The UK's Gone MadBy: John Honovich, Published on Oct 02, 2012
Unfortunately, the technical claims underlying these assertions are categorically false. The UK surveillance commissioner is either being grievously misquoted, is grossly incompetent or propagating lies to rally opposition to video surveillance.
Review of Claims
Here are the claims made and the reality of each:
Claim - Face + HD: Warns about "facial recognition systems and HD cameras are allowed to proliferate on high streets."
Reality - HD cameras and facial recognition are two entirely different things. There are millions of HD cameras already deployed in the world with 99.999% of them not using facial recognition. Facial recognition is extremely rarely used with surveillance cameras. Any criticisms of the two need to be separated.
Claim - Pick Out Half a Mile: "It's the ability to pick out your face in a crowd from a camera which is probably half a mile away."
Reality - Top of the line PTZ cameras can barely capture faces at 100 meters away with an operator manually zooming in on a person. Capturing a clear image of a face a half a mile away requires very special purpose extremely expensive cameras that are rarely used even by governments. Picking a face out in a crowd a half a mile away is far harder still.
Claim - Super Magic MP: "16-megapixel HD cameras are now very affordable, so people are buying a camera with a huge optical and digital zoom power."
Reality - No one is buying 16 MP surveillance cameras with optical zooms as the one manufacturer offering them only sells fixed cameras. Very few are even buying them at all as they are quite expensive and have limited capabilities (low frame rate, super high storage costs, no 3rd party integration, etc.). Digital zoom is a red herring and has no bearing on any camera's ability to capture images of individuals.
Claim Miracle Panoramics:"A tiny camera in a dome with a 360-degree view can capture your face in the crowd."
Reality - 360, aka panoramic cameras, can barely capture face details just 10 feet away from the camera, forget about crowds accross or down the street. These cameras max out at 5MP while looking in every direction, delivering very little image detail in any one.
Claim Face Rec Accuracy: "I've seen the test reviews that show there's a high success rate of picking out your face against a database of known faces."
Reality - These tests are, ironically, mostly skewed by (the few) vendors (left) pitching these offerings. And the few government tests do not factor in the operational difficulties of using it in the real world. The false matches are typically huge and the requirements for ideal lighting and direct head image capture make the systems essentially infeasible.
Bottom line: The use of HD cameras is growing rapidly around the world but, by itself, provides little risk to the safety or privacy of individuals. The use of facial surveillance is NOT growing rapidly and continues to suffer from major operational problems.
BBC embarrassing Report
As technologically illiterate as the UK Surveillance commissioner sounds in the Independent, the BBC's article might be worse. To justify the surge in HD surveillance, the BBC cites manufacturer figures of 129,299 HD CCTV cameras in the UK by the end of 2012 rising to over 3.7 million by 2016. First, those projections are for the world, not the UK. Secondly, those are the marketing dreams of a fringe trade organization that serious industry professionals do not believe. This projection requires a 130% CAGR for the next 4 years, which would require a miracle.
The rest of the BBC report treats regular features like mechanical cut filters, WDR, and HD resolution as if they were science fiction come to life. Unfortunately and ironically, what the BBC lacks is any credible assertion about succcessful facial surveillance vendors, which is not surprisingly considering there are essentially none of them.
Update: BSIA's Bad Response
As if things could get worse, the British Security Industry Association's technical director essentially agreed that the science fiction claims made were actually correctly. Their main defense - that only criminals should really care:
“Such accuracy in detecting and identifying known criminals should only be of concern to those who have committed a crime, and should reduce the risk of innocent people being wrongly accused or convicted of a crime.”
This is an incredibly dangerous debating tactic, especially when opponents may see this as calling them criminals and where the truth is that this cannot be done accurately and effectively anyway.