ABSTRACT
The most significant contribution from this paper is that we propose a new 2DMETS-based SIFT algorithm for the image target matching. By conducting the 2DMETS on the raw color images, the SIFT can more effectively and reliably identify the image targets. Moreover, in different conditions of the signal-to-noise ratio and image size, the 2DMETS-based SIFT can not only save a large amount of calculation cost, but also guarantee the stable performance on the image segmentation without any requirement of the targets’ prior feature information. Based on the 2DMETS processing, both the pixel gray and neighborhood spatial information will be used to construct a 2-D image gray histogram for the image threshold segmentation. Furthermore, the performance of the feature point searching and matching can be guaranteed by 2DMETSbased SIFT algorithm because the background noise and interference of edge pixel points are suppressed.
