Nkengne A, Bertin C, Stamatas G N, Giron A, Rossi A, Issachar N, Fertil B
INSERM, U678, Paris, France.
J Eur Acad Dermatol Venereol. 2008 Aug;22(8):982-91. doi: 10.1111/j.1468-3083.2008.02698.x. Epub 2008 Jun 6.
The facial appearance of a person does not always reflect the chronological age; some people look younger or older than they really are. Many studies have described the changes in skin properties (colour, wrinkles, sagging, micro relief, etc.) with age, but few of them have analysed their influence on the perceived age. The primary objective of this study was to assess the contribution of individual skin attributes of the face on the perceived age of Caucasian women. Secondary objectives were to assess the influence of age and gender of graders with regard to the age perception.
A random sample of 173 subjects of 20 to 74 years of age was taken from a database of more than 5000 healthy Caucasian women. A trained grader performed visual assessment of facial skin attributes (using a visual analogue scale), and a front face photograph was taken from each subject. Photographs were shown to 48 graders (20 men and 28 women, aged 22-64 years) who were asked to estimate the age of the subjects. Graders were classified as young (less than 35 years), middle age (35-50 years) and seniors (older than 50 years). Partial Least Square regression models were built to predict the chronological and the perceived age from the measured facial individual attributes. The contribution of each attribute within the regression model enabled to measure the relevance of this attribute with regards to age prediction.
The eye area and the skin colour uniformity were the main attributes related to perceived age. For age prediction, older graders' estimations were more driven by lips border definition shape and eyes opening, whereas younger graders' (older than 50 years) estimations were more driven by dark circles, nasolabial fold and brown spots. There were statistically significant differences in graders' age perception between gender and among age ranges. Our findings suggest that female graders are more accurate than male, and younger graders (under 35 years) are more accurate than older (over 50 years) to predict Caucasian women age from facial photographs.
Different skin attributes influence the estimation of age. These attributes have a different weight in the evaluation of the perceived age, depending on the age and of the observer. The most important attributes to estimate age are eyes, lips and skin colour uniformity.
人的面部外观并不总是反映实际年龄；有些人看起来比实际年龄年轻或年老。许多研究描述了皮肤特性（颜色、皱纹、松弛、微观纹理等）随年龄的变化，但很少有研究分析它们对感知年龄的影响。本研究的主要目的是评估面部个体皮肤属性对白人女性感知年龄的贡献。次要目的是评估评分者的年龄和性别对年龄感知的影响。
从一个包含5000多名健康白人女性的数据库中随机抽取173名年龄在20至74岁之间的受试者。一名经过培训的评分者对面部皮肤属性进行视觉评估（使用视觉模拟量表），并为每个受试者拍摄正面照片。将照片展示给48名评分者（20名男性和28名女性，年龄在22 - 64岁之间），要求他们估计受试者的年龄。评分者被分为年轻组（小于35岁）、中年组（35 - 50岁）和老年组（大于50岁）。建立偏最小二乘回归模型，根据测量的面部个体属性预测实际年龄和感知年龄。回归模型中每个属性的贡献能够衡量该属性在年龄预测方面的相关性。
眼周区域和肤色均匀度是与感知年龄相关的主要属性。对于年龄预测，老年评分者的估计更多地受唇缘定义形状和眼睛睁开程度的驱动，而年轻评分者（大于50岁）的估计更多地受黑眼圈、鼻唇沟和褐斑的驱动。评分者的年龄感知在性别和年龄范围之间存在统计学上的显著差异。我们的研究结果表明，从面部照片预测白人女性年龄时，女性评分者比男性更准确，年轻评分者（35岁以下）比年长评分者（50岁以上）更准确。
不同的皮肤属性会影响年龄估计。这些属性在感知年龄评估中的权重不同，这取决于观察者的年龄。估计年龄最重要的属性是眼睛、嘴唇和肤色均匀度。
