Face Recognition based on Discrete Wavelet Transform and Euclidean Distance
Silvester Tena, Amin Ajaib Maggang, Imanuel A. D. Unu · 2023
A face recognition application consists of feature extraction and face identification process. The two processes have been widely used as methods for recognizing various faces. The challenge in face recognition is how a face recognition application could recognize faces with expressions, poses and facial structures that always changes by age and emotion. This research aims to compare wavelet family, Haar and Daubechies (at the maximum level of decomposition), as they are used for feature extraction on a face recognition application. Furthermore, Euclidean distance (ED) was utilized in the application as face identification method. The developed system has 240 face images in its database, collected from 40 different people with 6 different types of expression for each person. The results show that the developed application is able to produce success rate and accuracy of recognition up to 95% from 80 tested images. Haar Wavelet, at fifth decomposition level, shows the best performance as it could produce minimum feature vector while maintaining the success rate and accuracy at 91.25%.