Photo Forensics

[John Dolva]

I'm starting this post as a response to questions on photo analysis. I don't know if the term 'photo forensics' already exists, perhaps in another format. Nevertheless it seems to me as this is something of interest to many it might be an idea to pool thinking on it.

[John Dolva]

i dont know quite how to kick this off so I'll refer to a recent post as illustration.

the image in the first row was presented as a photo.

it became an issue to determine forgery. while I'm not about to make a statement on whether it is or not. I'd like to point out how the human eye may be a poor judge of such things. Firstly, the image appears to be a black and white one, however it is not presented as such. In the first row below, the first image is split into its RGB values. As it appears, this is weighted evenly in all splits. Hence it's reasonable to say it is 'black and white'.

Actually it's grayscaled. Grayscale is a spectrum of black to white in 256 steps. The images following in the second row are stepped down from 255 to 230 in roughly even increments, the human eye cannot accurately distinguish between the grades in the extremes of the spectrum, but one can use computers to isolate these and therfore answer such questions as 'which part is whitest'.

To a large extent, in this endeavour (the murder of JFK) which contains a large body of images, these things are important. Further, photographers generally are also aware that colour values influence how adjacent values are percieved by the eye. Digital analysis can aid in separating such influences and make accurate analysis more likely.

Also it is surprising how visual clues as to depth can decieve the eye as to true dimensions of object in a 2d representation of a 3d scene.

Edited July 9, 2005 by John Dolva

[John Dolva]

with reference to the photo posted by Jim Root in Stella Polaris http://educationforum.ipbhost.com/index.php?showtopic=3981

The figure standing behind Walker as well as the figure behind prince Olav both have facial dimensions outside the norm. when as shown here in this analysis the

faces are cut out and added to or subtracted from the photo the dimensions 'get right' and perhaps can be used to identify them. Once faces has been pasted, the only way i'd imagine that the person they are masking can be identified would be by looking at the uniforms, recollection of others in the photo or seeing an unaltered photo. I suppose there is a possibility that these faces while out side the norm are unaltered as well. Anyway the point here for me is that known anatomical features like facial proportions can be used to determine alteration.

A minor point perhaps, but significant in looking at the backyard photos, where by the way so far I can't see evidence of alteration. Not to say there is none.

John

[John Dolva]

I find the whole xray thing hard to understand. The anatomy and radiography texts that I have read basically leads me to say that the xrays as available to me appear to be useless for serious analysis. For example, using the scale provided in a text on forensics that puts lead as the most opaque substance the following images tell me that the bullet fragmentation smearing is different from that of the reports. These are 3d 'graphs' of the grayscale values of the side xray show that the most opaque area is a broad vertical smear on the rear half of the skull. yet another reason for release of quality info to researchers.

the middle image is a view top down on graph, the first is a view of graph from the bottomof the xray, and the third a view from the front. (I hope these are easy to understand, I can twirl the graph about and view it from any angle to take these snapshots, so I'm familiar with them, is it an OK way to present info. to someone who hasn't seen it before?)

[John Dolva]

There is more of information in the films and photo's than one might think. For example in another topic there is indication that speeds of movements of objects can be derived by analysing blurs (Limousine stopped http://educationforum.ipbhost.com/index.php?showtopic=4280 ) and revolutions of wheels etc . Another approach recognises the 'film's within films', for example flipping the zfilm shows indistinct images of what may be going on outside the range of the frame. Here is a flipped/cropped part of zf313 showing what may be the s-e corner of Dealey plaza. See reflection in rear of limo.

Edited July 10, 2005 by John Dolva