Skip to Main Content
Local descriptors are a relatively new type of image registration methods and there exist numerous studies in the area over the past few years. The methods can be divided into three stages: region detection, descriptor formation and descriptor matching. This paper looked at the performance of current state of the art local descriptor methods for registering images of Maori artefacts under various image transformations such as rotation, scale, tilt and viewpoint changes. In particular, the focus was on the second stage which is the formation of local descriptors. The problems encountered in a previous experimental work conducted were discussed and based on the results, two new local descriptor methods were proposed which aimed to take advantage of the additional information provided by using colour images instead of greyscale images. Experimental work conducted indicate that the proposed methods out-perform existing methods when registering images of Maori artefacts with gains in matching accuracy of up to 10% depending on the image transformation involved.