Eyes versus Eyebrows: A Comprehensive Evaluation Using the Multiscale Analysis and Curvature-Based Combination Methods in Partial Face Recognition

Abstract

This work aimed to find the most discriminative facial regions between the eyes and eyebrows for periocular biometric features in a partial face recognition system. We propose multiscale analysis methods combined with curvature-based methods. The goal of this combination was to capture the details of these features at finer scales and offer them in-depth characteristics using curvature. The eye and eyebrow images cropped from four face 2D image datasets were evaluated. The recognition performance was calculated using the nearest neighbor and support vector machine classifiers. Our proposed method successfully produced richer details in finer scales, yielding high recognition performance. The highest accuracy results were 76.04% and 98.61% for the limited dataset and 96.88% and 93.22% for the larger dataset for the eye and eyebrow images, respectively. Moreover, we compared the results between our proposed methods and other works, and we achieved similar high accuracy results using only eye and eyebrow images.

Curvature Best Basis: A Novel Criterion to Dynamically Select a Single Best Basis as the Extracted Feature for Periocular Recognition

“…Then the images are filtered with a Gaussian filter G σ (mn) (38). The approach produces a smoother version of the original image Y (m,n) (39), and this smoothing filter has improved the performance results [16].…”

“…4 white dashed-box). To go further to more levels may produce lower performance results [16], [31], [32]. Using filters from (3)-( 22), we construct 21 WPT coefficients (nodes) with J , J-1, and J-2 and respected τ index as shown in Fig.…”

“…The Gaussian curvature is an intrinsic measure, and the mean curvature is an extrinsic measure [51]. In [16], the extracted curvature that yielded the best performance was derived from the mean curvature. Later in our experiments, we also show that the extracted mean curvature creates more data fluctuations than the Gaussian curvature.…”

“…Our recent work [16] employed multi-resolution analysis methods (discrete wavelet transform and scale-space) and combined them with surface curvature. Although the work in [16] produced satisfying results, the information on multiresolution analysis has been chosen as a static option, and the performance in each multi-resolution point of view was calculated one by one in the recognition system. There has been a gap in the dynamic selection of only the best information from the multi-resolution analysis.…”

“…The dimensional reduction-based such as principal component analysis (PCA) [12] and linear discriminant analysis (LDA) [13] have been evaluated. Multi-resolution analysis methods, such as wavelet-based [14], [15], and scale-space [16], have been employed to evaluate periocular regions. The periocular recognition system has also been tested using feature descriptor methods such as scale-invariant feature transform (SIFT), speeded up robust features (SURF), binary robust invariant scalable keypoints (BRISK), oriented FAST, and rotated BRIEF (ORB) [17], [18], deep learning and neural network-based [19], [20].…”