Review ArticleConceptual transitions in methods of skull-photo superimposition that impact the reliability of identification: A review
Introduction
Since the innovation of the skull-photo superimposition method reported by Professors Glaister and Brash in 1935 [1], the photographic process of image overlay has undergone significant technical progression [2], [3], [4], [5], [6], [7], [8], [9], [10] (Table 1). It has evolved into a real-time video process [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30] (Table 2) that has evoked considerable research aimed at automating the procedure using computers and software [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42] (Table 3). A much-cited report on the reliability of the video superimposition method by Austin-Smith and Maples [22] indicated a failure to exclude identifications of approximately 9% when comparing three skulls with 100 front-view photographs of living individuals. Recently, Gordon and Steyn [42] superimposed 40 digitized facial photographs with 3D digital images of 10 skulls each (a total of 400 superimpositions were done for each of the morphological and landmark methods) and observed that the failure to exclude (false positives) was 17.3% and 32% for the morphological and landmark methods, respectively. In addition, these authors [42] reported that the failure to include (false negatives) was 15% for the morphological method and 20% for the landmark method. The fact that skulls failed to match their related faces during superimposition in 15–20% instances appears counterintuitive. This finding raises troublesome questions because it has been relatively well accepted that the skull is the matrix of the living head … and face in life [43] and creates the architectural form of the head and provides the basic structure for the face [44], which is a concept shared by many other authors [20], [45], [46]. Furthermore, in view of the number of real-world cases in which the superimposition method has provided reliable results, the rate of failures reported by researchers [22], [42] appears inflated and has prompted a review of the concepts underlying the methodologies recommended by practitioners and those used by researchers to identify divergences, if any.
Section snippets
Practitioners’ and researchers’ methods: specific conceptual variances
The reported instances of applications of the superimposition method for identification purposes in routine case work include 25 in the CA Pound Laboratory, Florida [25], 52 in Japan [23], 108 in China [47], 71 in Hungary [48] and 251 in India during 1970–1989 [18], which increased to 1800 during 1990–2010 [49], together totaling 2307. A recent national level survey in India revealed that the number of superimposition-based court testimonies in the Tamil Nadu state during 2005–2009 was 200 [50]
‘Life-size’ face-image and skull orientation: the two critical requirements
Fundamentally, skull-photo superimposition is an image-mixing process that helps examine the appropriateness of the salient features in the image of the skull in question (that has been anatomically oriented to correspond to the posture in the face-image) for the missing individual's face-image that has been magnified appropriately; the classical example is that described by Glaister and Brash [1]. Subsequent authors describing the photographic process of overlaying transparencies [2], [4], [6]
On the concepts and methods relating to the use of ‘life-size’ enlargements
When discussing the conceptual basis for using ‘life-size’ images, Glaister and Brash [1] concluded that it was a “safer plan to make the enlargements of both the skull and portrait natural size, the former exactly so, the latter as near as might be feasible” because it would form “a more crucial experiment” than to fit the skull and face images without regard to their actual sizes. The concept underlying the use of ‘life-size’ images is that a comparison achieved by overlaying two physically
Conceptual departure when using ‘life-size’ enlargements
Ubelaker et al. [32] adjusted the size of the face-image until it filled at least 67% of the monitor screen, which departed from the use of ‘life-size’ images during superimposition. Subsequently, Delfino et al. [33] centered the skull-image on the monitor at approximately two-thirds the size of the screen. Seta and Yoshino [21] and Yoshino et al. [23] used half-size skull-images for superimposing because the skull image was focused on ground glass of that size. The work by Austin-Smith and
Conceptual basis for orienting the skull: use of points at two different planes
Glaister and Brash obtained the ‘primary’ orientation of the skull by either applying the angle of orientation of the tiara observed in the ‘life-size’ photograph of the skull or setting the skull to correspond to the salient features observed in the face in the ‘life-size’ transparencies [1]. These authors [1] recognized the requirement for slight movements of the skull to obtain its final orientation. The practice of achieving an appropriate orientation by overlaying the transparencies of the
On the use of Whitnall's tubercles during superimposition
Practitioners have long used orbital measurements from the skull to orient it relative to the posture of the face [11], [15], [24], [26], [27], [30]. However, there has been a lack of uniformity in the point in the orbital zone that is used. Morphological placement of the eye, i.e., placing it in conformity with the orbit, may shift the eye angle in the face away from the location of the Whitnall's tubercle in the orbit, as shown by some researchers [40], [42]. Chandra Sekharan [5] assumed the
Transitions in the criteria for assessing goodness of fit
Glaister and Brash relied upon salient features, such as chin height, for evaluating fit while superimposing skull tracings and face-images at ‘life-size’. Their reliance of cranial landmarks is limited – such as the use of prosthion seen exposed in the face photographs [1]. Of the seven authors reporting on photographic superimposition, six have relied on morphological features, of whom two have included asymmetries (Table 1). Of the 21 studies and case reports of video superimpositions, all
Conceptual basis for the reliance on cranial and facial landmarks
The research of Gordon and Steyn [42] has indicated, for the first time, that when other parameters are equal, the landmark method performs less reliably compared with the morphological method. Thus, when using any anthropometric landmark, the following must be considered: (a) the size of the marking of the landmarks in relation to the magnification of the images, (b) the ability to mark the landmarks on the face- and skull-images repeatedly, (c) the discriminatory power of the landmarks over
Fade and wipe resources for assessing the goodness of fit during superimposition
With regard to the fade (mix) and wipe resources available when using image-mixing devices, it has been well accepted that a very slow fade may provide the illusion of a perfect match during the period that one image is almost imperceptibly replaced by another [79], [80]. İşcan [80] cautioned that the fade approach must be avoided and recommended an initial complete wipe to be followed by a dissolving or mix analysis. Many authors have found the wipe (sweep) technique more useful [13], [15],
Reliance on asymmetries while assessing the goodness of fit during superimpositions
Arguably, in any superimposition that considers morphology, an asymmetry would naturally be included to help assess whether a match exists. However, Glassman [81] conceded that professionals who are typically trained to assess the size and conformity of anatomical structures are likely to miss aspects of contour or asymmetries that require an artistic ‘eye’ and suggests involving a facial reconstruction artist when assessing asymmetries while evaluating video superimpositions. Indeed, some of
On the use of 3D digital images of skull
Gordon and Steyn indicated that the general issue in obtaining high-quality digital scans and photographs may also contribute to the relatively poor accuracy reported in their study and considered that traditional, more laborious methods using video cameras, for example, may yield better results [42]. The authors observed missing areas in the 3D skull-images, which were attributed to the inability of the scanner to scan curved surfaces in the skull [42]. The use of digitized 3D images of skulls
Conclusion
The concepts relating to superimposition method proposed by early reports summarized here and have been accepted by several practitioners when identifying skulls in real-life cases include the use of ‘life-size’ images as a “safer” procedure [1], the use of auditory meatus-ear relationships in addition to orbital relationships while orienting the skull [5], a reliance on wipe images permitting the “better comparison of bony points” compared with the “more pleasing” mix images [13], and the
Acknowledgements
Universiti Sains Malaysia is acknowledged and thanked for the financial assistance through RU grant 1001/PPSK/813011 and Short Term Grant 304/PPSK/61312009 which supported this review.
References (95)
Identification of dismembered part: the medicolegal aspects of the Nagaraju case
Forensic Sci.
(1973)The positioning and magnification of faces and skulls for photographic superimposition
Forensic Sci. Int.
(1989)- et al.
Evaluation of anatomical consistency in cranio-facial superimposition images
Forensic Sci. Int.
(1995) - et al.
Facial identification: skull-photo superimposition
- et al.
Cranio-facial morphanalysis: a new method for enhancing reliability while identifying skulls by photo superimposition
Forensic Sci. Int.
(2001) - et al.
Skeletal remains of Dr. Eugenio Antonio Berrios Sagredo
Forensic Sci. Int.
(2001) - et al.
Victim of a dictatorial regime: identification of Mr. Roberto Gomensoro Josman
Forensic Sci. Int.
(2005) - et al.
The identification of a dismembered human body: a multidisciplinary approach
Forensic Sci. Int.
(2003) - et al.
An unusual case of cranial image recognition
Forensic Sci. Int.
(2005) - et al.
An experimental study on the applicability of evolutionary algorithms to craniofacial superimposition in forensic identification
Inf. Sci.
(2009)