Fractional speeded up robust features detector with the Caputo-Fabrizio derivative
- PDF / 3,205,014 Bytes
- 16 Pages / 439.642 x 666.49 pts Page_size
- 54 Downloads / 195 Views
Fractional speeded up robust features detector with the Caputo-Fabrizio derivative 2 ´ 1 · J. F. Gomez-Aguilar ´ J. E. Lav´ın-Delgado1 · J. E. Sol´ıs-Perez 1 ´ R. F. Escobar-Jimenez
·
Received: 25 October 2019 / Revised: 8 July 2020 / Accepted: 5 August 2020 / © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract In this research, a fractional-order method for distinctive keypoints detection and to the image matching based on the Caputo-Fabrizio derivative and in the Speeded-Up Robust Feature (SURF) algorithm is presented and experimentally tested. The main advantage of introducing the fractional-order derivative is the improvement of the texture details detection, by combining this derivative with the SURF algorithm, the images feature extraction is improved to reach accurate images matching. The proposed method is compared experimentally with conventional SURF and SIFT algorithms. The experimental results showed that the proposed method has a high capacity for detecting points of interest in a region of the image with low contrast and weak texture. Keywords Fractional calculus · Speeded-up robust feature · Caputo-Fabrizio derivative · Interest point detection · Image matching
1 Introduction The task to search for corresponding points between two images from the same scene, or in objects taken at different times is one of the problems in the computer vision area to solve, because it has many applications such as object recognition and tracking, robot navigation, 3D reconstruction, panoramic image generation, to mention a few. For this reason, in the last decades, several investigations have been done on this issue. In [12] was presented a Scale Invariant Feature Transform (SIFT) method for extracting interest points from images that are useful for stereoscopic image matching. The detected keypoints were invariant to the scale and rotation of the image, and partially invariant to changes in 3D viewpoint and changes in illumination. Also, highly distinctive, this property is used to perform matching J. F. G´omez-Aguilar
[email protected] 1
Tecnol´ogico Nacional de M´exico/CENIDET, Interior Internado Palmira S/N, Col. Palmira, C.P. 62490, Cuernavaca, Morelos, M´exico
2
CONACyT-Tecnol´ogico Nacional de M´exico/CENIDET, Interior Internado Palmira S/N, Col. Palmira, C.P. 62490, Cuernavaca, Morelos, M´exico
Multimedia Tools and Applications
image based on their descriptors and the nearest-neighbor algorithm. Experimental results demonstrated that the method has a perform reliably for object recognition tasks. An interest point detector and descriptor called Speeded-Up Robust Feature (SURF) was presented in [5]. It outperforms the SIFT method in terms of repeatability, distinctiveness and computation time. This approach considers the use of integral images for image convolutions, as well as the Hessian matrix approximation based on box filters. Experimental results on both an image set and an object recognition application showed a desirable performance. A faster approximation of t
Data Loading...
