PDF version of this document

Experiments

AART was used to carry out all our experiments in a stable environment which is straightforward to handle. Outputs varied from data files to graphs, images and videos of several types. Long-running simulations provided the required comparative results which revealed the strengths of the algorithms proposed in this paper.

Fig. 1. AART investigation of the dynamic registration behaviour by displaying video sequences

The application outputs, in particular the many types of graphs and videos have been thoroughly analysed by a panel to derive conclusions regarding the behaviour of the objective functions and perform subsequent experiments accordingly. The exact objective functions and optimisation regimes for all methods are beyond the scope of this paper.

The following concepts are worthy of further elaboration. Some of these have never been used before so their potential can be truly comprehended for the first time.

Probability Density Function (PDF): Such functions describe the volume of data distributions. More uniform data, as we aspire to achieve across all images during registration, will result in lower such values. An exponential PDF was used in the experiments by default although over a dozen others are available in AART, including a Gaussian one.

Wavelets: As compression [19] is closely related to MDL, these can provide an accurate estimate of the complexity of data and abundance of patterns within that data. An extensive group of different wavelets are offered by the application and, by default, Daubechy was used in the experiments. Computationally cheaper alternatives to the wavelets are Fourier and Hough transforms, but these have not yet been incorporated into AART.

Mutual Information: This strand of methods [18,12] will analyse the peaks of image histograms. Normalised MI is currently one of the most robust and widely-used methods for 2-D data.