Image matching based on the adaptive redundant keypoint elimination method in the SIFT algorithm
Zahra Hossein-Nejad,Hamed Agahi,Azar Mahmoodzadeh
DOI: https://doi.org/10.1007/s10044-020-00938-w
IF: 2.307
2020-11-12
Pattern Analysis and Applications
Abstract:Scale invariant feature transform (SIFT) is one of the most effective techniques in image matching applications. However, it has a main drawback: existing numerous redundant keypoints located very close to each other in the image. These redundant keypoints increase the computational complexity while they decrease the image matching performance. Redundant keypoint elimination method (RKEM)–SIFT are incorporated to eliminate these points by comparing their distances with a fixed experimental threshold value. However, this value has a great impact on the matching results. In this paper, an adaptive RKEM is presented which considers type of the images and distortion thereof, while adjusting the threshold value. Moreover, this value is found separately for the reference and sensed images. In an image, the adaptive RKEM finds the histogram of the keypoints distances, for which the number and the width of the bins are determined based on the number of keypoints and the distances distribution metrics. Then, a maximum value for searching the optimal threshold value is determined. Finally, for each integer value smaller than the mentioned maximum, a set containing distances smaller than that value is created and the one with the smallest variance is selected. The integer value corresponding to that set is chosen as the adaptive threshold for that image. This approach can improve the efficiency of the RKEM-SIFT in eliminating redundant keypoints. Simulation results validated that the proposed method outperforms the SIFT, A<sup>2</sup> SIFT and RKEM-SIFT in terms of the matching performance indices.
computer science, artificial intelligence