A Comparative Study of Matching Algorithms for Natural Markers

Natural markers are increasingly being adopted by the Augmented Reality (AR) community. These systems require several steps of data processing to be carried out with the aim of allowing a less parameterized image be recognized by the computer system. Among the possibilities for this task there are the proposals that employ interest points (or feature points). The steps needed to process the natural markers using interest points include the matching (usually between one image representation and the video stream). Even having these options well known in the community, try to minimize the processing time or improve the results in terms of accuracy of matching is especially important to the development of AR applications. In this paper, we discuss the results of three comparative studies that demonstrate the advantages and disadvantages of the algorithms FLANN and Brute Force when employed in the matching step of natural markers using interest points. As one of the results, we show that in some cases, the algorithm with most computational complexity can be the better choice.