Posted: January 4th, 2016

Application of Artificial Intelligence

An automatic car number-plate recognition (ANPR) system is to be used to identify vehicles entering and leaving a proposed congestion charge area in Southampton. This is to be achieved using a camera taking images of the vehicle as it passes into the zone and using software to identify the number.

The customer has asked that you provide a reasoned argument that describes, compares and contrasts two different AI approaches to the problem of identifying the characters (letters and numbers) on the number plate i.e. a symbolic solution and a connectionist solution together with any underlying mechanisms that are needed.

In doing this, you need to:

a) Identify a number of features of the characters on UK number plates that might be useful in this kind of recognition system. (You do not need either a legal description or the history of number plates, nor do you need to describe features that are not relevant to the identification of letters and numbers.)

b) Describe sufficient low level vision processing mechanisms that would need to be carried out in order to identify the features described in a).

c) The system will be required to take features that have been recognized by pre-processing and output the vehicle’s identity. Describe two different AI mechanisms that may be used. Compare and contrast the two mechanisms. For each of these systems you should also include reasoning for whether either fuzzy logic or statistical methods should be incorporated.

d) Reflect on whether or not an AI solution is significantly easier or better than a solution developed using more traditional software development technologies.