Analysing images and other sorts of unstructured data requires plenty of data preprocessing efforts. If the objective is to apply decision trees in order to classify images in a big dataset, which is the most important data pre$-$processing activity from the below list?

Dimensionality reduction

Duplicate elimination

Missing values

Outlier Detection