The Critical Consideration Behind the Hardware of Body-Worn Camera and Images

When we think about body-worn cameras (BWCs), we generally think about the little box that sits on the policeman’s jacket or the motor-cyclist’s helmet but there are several more things that need to be specified beside the lens and the pixels.

We need to know how the information is going to be used as this will dictate the features of the device.

For example, we may ask if we want to use facial recognition software, for people directly in front of the camera or at some distance. We may ask if we want to see what is directly in front of the camera with good clarity or do we want a panoramic view, say closer to 180°. It is a mistake to compare human vision with machine vision as human vision has developed for specific functions unconnected with the needs of a body-worn device. A BWC can accurately log anywhere between 50° and 120° field of vision (FOV) depending on the choice of model in a perfect cone (as opposed to a human’s binocular FOV of 114° side to side and 35° top to bottom). Also, considering “tunnel vision”, commonly when under intense pressure or stress, people’s FOV narrows further.

As the requirements grow so do the number of pixels required. As the pixel requirement increases so do the problems of file size. Should we store the data or stream it to the cloud? If the latter, we need to consider signal availability and strength.

Moreover, if we stream the data, we need to consider ‘chain of custody’ otherwise our record is dubious when confronted with the rigour of the law. Storing data may not be the problem that we once thought it might be with the introduction of Sandisk’s 1TB memory card. That’s still a lot of memory cards if we calculate for 500 officers over a two-year period (which is the length of time such records should be kept), then there needs to a searchable archive of 120,000 memory cards.

We also need to consider who controls the recording. It is perfectly feasible for the officer wearing the camera to have complete control, and this is generally the norm, but it is also not unreasonable for staff based back in the office or station to decide when to start recording, based on feedback from the officer on-site or from an audio input received. It is equally feasible for every camera to start recording when shift starts and stop when shift ends although that might result in deterioration of working relations between officer and supervisors.

We need to consider battery life, file format, date stamping, recording speed, editing prohibition, weight of camera (let’s remember the poor police officer is already lugging around nearly 10kg of other equipment), water-resistance, low-light capability and numerous other factors such that the technical specification for such systems, because we are talking about complicated systems now not just the front-end camera, become detailed and quite voluminous.

Strangely enough we need also to consider concomitant audio recording and the quality of the microphone which is a bit more complicated than might initially be thought. Audio recordings are pretty much taken for granted since the invention of the phonograph in the late 1870s. The Frenchman, Édouard-Léon Scott de Martinville, in 1860 recorded a lady singing but it was ghastly. Edison’s in 1878 was a tad better (and still a lot better than anything Justin Bieber has produced) but it took until the 1920s for recordings to be accepted as faithful reproductions. Unfortunately, modern technology has superseded those early efforts and now anybody’s voice can be reproduced saying things that they never said. Generally, background sounds can be used to verify the correctness of a recording but even these are not sacrosanct so the ‘chain of custody’ becomes of paramount importance.

What is actually required is a combination of all of the above. From this we can derive that the more pixels can be better for the viewer however can sometimes undermine police officer’s perspective as today’s camera sensors are superior to most police officer’s vision. The same can be said for a greater angle width.

But the data itself becomes of almost secondary importance when we consider all the other constraints imposed on the collection, retrieval and submission for action of such data. For more information on Valour Consultancy’s latest report on the enterprise body-worn camera systems report, click here.