مقاله انگلیسی رایگان در مورد الگوریتم استخراج سریع نقطه کلیدی شبکیه چشم – IEEE 2019

The feature extraction for multispectral images plays an important role in many computer vision applications. Recently, geometric algebra (GA) based scale invariant feature transform algorithm (GA-SIFT) and GA based speeded up robust Features algorithm (GA-SURF), have been proposed to extract feature of multispectral image in GA space. However those methods are difficult to be implemented in realtime applications. Now, the challenge is to design a new algorithm to extract the features of multispectral image more efficiently and quickly, so that it can be used in real-time applications. Although the proposed fast retina keypoint (FREAK) algorithm is faster to compute and more robust than SIFT and SURF, it can not be utilized to extract features directly for multispectral images. In this paper, we propose a novel fast retina keypoint extraction algorithm based on GA, named as GA-FREAK, for multispectral images. Firstly, the multispectral images are represented as multivectors in GA space, then the interest points are detected with the procedure of FREAK in GA space. Finally, our experiments demonstrate that the GA-FREAK is faster and more robust than some previous algorithms in multispectral images. It is expected that the proposed GA-FREAK will be a competitive alternative in real-time applications of multispectral images.

Introduction

Recently, multispectral images have been wildly used in many fields, such as computer vision, biology, remote sensing, astronomy, medicine, and digital photography [1]–[7]. Multispectral images consist of various image data captured at specific wavelength ranges across the electromagnetic spectrum and have more than three bands to contain more information than grayscale images. Feature extraction analysis is becoming more and more popular in multispectral image applications [8], [9], and it is applied to many aspects, such as object recognition, image stitching, and pattern recognition [10], [11]. More and more research works focus on the design of effective feature extraction algorithms in embedded devices with restricted memory and computation. The scale invariant feature transform (SIFT) proposed by Lowe is one of the most popular feature extraction algorithm [12]. Many SIFT based feature descriptors for color images are proposed, such as RGB-SIFT [13], HSV-SIFT [13], PCA-SIFT [14], and so on. However, SIFT based image feature extraction methods suffer from large computational burden. The speeded up robust feature (SURF) [15] proposed by Bay is a computationally-efficient replacement to SIFT, which outperforms SIFT in robustness and speed [16]. Since then, many algorithms based on SURF extensions are proposed, such as SURF-DAISY [17], Gauge-SURF [18], and SSURF [19]. Baig et al. [20] proposed a novel robust image representation for the content-based image retrieval (CBIR), which is based on complementary visual words intergration of SURF and cooccurrence histograms of oriented gradients (CoHOG).