Contextual Saliency for Non-rigid Landmark Registration and Recognition of Structured Natural Patterns

In this paper we develop a method for injecting within-pattern information into the matching of point patterns through utilising the shape context descriptor in a novel manner. The algorithm is motivated in Visual Animal Biometrics, a developing field concerned with the non-invasive, automatic recognition of animals. Landmark distributions on animal coats are commonly used as characteristic features in the pursuit of individual identification and are often derived by imaging surface entities such as bifurcations in scales, fur colouring, or skin ridge minutiae. However, many natural distributions of landmarks are quasi-regular, a property with which state-of-the-art approaches in point-set matching struggle. The method presented here addresses the issue by guiding matching along the most distinctive points within a set based on what we term contextual saliency. Experiments on synthetic data are reported which show the contextual saliency measure to be tolerant of many point-set transformations and predictive of correct correspondence. A general point-matching algorithm is then developed which combines contextual saliency information with naturalistic structural constraints in the form of the thin-plate spline. A comparative case study shows the presented algorithm to outperform two widely used point-matching techniques on a real-world manta ray data set.