Abstract
Skin quality perception is a discrimination process that determines many of our attitudes towards a person. However, the perception mechanisms involved in this process are not fully understood. Vision-based approaches can contribute to understand the role of human vision perception when confronted to visual quality estimation of complex materials such as the human skin. Here, we analysed how the colorimetric changes affected the visual perceived quality of human skin when modified by different makeup styles. A makeup artist was given the task to created 4 different makeup styles based on four given keywords: flat, fluffy, wet and sharp. The makeup was applied to the same woman on the same day. First, we took photographs of each makeup style in a controlled environment. Then, we randomly presented these images to subjects that judged the visual quality of the skin. The experimental results revealed that the applied makeup styles had different degrees of quality perceptions enabling their use in quality discrimination experiments. Moreover, the results were correlated with luminance statistics for the visibility of pores and glossiness perception. However, the visibility of makeup was uncorrelated with luminance statistics. Furthermore, we created artificial stimuli with only variations in their luminance distribution and without any change in the chromatic information. When presenting the artificial stimuli, the experimental results showed that the perceived quality of the skin changed, supporting the hypothesis that luminance distribution plays an important role as a visual cue for the determination of visual quality of human skin texture. Our results complement previous research findings, and open the possibility for the creation of new automated cosmetic efficacy estimators. Moreover, the development of cosmetics focusing on the luminance distribution of the skin can represent new alternatives for foundations that keep the original color of the skin and therefore give a more natural impression.
Meeting abstract presented at VSS 2014