Newton’s Law of Gravitational Force (NLGF) based Machine Learning Technique for Uneven Illuminated Face Detection

Aug 3, 2026

Face detection and recognition enable efficient photo gallery management and reliable memory search, yet performance degrades under uneven illumination. This research introduces a novel machine learning approach grounded in Newton’s third law of gravitational force to model pixel relationships for feature extraction. The method characterizes discriminating local traits to maximize dissimilarity between different identities while minimizing intra-person feature differences. It targets noisy, unevenly illuminated, and rotationally invariant face images with accurate and efficient processing.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages18

Reading Time45 min

AccessFree to download

Introduction

Illumination-invariant face recognition challenge

Existing illumination handling methods

Image descriptors and feature extraction

Local vs global descriptors

Hand-crafted and deep descriptors

Feature detection criteria and corner detection

Newton’s Law of Gravitational Force (NLGF) based Machine Learning Technique for Uneven Illuminated Face Detection M. Shalima Sulthana1, C. Naga Raju2 1Research Scholar [m.s.sulthana2012@gmail.com](m.s.sulthana2012@gmail.com) Department of Computer Science and Engineering YSR Engineering College of YVU Yogivemana University-Kadapa 2Professor Department of Computer Science and Engineering YSR Engineering College of YVU Yogivemana University-Kadapa Abstract—A photo gallery is crucial for organizing your photos, presenting them in beautiful categories, and doing sophisticated memory searches. The photo gallery is portrayed in a vocabulary of nonlinear similarities to the prototype face image collection. One of the difficult research ideas for machine learning technologies is the maintenance of a photo gallery using facial recognition. Based on changes in the faces'appearance, faces are identified. This research proposes novel machine learning algorithms to recognize faces by characterizing the majority of discriminating local characteristics, which maximizes the dissimilarity between face photos of different persons and reduces the dissimilarity between features between face images of the same person. This method relies on Newton's third law of gravitational force to determine the relationship between pixels to extract the features of noisy accurately and efficiently, unevenly illuminated, and rotationally invariant face images. Keywords-NLGF, ENLGF, Gravitational Force, Face recognition, Gamma-correction. I. INTRODUCTION Face recognition is a popular research area, On account of its extraordinary success and vast social applications. Every human face in this world has the same facial features like eyes, ears, lips and nose, etc. however, every face shows different structures. Several facial recognition algorithms have been proposed during the last few decades, such as Bayesian facial recognition [1], facial recognition with support vector machines [2], and Eigenfaces [3], However, the images that are obtained in good environments were used to test these algorithms in the past, but failed to show the best results on illumination invariant images. To solve this illumination invariant problem, several existing methods have been applied, including method1: in this method face images are intended to decrease the intensity first, the fundamental techniques come under this method includes gamma-correction [4], logarithmic modification [5], and further advanced methods are proposed to measure this intensity i.e., histogram equalization [6], etc. Method2: modeling strategies are used to detect facial images in various illumination situations i.e., 3Dimentional texture of a human face, in [7] explains the 3Dimentional structure of human faces with various levels of illumination conditions. Method3: invariant feature method, this method is used to detect static features such as edge mapping, Gabor filter and image stabilization of illuminated images. All these approaches reflect the common assumption that wavelengths they produce from all radiational sources are the same. Every image descriptor provides an explicit description of features such as shape, color, texture, or movement that is visible in the images or algorithms or videos. These image descriptors are divided into global descriptors and local descriptors. Global-descriptors are used to describe the texture of an image and color of an image, these global descriptors are outmoded due to their inconsistent performance in the form of illumination and occlusion images [8] . Local descriptors are traditionally categorized as data-driven descriptors, used to enhance the local performance of an image and handcrafted descriptors used to characterize an image localpatches [9] i.e., the key points in the image. Data driven descriptors have been designed and structured in accordance with training data. To identify the optimal configuration in the descriptor pa

FAQs

Why does uneven illumination reduce face recognition accuracy?

Many prior algorithms are tested under good imaging conditions and assume consistent radiational wavelength behavior. They fail to achieve best results when illumination varies across the image.

What is the core idea of the proposed NLGF-based technique?

The approach uses Newton’s third law of gravitational force to determine relationships between pixels, enabling extraction of discriminative local features from noisy, unevenly illuminated, and rotation-invariant face images.

How does the method improve discrimination between different people?

It characterizes local traits to maximize dissimilarity between face photos of different persons while reducing dissimilarity between feature representations of the same person.

Aug 3, 2026

Upload by Lucas Martin