Download
1 / 33
330 likes | 529 Views
Automatic Discovery of Connections between Vietnamese’s Anthropometric Features – Master Thesis. Dinh Quang Huy Iba Laboratory. 2. 3. 4. Our Approach. Evaluation and Result. Conclusion and Future Work. Contents. 1. Background and Motivation. Facial Reconstruction.
E N D
Automatic Discovery of Connections between Vietnamese’s Anthropometric Features – Master Thesis DinhQuangHuy • Iba Laboratory
2 3 4 Our Approach Evaluation and Result Conclusion and Future Work Contents 1 Background and Motivation
Facial Reconstruction • Recreating the face of an unidentified individual from their skeletal remains. • Forensic: determining the face of victims in the murders • Archaeology: verifying the remains of historic figures. • Anthropology: approximating the look of early hominid forms.
Facial Reconstruction Techniques • Three methods • 2D reconstruction • Manual 3D reconstruction • Computer-aided 3D Reconstruction
2D Reconstruction • Requires a forensic artist who draws the picture • Usually requires cooperation between artist and forensic anthropologist • Reconstruction process • Tissue depth markers are fixed on the skull • Taking photograph • Draw the portrait
Manual 3D Reconstruction • Clay modeling approach. • Landmark dowels are placed on the predefined landmarks on the skull • The lengths of these dowels are the tissue depth at the dowel’s landmark. • Clay is then filled between the dowels to build up the tissue depth of the unknown person • Effectiveness of clay reconstruction depends wholly on the skill of the sculptor
Computer-aided 3D Reconstruction • Manual approaches have limitations such as time consuming and sculptor experience dependence • Most computer-aided 3D reconstruction systems use the same process • The skull is scanned to obtain its 3D model • Facial landmarks on the skull are detected or manually selected • Tissue depths at these landmarks are referenced • Based on these landmarks and tissue depths, the face is built by interpolation or morphing
Soft tissue thickness prediction • The critical issue is soft tissue thickness prediction • All known facial reconstruction systems collect tissue depths at predefined landmarks of a population and use the average value in the reconstruction. • There are some methods to collect tissue depths information • Invasive technique: needle technique • Non-invasive techniques: ultrasound, MRI, CT
Problem and Motivation • Lack of a suitable method for soft tissue thickness prediction • The models generated automatically by the systems are often modified manually by experts to have the final model. • Do facial features such as eye and nose shapes depend on the skull? • Many researchers have been working to prove that these features depend on the bones.
2 3 4 Our Approach Evaluation and Result Conclusion and Future Work Contents 1 Background and Motivation
Our Approach • We treat the soft tissue thickness prediction issue the missing data problem • The tissue depths are the target • The skull shape is the input • One solution for the missing data problem is to • Collect data of pairs of input and target • Apply machine learning techniques to obtain a model that best describes input-target relationship
Data Description: Target Data • Soft tissue thicknesses at facial landmarks • landmarks are believed to contain the most information about the skull shapes • Facial features: distance between 2 pupils, nose height, and nose length.
Data Description: Input Data • Distances between facial landmarks • These distances must be measured from the skull itself. In the case of facial reconstruction, only these distances are known
Data Description (cont) Need to predict these information These information can be measured from any discovered skull
Data Collecting • Available techniques • Invasive: needle technique • Non-invasive: ultrasound, MRI, CT • Needle technique can not be used on living people, so its accuracy is very low • Needle technique can not be used to measure distances between facial landmarks, either. • Among the non-invasive techniques, CT is the best option • CT image is in high quality and the distances between facial landmarks can be measured visually.
Data Collecting (cont) • Collected data of 220 Vietnamese candidates • 98 males, 122 females • Age varies from 17 to 82 • Weight varies from 38kg to 75kg
Discovery of Relationships • Two methods • Linear Regression • Fitting a linear equation to the observed data • Can only discover one to one relationships • Artificial Neural Networks • Non-linear nature • Work in two modes, training and testing
Applying Linear Regression • Linear Regression is applied for all pairs of input-target • For each target field, the pair with best performance (in term of MSE) is chosen
Neural network approach • This structure can represent any functional relationship between inputs and outputs if the hidden layer has enough neurons • two-layer feed forward network • tan-sigmoid transfer function in the hidden layer • linear transfer function in the output layer • Every target field has its own neural network prediction model
3 4 Evaluation and Result Conclusion and Future Work Contents 1 Background and Motivation 2 Our Approach
Evaluation • Ten-fold cross validation • Linear Regression and Neural Network techniques were implemented • The results are compared with the ‘average method’
4 Conclusion and Future Work Contents 1 Background and Motivation 2 Our Approach 3 Evaluation and Result
Conclusions • Proposed an approach for automatic discovery of relationships between anthropometric features • Critical in facial reconstruction systems • Evaluation results show that the approach has better performance than the ‘average method’ which is used in most facial reconstruction systems • Although the data is collected from Vietnamese, the study can be applied for any race
Publications • QuangHuyDinh, ThiChau Ma, The Duy Bui, TrongToan Nguyen, DinhTu Nguyen. “Facial soft tissue thicknesses prediction using anthropometric distances.” Studies in Computational Intelligence(351/2011), page 117-126, Springer, (2001) • ThiChau Ma, DinhTu Nguyen, QuangHuyDinh, The Duy Bui. “3D facial reconstruction system from skull for Vietnamese.” Proceedings of the 3rd International Conference on Knowledge and Systems Engineering, page 120-127, IEEE, (2011).
