-------- Original Message -------- Subject: Re: Re: right/left looking photos for human face shape Date: Thu, 3 Nov 2011 11:04:03 -0400 From: [email protected] Reply-To: [email protected] To: [email protected] I fully agree with Dean: the solution is data dependent. In the small dataset I played with, we had landmarks only on half of the face (left side, I think) and that's possibly why part of the differences due to angles were uniform. Also, it was a forensic application where all it mattered was to 'get the target right' (no interpretation of biological variation), and removing the uniform component increased the accuracy with which individuals were matched regardless of head rotation. Cheers Andrea -----Messaggio originale----- Da: morphmet <[email protected]> Inviato il: 3 Nov 2011 - 10:25 A: morphmet <[email protected]> -------- Original Message -------- Subject: Re: right/left looking photos for human face shape Date: Thu, 3 Nov 2011 09:52:07 -0400 From: Dean Adams To: [email protected] Mark, I played around with this issue (parallax) years ago, and agree that the orthogonal projection approach off PC1 by Gharaibeh is often a useful solution. As mentioned by Andrea, regression residuals may also do the trick. However, I found that the best solution is often data-dependent. Thus I recommend exploring both possibilities with your data to determine whether the lack of standardization of face position generates a predictable pattern (see Fig 1 in Gharaibeh). That said, I would caution against eliminating the uniform component from the analysis. While this may improve the fit in some circumstances, it will also of course eliminate all uniform shape variation, which can be considerable in some datasets. Additionally, it turns out that the shape distortions generated by non-standardized specimens in front of the dig itizing plane are often highly non-linear, such that most of this variation will not be embedded in the uniform shape components. In which case, eliminating the uniform component will remove actual uniform shape variation, but not variation due to lack of positional standardization. Dean -- Dr. Dean C. Adams Associate Professor Department of Ecology, Evolution, and Organismal Biology Department of Statistics Iowa State University Ames, Iowa 50011 www.public.iastate.edu/~dcadams/ phone: 515-294-3834 On 11/2/2011 5:17 PM, morphmet wrote:
-------- Original Message -------- Subject: Re: right/left looking photos for human face shape Date: Wed, 2 Nov 2011 06:14:49 -0400 From: andrea cardini To: [email protected] Dear Mark, years ago, on a small dataset with the same individual photographed under two different angles, I managed to control for the angle by simply using residuals of a regression of shape onto the angle. By excluding the
uniform
component of variation from the regression, the accuracy of the model was improved. Likely a large amount of differences between the two pictures were approximately uniform. Possibly you may be able to do something similar, validate the regression model on subjects for which you have two pictures and then apply it to those for which you have only one picture. If you manage to get a pdf of the chapter mentioned by Dennis, I'd greatly appreciate if you could send me a copy. Thanks in advance. Cheers Andrea At 14:33 01/11/2011 -0400, you wrote: > > -------- Original Message -------- > Subject: right/left looking photos for human face shape > Date: Tue, 1! Nov 201 1 00:10:00 +0000 > From: Mark Belk > To: [email protected] > > > > > Does anyone have a suggestion for how to correct for right/left lookers > in human face shape. A colleague has a series of historical photos of > people with known pedigrees, but many of the subjects are turned a few > degrees to the right or left. Any suggestions on how to adjust for this > rotation in the landmarks? > > Thanks, > > Mark > > Mark C. Belk > > Professor of Biology > > Editor, /Western North American Naturalist/ > > Brigham Young University > > 801-422-4154 > > > Dr. Andrea Cardini Researcher in Animal Biology Dipartimento di Biologia, Universitá di Modena e Reggio Emilia, via Campi 213, 41100, Modena, Italy tel: 0039 059 2055017 ; fax: 0039 059 2055548 Honorary Fellow Functional Morphology and Evolution Unit, Hull York Medical School University of Hull, Cottingham Road, Hull, HU6 7RX, UK University of York, Heslington, York YO10 5DD, UK Adjunct Associate Professor Centre for Forensic Science , The University of Western Australia 35 Stirling Highway, Crawley WA 6009, Australia E-mail address: [email protected], [email protected], [email protected], [email protected] Webpage: http://sites.google.com/site/hymsfme/drandreacardini Datasets: http://ads.ahds.a! c.uk/cat
alogue/archive/cerco_lt_2007/overview.cfm#metadata <http://ads.ahds.ac.uk/catalogue/archive/cerco_lt_2007/overview.cfm#metadata>
Editorial board for: Zoomorphology: http://www.springer.com/life+sciences/animal+sciences/journal/435 Journal of Zoological Systematics and Evolutionary Research: http://www.wiley.com/bw/journal.asp?ref=0947-5745&site=1
<http://www.wiley.com/bw/journal.asp?ref=0947-5745&site=1>
Hystrix, the Italian Journal of Mammalogy: http://www.italian-journal-of-mammalogy.it/
Dr. Andrea Cardini Researcher / Lecturer in Animal Biology Dipartimento di Biologia, Universitá di Modena, via Campi 213, 41100, Modena, Italy tel: 0039 059 2055526 ; fax: 0039 059 2055548 Honorary Fellow Functional Morphology and Evolution Unit, Hull York Medical School University of Hull, Cottingham Road, Hull, HU6 7RX, UK University of York, Heslington, York YO10 5DD, UK Adjunct Associate Professor Centre for Forensic Science , The University of Western Australia 35 Stirling Highway, Crawley WA 6009, Australia E-mail address: [email protected], [email protected], [email protected] Webpage: http://sites.google.com/site/hymsfme/drandreacardini Datasets: http://ads.ahds.ac.uk/catalogue/archive/cerco_lt_2007/overview.cfm#metadata ------------------------------------------------------------------------------- Valore legale alle tue mail InterfreePEC - la tua Posta Elettronica Certificata http://pec.interfree.it -------------------------------------------------------------------------------
