It depends what is the goal of your analysis and the type of data that you are working with. For example, if you are working in machine learning (specifically support vector machine classification) then the RBF kernel function would be a good/standard approach for detecting similarity between two features.
Ritam Guha : Cross-correlation (using pearson's /spearman's cofficient) and auto-correlations are simple approaches for assessing "similarity" between feature sets.