Automatic Anthropometric Measurement Systems from 3D Hand Scans: Accuracy of a Developed Algorithm for Large Number and Various Types of Dimensions
2022-08
Loading...
View/Download File
Persistent link to this item
Statistics
View StatisticsJournal Title
Journal ISSN
Volume Title
Title
Automatic Anthropometric Measurement Systems from 3D Hand Scans: Accuracy of a Developed Algorithm for Large Number and Various Types of Dimensions
Authors
Published Date
2022-08
Publisher
Type
Thesis or Dissertation
Abstract
There is a research gap in using an automatic dimension extraction process to extract a large set of dimensions and different types of measurements from 3D hand scans that are developed from the physical hand. In this thesis study, an algorithm was developed, tested, and evaluated to address that gap. Twenty participants from the Human Dimensioning Lab at the University of Minnesota-Twin Cities were used. Sixty-eight Landmarks were used to extract 121 measurements (50.4% linear, 38% surface arc, and 11.6% circumference) from these scans automatically and manually in the digital medium by a measurer. Paired t-tests, mean absolute difference (MAD), and percentage of differences (diff%) were used for statistical analysis. ISO standard for maximum allowable limit is used for MAD acceptance. The automatic processing successfully measured 99.25% of all dimensions. It extracted the dimensions that the manual measuring process couldn’t measure. The automatic method extracted 87% linear, 30.4% surface-arc, and 43% circumference dimensions accurately without statistical significance. For surface-NC measurements, it measured 14 dimensions within the acceptable MAD limit. More than 50% of the dimensions that were found statistically significant had a single or double participant with a very large value that impacted the MAD value. For 60% of statistically significant dimensions, the percentage was less than 5%. For circumference measurements, 8 measurements were statistically significant and three of them have a difference percentage within 3%. Linear dimensions had statistically significant differences due to the difference in dimensions extraction procedure for breadth and depth measure between the manual and automatic measuring systems. The cause of the difference in surface-arc and circumference measurements are seemed to be similar. Thumb movement, hand posture, and scan quality were the biggest barriers to measuring these dimensions accurately. Overall, the proposed technique was performed in an acceptable form under conditions for surface-NC and circumference measurements when it is working well for linear measures. A new method is needed to validate automatic processing from the manual measuring process when they both measure from a 3D scan. Both scanning protocols, landmarking, and software need to be developed together to improve the accuracy of automatic measurements. By making the automatic process adaptive to the scan, the automatic dimension extraction process could be a game changer in 3D hand anthropometry and product design.Keywords: automatic dimension extraction, 3D hand scan, digital anthropometry, validation of measuring process, algorithm, anthroscan, manual measurement, 3D scanning
Description
University of Minnesota M.S. thesis. 2022. Major: Design, Housing and Apparel. Advisors: Linsey Griffin, Elizabeth Bye. 1 computer file (PDF); 124 pages.
Related to
Replaces
License
Series/Report Number
Funding information
Isbn identifier
Doi identifier
Previously Published Citation
Suggested citation
Bhuyan, Md Arif-Ul-Anwar. (2022). Automatic Anthropometric Measurement Systems from 3D Hand Scans: Accuracy of a Developed Algorithm for Large Number and Various Types of Dimensions. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/250025.
Content distributed via the University Digital Conservancy may be subject to additional license and use restrictions applied by the depositor. By using these files, users agree to the Terms of Use. Materials in the UDC may contain content that is disturbing and/or harmful. For more information, please see our statement on harmful content in digital repositories.