In an image processing system, a subject object (400) is placed on a calibration object having the form of a photographic mat (34) or a three-dimensional object (2500). The calibration object has a pattern comprising features, and at least some of the features are grouped into clusters such that the features in any cluster satisfy at least two different relationships based on the position, connectedness, shape, size and/or colour of the features. Images (420-426) each showing at least part of the subject object (400) and the calibration object are recorded from different positions and orientations. The image data is processed to calculate the positions and orientations at which the images were recorded. In processing data for an image, features representing the features on the calibrationobject are distinguished from features representing features on the subject object (400) and each feature is labelled as a unique feature on the calibration object. This is done by performing processing using the clusters and the relationship between the features therein.