Automated landmarks detection on 3d human facial image / Ngo Chee Guan
One of the methods of craniofacial anthropometry is indirect anthropometry. A measurement performs on digital facial images or x-ray images. In order to get this measurement, few definable points on structures in individual facial image are needed and mark as landmark points. Currently, most of the...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2014
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/6408/1/NgoCheeGuan_SGJ130008.pdf http://studentsrepo.um.edu.my/6408/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | One of the methods of craniofacial anthropometry is indirect anthropometry. A measurement performs on digital facial images or x-ray images. In order to get this measurement, few definable points on structures in individual facial image are needed and mark as landmark points. Currently, most of the work in anthropometric studies uses landmark points that are manually marked on 3D facial image by examiner. This method leads to time consuming and human bias, which will vary within intra-examiner himself and among inter-examiners when involve large data sets. Biased judgment as well leads to a wider gap in measurement error. Thus, this work aims to automate the process of landmarks marking that will help in enhancing the accuracy of measurement.
In this work, automated landmarks detection on 3D human facial image system is produced to identify nasion (n), pronasale (prn), subnasale (sn), alare (al), labiale superius (ls), stomion (sto), labiale inferius (li), and chelion (ch). These landmarks are detected on 3D human facial image in .obj file format and used for obtaining measurements. These measurements are important in craniofacial analysis such that the 3D facial image represents the digital print of the human facial image.
Relationship between 3D geometry characteristic with human feature points is studied. Pronasale (prn) are located by searching for vertex with global maximum z coordinate while others are located by seeking for local maximum or minimum coordinate at specific region. Fine tuning is implement on certain landmarks by using dot product of two vectors at 3D Euclidean space to improve accuracy.
The system has successfully identify landmarks and obtain measurements with accuracy at nose region is better than orolabial region as landmarks location at orolabial region will be affected by facial expression. |
|---|