Face recognition systems have been a hot topic in computer vision research. Three-dimensional face recognition, as a result of the evolution of 3D imaging systems, has the ability to attain a high recognition rate when compared to two-dimensional face recognition. However, conventional 3D face recognition method has some shortcomings like low recognition rate owing to the pose and illumination changes and high computational overhead. This paper built a 3D face recognition system combining multi resolution analysis and a machine learning model to overcome such restrictions. For preprocessing 3D face image, a new filtering technique is proposed. Preprocessed images are decomposed using wavelet transform and then facial features are selected using modified sine cosine algorithm. A machine learning model named Support Vector Machine (SVM) is used to complete the recognition task. The developed system's efficacy is tested using FRGCV2.0 and the Texas 3D database. According to numerical statistics, the developed system achieves a recognition rate of 98.7% and 99.8%, respectively, for neural vs all investigations upon that FRGC V2.0 and Texas 3D datasets, with a 0.1 percent incorrect admittance rate.
Volume 12 | Issue 2
Pages: 3534-3544