This paper presents a modified kernel-based Active Shape Model for neutralizing and synthesizing facial expressions. In recent decades, facial identity and emotional studies have gained interest from researchers, especially in the works of integrating human emotions and machine learning to improve the current lifestyle. It is known that facial expressions are often associated with face recognition systems with poor recognition rate. In this research, a method of a modified kernel-based active shape model based on statistical-based approach is introduced to synthesize neutral (neutralize) expressions from expressional faces, with the aim to improve the face recognition rate. An experimental study was conducted using 3D geometric facial datasets to evaluate the proposed modified method. The experimental results have shown a significant improvement on the recognition rates.
Feature extraction for face recognition via Active Shape Model (ASM) and Active Appearance Model (AAM)