. The selected participants were different in gender and age. All of the participants are right-handed. A sample is presented below.
The horizontal stroke “一” is omitted in the analysis this time, as indicated by the pilot test that the cross section of a horizontal stroke cannot offer much information.
Another assumption derived from the experience of handwriting examiner is that the end of a stroke tends to be more distinguished, which is due to the relatively strong force one applied when one finishes a stroke. Therefore, each “丨” stroke was analyzed twice and different locations.
If the results from the tests are satisfactory, we would continue the tests in less tightly controlled environments. This is to better imitate the real writing conditions so that the results would have a higher external validity. We include other types of pens and different writing surfaces in this stage of experiments. [Figure 3] shows how the same character looks when written with ball point pen, pencil, and a black marker.
|Figure 3: The same character written with different types of writing instruments|
Click here to view
| Results|| |
The completed results are as the following. Here, the capital Y denotes that the indentation on the paper clearly indicates that the pen was slanted to the correct direction while writing the letter, small y denotes that the indentation can somewhat indicate that the pen was slanted, and N denotes that the indentation cannot correctly indicate that the pen was slanted.
All trials were conducted by the same right-handed people as shown in [Table 2]. Two people participated in this experiment. Results from using the right hand writing the same strokes are presented again as a comparison.
The results showed that 27 out of the 36 trials are able to indicate which hand was used. Among them, 18 out of 27 offered strong evidence, and the rest offered moderately strong evidence. It is worth noting that most of the unsatisfactory results occurred when the left hand was used.
Here, two screenshots are attached. [Figure 4] shows the data of the cross-section when the left hand was used, and [Figure 5] shows the same data but using the right hand.
It is obvious from the comparison that when the stroke was formed with the left hand, the deepest indentation is on the left side, and when the stroke was formed with the right hand, the deepest part is on the right.
Pen holding position
Here, sixteen trials were conducted by four different people as shown in [Table 3].
The first thing to notice from the results is that when holding the pen closer to its tip, i.e., when the pen is tilted for only 60°, the resulting indentation made on the paper is not very indicative of its shallow angle. This is consistent with the results from the pilot tests. In other words, the indentation was symmetrical, which suggests that the pen was held more or less upright. Only a few cases show that there is an angle to the pen handling, and those few cases mostly clustered to the “丿” stroke. As for the “╰” stroke, all the results indicates that the pen was held more or less vertical.
Equally obvious is that in the case of 45°, the stroke “丿” can almost always indicate that the pen was held with an angle, and the stroke “╰” almost always fails. It can be easily observed that while the “丨” and “丿” strokes in the character can indicate the angle of the pen, the last stroke and the one to the right, “╰,” can not.
The behavior of the lower portion of “丨” is more satisfying than that of the upper portion, as made clear that there are more “Y” or “y” in the last column than in the second to last.
Results from generalized experiments
The followings shown in [Table 4] are the results from further generalized experiments. The writing conditions are established based on results from basic tests. Compared to those, these tests have fewer repetitions but broader parameters. All of the following experiments were conducted with right hand when the pen was tilted 45°. We only increased the hardness of the writing surface, as it is to be expected that the harder the surface is, the harder it is to form an indentation, and a softer writing surface would only make the indentation clearer. Specifically, we reduced the paper underneath the page written on from two pages to one and zero. Only the lower portion of the “丨” stroke was analyzed, as it has been demonstrated that this yields a clear image than the upper portion. Considering felt tip pen only yields negative results when one page is included, it is not used when writing on glass panel.
| Discussion|| |
Before we dive into discussions in detail, it is important to keep in mind that the results are obtained from the cross-section of a stroke, which means that there are about many cross-sections in any given stroke. The assumption is that the behavior of the indentation would remain constant in a single stroke, but it might not always be the case. We already considered the possibility that the lower part of “丨” might be more reliable than the upper part. Therefore, it is possible that some parts of a stroke can indicate the position of the pen better than others. Bearing in mind the possibility of the presence of false positives and false negatives in the above data, we can explore what those data means.
The experiments regarding dominant hand indicated that it is reliable to deduce the writing hand based on the indentation left on the paper. The right-hand trials seem to be more satisfactory than the left-hand ones, given that the results are all categorized as capital “Y”s, but it should be noted that the writer is right handed, so there was some awkwardness while using the left hand to write, which might account for the fact that there are a few “y”s in the left-hand experiments. Nevertheless, all the results are either Y or y, which means all the tests show the machine can tell which hand the writer used based on the indentation of the stroke. This means that there is a great potential of using where the deepest indentation on the cross section of a stroke lays to determine which hand the writer used.
Pen holding positon
The results from trials concerning 60° are as expected since the pilot test already indicates that the difference between 60° and 90° is too subtle to be captured using this instrument. Therefore, the pen angle can be detected in only a few cases.
As for the data from trials with 45° of slanting, the stroke “丿” yields the most satisfying results, whereas the stroke “╰” is the worst. One explanation is that as the writer is right-handed, so the pen has a smaller degree of slanting when writing a stroke that appears on the right-hand side of the page, which is the case of the “╰” stroke. On the other hand, “丿” appears on the left of the character and hence was written when the pen was held more horizontally.
The justification for testing both upper and lower portion of the “丨” is that as one finishes a stroke, more force tends to be applied. Therefore, the discriminating features could be more distinguishable in the lower part. As the data above indicate, when the pen is slanted for 45°, there are four cases, trial 5, 6, 8, and 11, where the lower portion of “丨” yields considerably more information than the upper one. This confirms our predictions to some degree.
There are some inconsistencies between the first part of the experiments, writing single strokes, and the second part, writing full characters. Aside from natural fluctuations of data, two explanations could be offered. First of all, although the mechanism is the same, writing a stroke is more unnatural than writing a full character, the latter is part of daily life. Second, when writing a character, only the pen was moving and the hand was usually kept still, which can account for some angle change in one character. In other words, the pen is more vertical when writing the portion that appears on the relative right of the character.
The hypothesis is that when there are less than two pages between the table and the paper one writes on, the surface would be harder and therefore the indentation less clear. The results indeed showed that there are less positive outcomes in the generalized tests. In addition, the type of pen also has an influence over whether information can be drawn.
There is no significant difference between writing with one page underneath the writing surface and when writing on a glass panel. However, it is clear that both cases were worse than writing with two pages underneath the writing surface, which hence formed a rather soft surface.
When a ball pen is used, which is the closest thing to the gel pen among the newly introduced writing materials, there are more positive results than when other types of instruments were used. However, the “丿” stroke shows some eccentricity. Based on previous experiments, the “丿” stroke should be the most capable in indicating the shallow angle of the pen because it is situated in the most left of the character. This incongruity could be due to the mechanism of the ball pen, in the sense that it works less fluently when moving it from right to left. When a pencil was used, the result is less clear than that of the pen, in the sense that only “丿” stroke can somewhat tell whether the pen was held with a degree, but this is consistent with previous results using gel pens. When the pen was felt tip (the one used in this experiment is a black mark pen), it is more difficult to detect the angle of writing, which means that a soft tip writing instrument should not be relied upon to produce information on how characters are formed.
Based on the results presented above, it is sufficient to conclude that there are two ways examining the deepness of a stroke can contribute to handwriting analysis based on the indentation on the paper.
First, the practitioners can infer whether the writing sample is formed by a left-handed person or a right-handed person. As the data above indicate, when the stroke is formed with the right hand, the deepest indentation tends to appear on the right, whereas when the stroke is formed with the left hand, the deepest indentation tends to appear on the left. This finding can be used where the author of a writing sample is in dispute. However, it is very important to note that although knowing which side of the hands was used is useful, it cannot be used for individualization purposes. In other words, it can only say that a group of people is more likely to have written a character than another group.
The second conclusion is that it is possible to distinguish the pen holding position to some degree based on the indentation. If the indentation is not symmetrical, it suggests that the pen is holding relatively far from the tip, hence resulting in a more slanted pen position. A similar caution has to be made as in the first implication. The results of this test are only applicable in narrowing down a group of likely candidates.
Furthermore, upon extending the material of the experiments and include the more real-world writing conditions, the author found that the results can be generalized to some of the other types of pens and potentially to different degrees of softness of the writing surface. This justified the applications of the results to real case works.
Since this experiment is based on the indentation left on the page, it follows that even without the ink, the experiment can still be conducted. For example, if the only available material is the page that is below the one that was written on, it is still possible to obtain information based on how clear the indentation is. This is a possibility that has a great forensic value and should be explored by future experiments, so that information can still be obtained from the page under the original sample which might be ruined or otherwise unavailable.
Limitations and recommendations
For determining dominant hand, the experiment is carried by two people who are right handed. Therefore, more experiments should be done with a more diverse sample. It is ideal to conduct the experiment with right-handed writers, left-handed writers, as well as ambidextrous writers. Furthermore, we considered the effect of the condition of the writing surface. We changed the numbers of pages underneath the page we were writing on to vary the hardness of the surface, but different textures could also be included. For example, writing on glass, on desk, or on dining table. In addition, it would improve the external validity of this study if more writing samples and more participants are included.
Moreover, it is needless to say that this study only applies to traditional handwriting samples, which exclude anything written on tablet or other types of electronic devices. Further experiments can be explored with tablets that can record the vector of strokes.
The foremost limitation of this study is that it only applies to comparing samples which are formed with different pen holding positions. For example, if both samples were written with the pen held more than 60°, then it is very difficult to tell them apart based on the form of indentation on the page alone. In this regard, the author wants to emphasize that the results from this study cannot be used for individualization purposes since the pen holding degree is not something unique to one individual. Instead, it behaves like footprint in the sense that it can only eliminate a group of people.
The author mentioned the effects of cognitive bias in previous sections, and the reader has to keep in mind that although most of the experiments rely on an instrument, there is still some human input. For example, it is the author who decided which category to put a certain result, whether it is Y, N, or y. Despite the author's best efforts, confirmation bias could come into play as the author knew which hand and which holding position was used to write the sample. Therefore, one way to improve this study is to ask another participant who is unaware of the writing conditions to rate the outputs.
To complicate the matter further, the examiner has to consider that a person might change their writing habits. As Moszczynski pointed out, a person might develop two or more different styles of writing in their life. Therefore, the examiner has to note the possibility that the inconsistencies between handwriting samples might be arisen from innocent reasons.
| Conclusion|| |
To summarize, this pilot study is an experiment-driven project. The author designed and conducted a number of tests in a systematic way to demonstrate a novel approach to handwriting or signature examinations. The results from the experiments demonstrated that the indentation of the stroke can indicate which handheld the pen and how it is held. Although this experiment has a few limitations, both in terms of it experiment design and its generality, it forms the foundation of using the indentation left on the paper to help determine how a character is formed. Future researches can be built on this to further explore the 3D property of handwriting samples.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Agius A, Jones K, Epple R, Morelato M, Moret S, Chadwick S, et al
. The use of handwriting examinations beyond the traditional court purpose. Sci Justice 2017;57:394-400.
Kumar R, Kumar V, Sharma V. Fourier transform infrared spectroscopy and chemometrics for the characterization and discrimination of writing/photocopier paper types: Application in forensic document examinations. Spectrochim Acta A Mol Biomol Spectrosc 2017;170:19-28.
Dror IE, Hampikian G. Subjectivity and bias in forensic DNA mixture interpretation. Sci Justice 2011;51:204-8.
Found B, Ganas J. The management of domain irrelevant context information in forensic handwriting examination casework. Sci Justice 2013;53:154-8.
Saini K, Kaur S. Forensic examination of computer-manipulated documents using image processing techniques. Egypt J Forensic Sci 2016;6:317-22.
Taroni F, Marquis R, Schmittbuhl M, Biedermann A, Thiéry A, Bozza S. The use of the likelihood ratio for evaluative and investigative purposes in comparative forensic handwriting examination. Forensic Sci Int 2012;214:189-94.
Chen XH, Champod C, Yang X, Shi SP, Luo YW, Wang N, et al
. Assessment of signature handwriting evidence via score-based likelihood ratio based on comparative measurement of relevant dynamic features. Forensic Sci Int 2018;282:101-10.
Moszczynski J. The multi-individuality of handwriting. Forensic Sci Int 2019;294:e4-10.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]
[Table 1], [Table 2], [Table 3], [Table 4]