CHAPTER 4 CIVIL LIBERTY IMPLICATIONS
The Data Protection Act
4.15 A fundamental
principle behind the Data Protection Act 1984 is that material
should be collected for specified and legitimate purposes, and
used and disclosed only in ways compatible with those purposes.
Yet, as already noted, the Act only applies to CCTV material in
the relatively rare circumstance that the data were recorded and
processed with respect to an individual.
4.16 In both written
and oral evidence from the Data Protection Registrar (QQ 222-259,
p 76) we were struck by the commonality of purpose behind
her remit and the objectives which we believe should be applied
to the control of CCTV. We share the view that "it is not
so much the existence of technology that poses a threat to the
civil liberties of the individual, but the use that is made of
the technology and the setting in which surveillance takes place"
(p 79). The principles of the Act, on the relevance of information,
fair processing and disclosure, are just those which should be
applied to public space surveillance.
4.17 As we complete
our report a new Data Protection Bill has been introduced in Parliament.
We would like to see the new law encompass CCTV and the DPR
be given responsibility to issue or oversee enforceable codes
of practice for such systems. The maintenance of public support
for CCTV would be further strengthened if the DPR were to be given
powers to check, that is to audit, the operation of these systems,
and not to have to wait for a complaint to be lodged before taking
|Box 6: Image recognition
|Most cameras are passive, unintelligent devices which take a picture and leave the processing and interpretation to human operators. If a computer is linked to an image input device (either a camera or a scanner) it is possible to automate some of the interpretation, to speed up the interpretation process dramatically and to cross-match the image to other databases. There are numerous applications of this technology but three have, perhaps, the widest implications for images used as evidence: character recognition from scanned documents, number plate recognition (from images taken by surveillance or enforcement cameras), and face recognition.
|Character recognition software is often used in conjunction with document scanners to convert the image into a word processing file. The document is scanned to produce a digital image and then the recognition software interprets areas of text within the image, translating data patterns into appropriate text files. There are a number of advantages over the original scanned image: the new file uses considerably less memory; the document can be treated as any other word processing file, rather than as an image; the quality of the document as displayed on screen is often greatly improved; and the document becomes searchable using standard techniques.
|Number plate recognition
|This was demonstrated to the Committee by the City of London Police who use a system of cameras to monitor vehicular access to and from a controlled zone within the City. The principle is similar to character recognition discussed above. A camera films the vehicles, software interprets which section of the image is the number plate, and then character recognition software interprets the picture elements to make a 'best guess' of the numbers and letters in the number plate. In the City of London operation the system processes around 114,000 vehicles a day. It takes about four seconds to identify not only the number plate, but also to cross-check it with local Police databases and the Police National Computer and then alert the control centre operators if there is a match (eg if the vehicle has been registered as stolen). The recognition software does make errors and so the video still image of any number plate which is difficult to identify (eg obscured by mud), and any number plate which is flagged for further attention because it cross-matches with a database entry, is displayed in the control room alongside the interpretation for human verification.
|Face recognition is still in the development stages because of the scale of complexity involved. As well as needing to define what makes faces different enough for a computer to distinguish individuals, a reliable system would have to overcome differences in facial expression, head angle, lighting, movement, and ephemeral features such as sun tans, hair style, facial hair, make-up, hats and sunglasses too! A comparative database of faces identified with individuals is a prerequisite, and such a database may need to contain multiple images of each person: a database of this kind does not exist, and there might be intense opposition if it were attempted to create one.
4.18 The Government
is looking at the provision of electronic services and these are
measures we firmly support. Data matching is a particular concern
for this Enquiry where image recognition techniques are used (see
Box 6). In the City of London we saw a number plate matching system
in operation and we were impressed by the speed with which it
could process vehicle information. The Symonds Group held the
view that: "if cameras are used in conjunction with image
tracking software then the effectiveness of surveillance is vastly
improved and thus the implications of civil liberties are increased"
(p 151). If the technology continues to develop as expected
it may in the future be possible to match data from the digitised
passport photographs that are coming into use (p 127) with
other databases, enabling the rapid identification of individuals
who were, for example, attending a football match. This might
cause justifiable unease. It is clear that image recognition technologies
have an important role to play in assuring security, but controls
on their use may be needed to strike a fair balance with civil
4.19 Although we did
not take evidence and did not consider all of the issues of image
recognition and data matching in detail, it is an area where the
technology is pressing ahead and the public debate lagging far
behind. It is possible that there will be disquiet over the implications
of data-matching in general and image recognition in particular.
We were therefore surprised that there was no published guidance
for Government Departments on data-matching and related technologies
4.20 We recommend
the Government produces guidance for both the public and private
sectors on the use of data matching, and in particular the linking
of surveillance systems with other databases.
4.21 Individuals may
not be aware of the existence of information on them derived by
reference to two or more unrelated databases. They will therefore
not recognise if there is a need to approach the DPR who can act
only if a complaint has been made. We recommend that the DPR
be given powers to audit the operation of datamatching systems.
4.22 Any form of surveillance
is also a form of intrusion. The question to be considered is
whether such intrusion can be justified (p 151). In this
chapter we sought ways to harmonise the interests involved. An
individual's right to privacy and the interest of both the individual
and society in the prevention of crime and disorder need to be
fairly balanced. Essential and unavoidable compromises must be
widely appreciated. Our recommendations for a degree of greater
control over the installation, use and release of images from
public space surveillance systems seek to chart a way to achieve
that balance. We want to see public acceptance of surveillance,
introduced to deter and assist in the capture and prosecution
of criminals, blossom and flourish. This is more likely to be
the case if there is wide public debate of the issues involved,
and we recommend that the Government should promote such debate.
28 The Data Protection Bill, HL Bill 61, 14 January