Enhanced facial expression using oxygenation absorption of facial skin

Abstract

Facial skin appearance is affected by physical and physiological state of the skin. The facial expression especially the skin appearances are in constant mutability and dynamically changed as human behave, talk and stress. The color of skin is considered to be one of the key indicators for these symptoms. The skin color resolution is highly determined by the scattering and absorption of light within the skin layers. The concentration of chromophores in melanin and hemoglobin oxygenation in the blood plays a pivotal role. An improvement work on prior model to create a realistic textured three-dimensional (3D) facial model for animation is proposed. This thesis considers both surface and subsurface scattering capable of simulating the interaction of light with the human skin. Furthermore, six parameters are used in this research which are the amount of oxygenation, de-oxygenation, hemoglobin, melanin, oil and blend factor for different types of melanin in the skin to generate a perfect match to specific skin types. The proposed model is associated with Blend Shape Interpolation and Facial Action Coding System to create five basic facial emotional expressions namely anger, happy, neutral, sad and fear. Meanwhile, the correlation between blood oxygenation in changing facial skin color for basic natural emotional expressions are measured using the Pulse Oximetry and 3D skin analyzer. The data from different subjects with male and female under different number of partially extreme facial expressions are fed in the model for simulation. The multi-pole method for layered materials is used to calculate the spectral diffusion profiles of two-layered skin which are further utilized to simulate the subsurface scattering of light within the skin. While the subsurface scattering is further combined with the Torrance-Sparrow Bidirectional Reflectance Distribution Function (BRDF) model to simulate the interaction of light with an oily layer at the skin surface. The result is validated by an evaluation procedure for measuring the accountability of a facial model via expressions and skin color of proposed model to the real human. The facial expressions evaluation is verified by calculating Euclidean distance between the facial markers of the real human and the avatar. The second assessment validates the skin color of facial expressions for the proposed avatar via the extraction of Histogram Color Features and Color Coherence Vector of each image with the real human and the previous work. The experimental result shows around 5.12 percent improvement compared to previous work. In achieving the realistic facial expression for virtual human based on facial skin color, texture and oxygenation of hemoglobin, the result demonstrates that the proposed model is beneficial to the development of virtual reality and game environment of computer aided graphics animation systems.