MODEL-BASED INTERPRETATION OF COMPLEX AND VARIABLE IMAGES

The ultimate goal of machine vision is image understanding-the ability not only to recover image structure but also to know what it represen ts. By definition, this involves the use of models which describe and label the expected structure of the world. Over the past decade, model -based vision has been applied successfully to images of man-made obje cts. It has proved much more difficult to develop model-based approach es to the interpretation of images of complex and variable structures such as faces or the internal organs of the human body (as visualized in medical images). In such cases it has been problematic even to reco ver image structure reliably without a model to organize the often noi sy and incomplete image evidence. The key problem is that of variabili ty. To be useful, a model needs to be specific-that is, to be capable of representing only 'legal' examples of the modelled object(s). It ha s proved difficult to achieve this whilst allowing for natural variabi lity. Recent developments have overcome this problem; it has been show n that specific patterns of variability in shape and grey-level appear ance can be captured by statistical models that can be used directly i n image interpretation. The details of the approach are outlined and p ractical examples from medical image interpretation and face recogniti on are used to illustrate how previously intractable problems can now be tackled successfully. It is also interesting to ask whether these r esults provide any possible insights into natural vision; for example, we show that the apparent changes in shape which result from viewing three-dimensional objects from different viewpoints can be modelled qu ite well in two dimensions; this may lend some support to the 'charact eristic views' model of natural vision.