Animation Results

Figure 17.17 shows selected frames of sample animation. The top row shows video images of the original speaker's face during a motion-capture session using the OPTOTRAK (Northern Digital Inc., Canada) system. This motion-capture data drives the original speaker's face model (second row) by the individual PCA-based technique discussed in Sections 17.2.4 and 17.2.5 (Kuratate et al. 1998). Then the linear-combination values a of this model are mapped to a face model generated from photographs...

Application Of Explanationtesting Methodology To The Catch Model

Here I will show that the application of the explanation-testing methodology to the experimental situation of the catch model can lead to self-contradiction and failure to reconstruct the target face. I assume two hypothetical situations and two performance rules primary and alternative. While the primary rule does lead to a successful reconstruction of the target face, the alternative rule does not. Situation A The correspondence between the primary performance rule and the participant's...

Averaging

The averaging procedure had several steps, as illustrated in Figure 6.4. First the computer generated an average delineation map from the individual delineation maps by calculating the average x and y coordinate for every point along the contour lines. The computer then warped each individual's face into the average template using a multiscale algorithm (Tiddeman 1998). The average color (with red, green, and blue components) of each pixel was calculated to produce the initial average face...

Collection And Processing Of Ct Database

A problem of basic forensic importance is to determine the identity of human skeletal remains. For this purpose, the skull often gives the best available information. One approach to the identification problem is to superimpose images of a person's face onto a given skull (Aulsebrook et al. 1995, Shahrom et al. 1996, Yoshino et al. 1997). This is essentially a method of verifying a hypothesized identification by determining the goodness of fit of a photographic image of a person with a...

Comparing Facial Images

Each 2D facial image obtained from the 3D facial image of subject No. 64 was compared to the adjusted 3D facial images of 132 subjects in each database set by FaceList. As shown in Figure 19.9, the 2D facial image of the target person is shown in the upper left corner of the retrieval result image. The 3D facial images of the database were ranked in batches of twenty and could be scrolled (Figure 19.9). The file name, rank, and status are shown in the lower left corner of the retrieval result...

Computer Software Solutions

In the ten years since subtraction radiography was only an aid to subjective analysis, much work has been accomplished. There are now reportedly no fewer than 38 computer applications suitable for comparing radiographs using applications of algorithms (Lehmann et al. 2000). An algorithm is a finite series of clearly defined logical steps to solving a problem. In this application, mathematically specific techniques are used to implement image-processing The white arrows indicate areas of barely...

Computerassisted Facial Image Superimposition

To make the comparison between a 3D facial image of a suspect and a 2D facial image taken at the scene of a crime, the 3D facial image is first reproduced on The fine framework images reproduced from the 3D data shown in Figure 18.8. (A) frontal view, (B) oblique view. The fine framework images reproduced from the 3D data shown in Figure 18.8. (A) frontal view, (B) oblique view. the display of the host computer from the MO disk and then the 2D facial image is taken with the color image scanner...

Face Estimation From Partial Face Feature Information

In this section, we describe a general approach to 3D face estimation using the 3D face database, and address the specific case of estimation from photographs. If partial information of a face is obtained, it can be used to estimate the entire face structure using a known relation between a subset of data points and the entire face. Time-series plots of the top six principal components used to synthesize the talking face after parametrization by the facial-marker positions at each time step....

Development Of The Method

The 3DCeph computer software running on a standard PC was able to correlate digitized landmarks seen from different perspectives and projections and convert the 2D location of each landmark into its 3D spatial coordinates (x, y, z). From those landmarks and their spatial coordinates, 3D wire-frame meshes can be established by connecting those landmarks in triangular fashion, creating multiple-unit surfaces. Finite-element modeling can be obtained following the creation of the wire-frame meshes...

Digitalization Of The Bone Fragments

The goal of this first step in the working process is to produce a true-to-scale 3D digital copy of the bone fragments. The following requirements have to be met by the measuring system in order to assemble the digitized bone fragments using a CAD program High local resolution of the bone fragments to capture the precise shape of the fractured surface as well as the anatomical features (e.g., slightly rough surface structures, suturae) Detailed documentation of the surface color Multiple task...

Discussion

The laser surface scanner generates accurate 3D data. The availability of a lightweight user-friendly surface laser scanner in the field of orthodontics makes it possible to analyze growth, soft-tissue changes, treatment simulation, appliance designs, and treatment effects in three dimensions. Three-dimensional Sample of three-dimensional head model taken from multiple views to eliminate undercuts using special headband to ensure precise integration of the segments. Sample of three-dimensional...

Estimation Of Face Postures

Once a 3D face model and surface texture are acquired, we can employ the database to estimate other facial expressions. In the principal-component space obtained from analyzing all faces in the database, individual face postures form a cluster around the neutral face, as shown in Figure 17.13. By describing the unknown input face in this space, similar faces in the database can be found by simply measuring the distance between the input neutral face and any neutral face in the database. Once...

Expression Estimation Results

Figure 17.14 shows estimation results for eight expression postures starting from the neutral face in Figure 17.8. We also applied this method to inanimate models, including a statue of Venus, which was scanned with the Cyberware scanner see Figure 17.15(a), like the human subjects. Of course only one posture was available, and a generic mesh was adapted to the data, resulting in the 3D model shown in Figure 17.15(b). Expression estimation results Figure 17.14 Eight postures estimatedfor...

Facial Image Database Sets

As model suspects, the 3D facial data of 132 Japanese male adult subjects were obtained using Fiore and stored in the host computer. In order to test identification performance of the improved software FaceList, the 2D facial images which showed various angles were made from the 3D facial image of subject No. 64 by 3D-Rugle3 for Face-To-Face. The right oblique 2D images were obtained at every 10 degrees (10, 20, 30, 40, 50, 60 degrees) from the 3D facial image at the horizontal position (Figure...

Forensic Imperatives

The success of medico-legal investigations of death in the past decade result, in part, from methodological developments within many research domains of forensic science. However, the recent marked increase in the popularity of all things forensic is also peripherally responsible for an active media feeding the public's fixation on a mystery and society's innate morbid curiosity. A symbiotic relationship flourishes between the media, science, the medico-legal community, and the public that...

Generation Of The Average Face

We used 130 standardized face photographs of females self-reporting the socially perceived group membership of European (mean age 22.3, sd 4.0 years), to generate average faces used here to represent the face color and texture of the constructed facial approximations. This average face was generated using the refined methods described in Chapter 6. The average female European face is presented below with its respective delineation map (Figure 11.1). The average socially perceived European...

Introduction

As emphasized by Bramble et al. (2001), two and three-dimensional computer-based reconstruction systems have been developed to make the reconstruction process faster, more flexible and to remove some of the subjectivity and inconsistencies associated with the traditional approaches (illustrative identikit and 3D clay based reconstruction). We can attempt to classify the 3D computer-based methods that have been presented in the last fifteen years into the three following categories. The user...

Manual 3d

A distinction is made between the anatomical (morphological, morphoscopic) method and the morphometric (sculptural) method. In the anatomical method, the muscles, salivary glands, and fat are positioned and then covered over by a layer of skin. The morphometric method sculpts the face following indications of the average thickness of certain salient anthropological points. Most researchers choose this method as it is probably easier and it has not been demonstrated scientifically that one...

Materials And Method

The purpose of this study is to develop a noninvasive, economical, and reliable method for measuring human facial soft-tissue thickness using 3D finite-element modeling. Custom computer software (3DCeph 2000, Department of Orthodontics, University of Illinois, Chicago, Illinois) using a stereopho-togrammetry algorithm has been found to be accurate to 1.5 mm for linear measurements. From 38 hard-tissue main landmarks and 36 soft-tissue main landmarks, 97 hard-tissue vectors, and 159 soft-tissue...

Applications Of Computergraphic Facial Reconstruction

CHAPTER 16 INVESTIGATION OF ETHNIC DIFFERENCES IN FACIAL MORPHOLOGY BY THREE-DIMENSIONAL AVERAGING Ashraf I. Shaweesh, C. David L. Thomas, and John G. Clement 17 ESTIMATION AND ANIMATION OF FACES USING FACIAL MOTION MAPPING AND A 3D FACE DATABASE Takaaki Kuratate, Eric Vatikiotis-Bateson, and Hani Camille Yehia FACIAL IMAGE IDENTIFICATION SYSTEM BASED ON 3D PHYSIOGNOMIC DATA 347 Mineo Yoshino 19 A NEW RETRIEVAL SYSTEM USING A 3D FACIAL IMAGE DATABASE Mineo Yoshino It is the common wonder of all...

Placing The Landmarks

Small pyramids were used to represent the traditional tissue-depth markers utilized in clay reconstructions (Figure 10.3). The square bases were placed perpendicularly to the bone surface at the appropriate craniometric points on the skull. The height of the pyramids was input in millimeters to the measurements specified in the literature (Rhine and Campbell 1980). For clarity, each pyramid was individually renamed for the point it represented (i.e., nasion-right or glabella-left). For the...

References

H. (1993) Inversion and Configuration of Faces, Cog. Psychol. 25, 281-316. Bartlett J. C., Searcy J. H. and Abdi H. (2003) What are the routes to face recognition , in Perception of Faces, Objects, and Scenes Analytic and Holistic Processes (M. A. Peterson and G. Rhodes, eds.). Oxford University Press, Oxford. Biederman I. and Kalocsai P. (1998) Neural and Psychophysical Analysis of Object and Face Recognition, in Face Recognition From Theory to Application (H....

Registration Of Images

The software uses a reference image and a follow-up image for image registration. These images are analogous to antemortem and postmortem images in forensic identification cases. Using the appropriate tool, the anatomical points are selected in the same order in both images. The points define the area of interest (AOI) that the software will analyze (see Figure 13.3). After registering and overlaying the two images, the software de-emphasizes similar pixels and enhances dissimilar pixels,...

Repairing The Damaged Skull

When skeletal remains are presented in a fragmented condition, this poses a challenge for the reconstructive sculptor. The skull may have been damaged by trauma, burning, or environmental or animal activity. The bones of the cranium and mandible may be fragmented in many places, and portions of either may be lost. The bones may also be in an extremely friable condition requiring great care if further damage is to be avoided. Such fragmented skulls can be rebuilt and held together with the aid...

Results And Future Work

The face of the Egyptian mummy generated using this technique is shown in Figure 10.7. As can be seen in the figure, there is a pronounced overbite on the upper lip. Initially, it was thought that this could be due to the lower mandible having sunk into the skull over the ages. However, this hypothesis was refuted by a forensic odontologist who confirmed that the position of the mandible was correct. This overbite on the upper lip particularly excited the Egyptologists, as the royal family of...

Rigid Alignment Of Skull

In order to standardize the database, each set of skull and face surfaces is aligned with a canonical model. This process starts by establishing a set of point correspondences between the scanned skull and the canonical skull. Once this is done an optimal metric transform (rotation, translation, and uniform scaling) is computed (Horn 1987). This processing step is performed on all scanned data. Figure 8.4 illustrates this process. This operation defines an overall alignment of the skull (and...

Skull Depth Alignment

Once the head-scan pixels have been aligned with the subject pixels, an adjustment must be made on the r intensity values. The approach taken in this paper is to define an adjusting surface 8r (0, h). Each head-scan pixel i with coordinates (0i, hi) will have its pixel intensity adjusted by adding the value of 8r (0i, hi). Once again the relationship between the head-scan skull and the subject skull will be used. The value of the adjusting surface cannot be measured for all points, since the...

The Dr Method Three Conditions For Reconstruction Of Past Events

Having described the catch model, I shall now describe the D-R method's three conditions whose fulfillment ensures successful reconstruction of a PE (i.e., the target face) from memory. The D-R method provides essential conditions for judging whether a given performance rule and an analysis rule will result in reconstruction of the PE from memory. The method is a procedural schema, in which one inserts a particular performance rule (such as the similarity judgement used in the catch model) and...

Threedimensional Quantification Of Facial Shape

School of Dental Science, University of Mebourne, Victoria 3010, Australia Anatomy is a subject that is going through a transition from being entirely descriptive to being a quantitative science based on measurement. The delay in transferring methods that are common in the physical sciences to biology is partly due to the innate complexity of biological systems and partly to the difficulties inherent in making sufficiently accurate measurements on the very complex objects created by biological...

Warping Of The Average Face

Three female individuals (not included in the average), self-identifying as Europeans, from Melbourne Australia, were photographed and delineated as described in Chapter 6. The average face was then warped to the exact same frontal face shape for each of these three individuals using the averaging software developed by Perrett and colleagues (Benson and Perrett 1992, Penton-Voak et al. 1999, Perrett et al. 1994, Perrett et al. 1998, Perrett and Rowland 1995, Tiddeman et al. 2001) see Figure...

What The Skull Can Tell Us Positional Relationships Between The Skull And The Facial Features

Facial features such as eyebrows, eyes, and eyelids, and to a lesser extent the ears, nose, and lips form the recognizable features of the face. As they are variable from individual to individual, only general information can be gleaned from a study of a population, particularly that of mixed ethnicity. The published literature has focused on single ethnic groups and some of the following generalizations need to be varied to account for differing populations. 3.4.1 THE RELATIONSHIP BETWEEN THE...

Memory Research Has Two Different Goals

Since the early 1980s, an expanding line of research in errors and distortions of memory has emerged (for a critical review see Koriat, Goldsmith & Pansky, 2000), encompassing the following eleven research areas Gestalt approach to changes in memory over time, spatial memory and distortion, memory for gist versus detail, schema-based effects on memory accuracy, source monitoring, illusions stemming from fluency misattribution, false recall and recognition, misleading postevent information,...

Face As A Microsoft Excel Sheet

3D-Rugle3 provides a facility for the export of facial scans to a comma-separated-value (.csv) file that can be imported into a Microsoft Excel spreadsheet. The inherent limitation of Excel for this purpose is that there are a maximum of 256 columns in a spreadsheet and thus exported images must be reduced in size by a scale factor of 0.5. After all images had been aligned they were saved as 256 X 256 arrays in Excel sheets, the number in each cell being the z value of the surface at the (x, y)...

Photography

Participants were photographed, in smiling and neutral expressions, in both frontal and right side profile, on a specially designed craniofacial photography rig that is similar to that described by Dobrostanski and Owen (1998). The rig is a permanent structure that enables frontal and profile photographs to be taken simultaneously (see for images Stephan 2003, Stephan et al. 2004). The rig uses two cameras that are fitted with 105 mm 1 2.8 Macro Nikor lenses, and the camera shutters are...

Appendix

Table of landmarks for the profile delineation map (Figure 6.3(b)). Left-side bilateral landmarks and landmarks falling midway between two other landmarks are not shown. Standard anthropological landmarks are shown in bold type. point at corneal junction with lower lid point at corneal junction with upper lid most lateral point of epicanthal fold most anterior point of epicanthal fold most anterior point on upper vermillion border labriale superius approximation (ls) point at junction of upper...

How Successful Is Facial Approximation

If the bones of the skull and the soft tissues of the face are interrelated, it stands to reason that the anatomy of the former will predict that of the latter, but is it predictable There have been many successes reported in the literature, and forensic sculptural approximation remains a popular if last-ditch attempt to identify unknown remains (Prag and Neave 1997b). Given all the problems associated with the reproduction of soft-tissue structures, it is perhaps not surprising that facial...