Photogrammetry is the science of determining measurements from photographs. It’s most general use is in surveying and mapping, but, in the forensic context, it is also used to determine measurements of crime scenes and in accident reconstruction; and even more specifically, it can be used to determine measurements of the physical size of perpetrators from surveillance camera images. In a simple two-dimensional situation, here is basically how it works:
Now, we don’t live in a two-dimensional world, so photographs will contain varying degrees of what’s called “perspective”. It’s critical to understand, and account for, this phenomenon when doing a photogrammetric analysis, so here is a brief summary of visual perspective:
Visual perspective is the phenomenon whereby objects appear to diminish in size as they get farther away from the viewer (or camera). For example, when looking down a long section of railroad track, the two rails “appear” to get closer together the farther down the track one looks. Many architectural renderings are drawn in perspective to show what a building will actually “look like”. The following drawing shows a staircase drawn in perspective.
This is what the staircase “looks like” when we see it. Focus on the very bottom step. We know that the two long sides of the step are actually parallel, but, as shown, if these sides are projected back away from the viewer, the two “rays” ultimately meet at a point called the “vanishing point”. And as the diagram shows, this applies to every straight-line edge in the drawing.
Now, just a word about surveillance cameras – surveillance cameras, in order to achieve a wide field of view, use short focal length lenses. As a consequence of this, the images captured by the camera will contain a fair degree of “perspective”. This is readily seen in the frames of a surveillance video that captured a robbery in process:
Straight-line edges that we know are actually parallel are not parallel in the captured frame, and they can be extended back to a vanishing point. Below is a link to the photogrammetric analysis I did of this photo to determine the height of the perpetrator:
The reference dimension for this analysis is the inside height of the door frame from the floor to the inside of the top frame, which was actually measured. You can see where perspective lines have been added to bring this known dimension back and over to where the perpetrator is standing. Allowance has been made for the fact that the head is bent forward, and he is wearing a ball cap. The determination is that the perpetrator (in bare feet) is 5’11” tall, ±¾”. Interestingly, the person currently serving prison time for this robbery is 6’4″ tall – and yes, we are working that case.
Here is one more simple example. This diagram was used to determined the height-above-road-surface of an entry wound in the head of the driver of a Jeep CJ7. Since all points of interest are very close to being in the same distance plane from the observer, this is basically a two-dimensional analysis.
The reference dimension here is the known 93.4″ wheelbase of the 1983 Jeep CJ7. The determined wound height distance was then used in a bullet trajectory analysis to determine the point of origin of the gunshot.
There are many photogrammetric software applications for sale that can range as high as several thousand dollars, but the analyses shown in this post were done “by hand”. Plus, I prefer to make my own judgements about where a pixel starts and where a pixel ends.
The American Society for Photogrammetrics and Remote Sensing (of which I am a member) has much information on their website: ASPRS.org ; however, most of it is related to mapping and surveying.
Forensic photogrammetry can be a useful tool, and it’s NOT junk science – it’s mostly just plain old geometry and trigonometry.