Optimizing Fuzzy Rule Base for Illumination Compensation in Face Recognition using Genetic Algorithms
Journal Title: EMITTER International Journal of Engineering Technology - Year 2014, Vol 2, Issue 2
Abstract
Fuzzy rule optimization is a challenging step in the development of a fuzzy model. A simple two inputs fuzzy model may have thousands of combination of fuzzy rules when it deals with large number of input variations. Intuitively and trial‐error determination of fuzzy rule is very difficult. This paper addresses the problem of optimizing Fuzzy rule using Genetic Algorithm to compensate illumination effect in face recognition. Since uneven illumination contributes negative effects to the performance of face recognition, those effects must be compensated. We have developed a novel algorithm based on a reflectance model to compensate the effect of illumination for human face recognition. We build a pair of model from a single image and reason those models using Fuzzy. Fuzzy rule, then, is optimized using Genetic Algorithm. This approach spends less computation cost by still keeping a high performance. Based on the experimental result, we can show that our algorithm is feasible for recognizing desired person under variable lighting conditions with faster computation time.
Authors and Affiliations
Bima Sena Bayu Dewantara, Jun Miura
Keywords
Related Articles
- EP ID EP170976
- DOI 10.24003/emitter.v2i2.27
- Views 84
- Downloads 0
Automatic Samples Selection Using Histogram of Oriented Gradients (HOG) Feature Distance for Tsunami Victims Detection
Finding victims at a disaster site is the primary goal of Search-and-Rescue (SAR) operations. Many technologies created from research for searching disaster victims through aerial imaging. but, most of them are difficult...
Modified Particle Swarm Optimization using Nonlinear Decreased Inertia Weight
Particle Swarm Optimization (PSO) has demonstrated great performance in various optimization problems. However, PSO has weaknesses, namely premature convergence and easy to get stuck or fall into local optima for complex...
Secure Communication and Information Exchange using Authenticated Ciphertext Policy Attribute-Based Encryption in Mobile Ad-hoc Network
MANETs are considered as suitable for commercial applications such as law enforcement, conference meeting, and sharing information in a student classroom and critical services such as military operations, disaster relief...
Performance Analysis of CP-Based and CAZAC Training Sequence-Based Synchronization in OFDM System
Orthogonal Frequency Division Multiplexing (OFDM) is a popular wireless data transmission scheme. However, its synchronization is still being a major problem when it is applied in real hardware. Cyclic Prefix (CP) based...
Classification of Radical Web Content in Indonesia using Web Content Mining and k-Nearest Neighbor Algorithm
Radical content in procedural meaning is content which have provoke the violence, spread the hatred and anti nationalism. Radical definition for each country is different, especially in Indonesia. Radical content is more...