We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
基于笔痕角度特征对高水平机器人书写字迹的识别研究.
- Authors
杨求凤; 韩星周; 王晓光; 秦 达; 闫海倩; 郭姿含; 郝红光; 王锦生
- Abstract
Whether the suspicious handwriting is written or not is the key in the handwritirg examination, which directly determines the conclusion. The traditional methods of forging handwriting, inclicding printing, copying and stamping, are easy to identify, because there are no stroke pressure. But the advances in robotic writing technology pose new challenges for forensic document examiners. Robotic writing can form three-dimensional stroke indentation, and it is easy to misjudge the handwriting formed by normal writing in the examination. Some researchers have modifi ed the Cartesian writing robot to make its handwriting more natural, and more similar to the handwriting written by human, which is difficult to distinguish. Forensic document examiners should be alert to this kind of high-level forged handwriting. In this study, by utilizing the optical tomography and depth-of-field technology of AB 8000 handwriting quantitative tester, we measured the strokes of natural handwriting samples and high-level robotic writing samples, and obtained the width and depth of stroke, calculated the ratio i of the horizontal distance between the deepest mark and the edge of stroke and the width of stroke ink. We alsa measured the distribution degree of i value by chart and differential value operation, and analyzed the aggregation and fl uctuation of the mark angles between high-level robotic writing samples and human handwriting samples. The results showed that the stroke angle of high-level robotic writing samples has high aggregation and low volatility, while the stroke angle of human writing samples has low aggregation and high volatility. Therefore, it is verifi ed the validity of stroke angle feature in recognizing the robotic writing. It is also observed in the experiment that the stroke angle of long strokes is more stable, and the angle data of turning strokes and crossing strokes will be disturbed, so it is more accurate to choose long strokes for measurement.
- Publication
Forensic Science & Technology, 2023, Vol 48, Issue 6, p563
- ISSN
1008-3650
- Publication type
Article
- DOI
10.16467/j.1008-3650.2023.0052