| Date | May 2018 | Marks available | 4 | Reference code | 18M.2.HL.TZ0.9 |
| Level | HL | Paper | 2 | Time zone | no time zone |
| Command term | Explain | Question number | 9 | Adapted from | N/A |
Question
An image processing system analyses colour images of architectural landscapes. The system compares any new images with existing ones stored in a knowledge base. The purpose of the system is to extend the knowledge base with images of the same scenes taken either from different perspectives or at different times of the year or with different weather conditions.
Explain how cluster analysis can be used to achieve the aims of the system that is described above.
Discuss the appropriateness of using genetic algorithms to compare the processed images with those stored in the knowledge base.
Markscheme
Award [4 max].
Cluster analysis will attempt to group together images;
Into groups that have certain similarities;
This will be achieved by looking at characteristics;
Such as the presence of horizontal/vertical lines/windows/spires etc.
Award [6 max]:
Award [2] + [2] + [2].
Award [2] for a description of GA, such as:
GA uses a fitness function (based on stored data);
To validate/discard the “adequacy” of some detected features;
Award [2] for one reason for appropriateness with amplification.
The features in images have been extracted in previous processing;
And can be represented in a suitable way to be used with genetic algorithms;
Award [2] for one reason against appropriateness with amplification.
One has to decide which features shall drive the recognition process;
Choosing the wrong feature may lead to incorrect results;
If the GA merges similar features to build broad classes;
Then the recognition may not be precise enough;
If the GA focuses on several small details taken in isolation;
A limited number of features may be recognised;
