A novel occluded face detection approach using Enhanced ORB and optimized GAN

IF 0.9 4区计算机科学 Q4 COMPUTER SCIENCE, SOFTWARE ENGINEERING International Journal of Wavelets Multiresolution and Information Processing Pub Date : 2023-11-04 DOI:10.1142/s0219691323500510

Abhilash Nelson, R. S. Shaji

引用次数: 0

Abstract

Objectives:This research presents an approach to detect occluded face images. In order to achieve this, the research presents a novel technique that involves feature extraction and occluded face recognition. Methods: Feature extraction is performed by the enhanced ORB algorithm, which is proposed by the modification of the Oriented Fast and Rotated Brief (ORB) algorithm, by adding a phase for contrast adjustment, together with CNN features. For occluded face recognition, a Generative Adversarial Network (GAN) optimized by the proposed SR-SSA is designed. SR-SSA is proposed by the integration of Search and Rescue Optimization (SAR) in the Sparrow Search Algorithm (SSA). Results: The experimental results demonstrate that the SR-SSA-based GAN algorithm outperforms existing methods in terms of accuracy of 0.956, FAR of 0.045 and FRR of 0.021.

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

一种基于增强ORB和优化GAN的遮挡人脸检测方法

目的:提出一种检测人脸遮挡图像的方法。为了实现这一目标，本研究提出了一种融合特征提取和遮挡人脸识别的新技术。方法:通过对ORB (Oriented Fast and rotating Brief)算法进行改进，增加一个相位进行对比度调整，并结合CNN特征，提出增强ORB算法进行特征提取。针对遮挡人脸识别问题，设计了基于SR-SSA优化的生成对抗网络(GAN)。SR-SSA是将搜救优化(SAR)与麻雀搜索算法(SSA)相结合而提出的。结果:实验结果表明，基于sr - ssa的GAN算法的准确率为0.956,FAR为0.045,FRR为0.021，优于现有方法。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文去求助

来源期刊

International Journal of Wavelets Multiresolution and Information Processing 工程技术-计算机：软件工程

CiteScore

2.60

自引率

7.10%

发文量

审稿时长

2.7 months

期刊介绍： International Journal of Wavelets, Multiresolution and Information Processing (hereafter referred to as IJWMIP) is a bi-monthly publication for theoretical and applied papers on the current state-of-the-art results of wavelet analysis, multiresolution and information processing. Papers related to the IJWMIP theme are especially solicited, including theories, methodologies, algorithms and emerging applications. Topics of interest of the IJWMIP include, but are not limited to: 1. Wavelets: Wavelets and operator theory Frame and applications Time-frequency analysis and applications Sparse representation and approximation Sampling theory and compressive sensing Wavelet based algorithms and applications 2. Multiresolution: Multiresolution analysis Multiscale approximation Multiresolution image processing and signal processing Multiresolution representations Deep learning and neural networks Machine learning theory, algorithms and applications High dimensional data analysis 3. Information Processing: Data sciences Big data and applications Information theory Information systems and technology Information security Information learning and processing Artificial intelligence and pattern recognition Image/signal processing.

期刊最新文献

Piecewise Scalable Frames in Hilbert Spaces A novel occluded face detection approach using Enhanced ORB and optimized GAN A Dilated Convolution-Based Feature Adaptation Method for Detection of High Aspect Ratio Objects in Aerial Images Fully Symmetric Frame Scaling Functions and derived Framelets Author index (Vol. 21)