面孔相似性判断中各部件的权重分配

采用多级估量法，通过分析不同隶属度的部件样例组合成的面孔样例对原型的隶属度，考察耳朵、脸型、眼睛和嘴巴等部件在面孔相似性判断中的作用。结果发现，面孔相似性判断中，人们对不同面孔部件赋予的权重并不相同，对脸型和嘴巴赋予的权重较大，而对耳朵和眼睛赋予的权重较小。表明，面孔各部件在面孔相似性判断中的作用不同，相对于耳朵和眼睛，脸型和嘴巴的作用更大；面孔内部特征和外部特征的划分方法并不一定适合于面孔相似性判断。

The Weight Distribution of the Face Part in the Face Similarity Judgment

The judgment of face similarity is generally considered to be one kind of fuzzy judgment. In contrast with the traditional evaluation method, the multistage evaluation method was more practical, which was frequently adopted to measure the similarity judgment. The present study used such method of multistage evaluation method, to investigate the weight conferred on the different parts of the human face. First of all, we constructed two faces as two experimental prototypes (which we called prototype 1 and prototype 2). Each face was comprised of six parts: nose, eyebrows, ears, face outline, eyes, and mouth. In terms of no interest of the first two face parts as mentioned above (nose and eyebrows) in the present study, the other four parts (including eyes, ears, mouth and the face outline) were considered as four parts prototype on each face, which were extremely different in two experimental prototypes. We took each part of two prototypes as the source face and the target face respectively. By using the Magic Morph (V1.95) image processing software, we deformed eight intermediate faces from the source part to the target part. For example, the ear of prototype 1 was as the source part, the ear of prototype 2 was as the target part. By deforming, 8 consecutive pictures between prototype 1 and 2 were construted. In total, there were 10 exemplars for each part of each prototype. Participants were asked to make a judgment of the membership of each exemplar to the corresponding part prototype. Then, as unit of each participant, three exemplars of each part were selected: the first exemplar was with high membership with prototype 1 but low membership with prototype 2; the second exemplar was with middle membership of both two prototypes; and the third exemplar was with low membership of prototype 1 but high membership with prototype 2. On the basis of this, twelve exemplars occurred for each participant, which consisted of 81 faces being used as the formal experimental stimulus. Participants were instructed to evaluate the similarity between each face and each prototype with multistage evaluation judgment. The membership was calculated as the dependent variable in the present study. The results showed that the weight of four different face parts varied to some extent: the highest was the face outline with .550, which was significant high than the other parts, then the mouth was in the middle (.250), and ears and eyes were at the lowest (.100). These results indicated that, different parts of face played different roles on the face similarity judgment. Compared with ears and eyes, the face outline and mouth were with more importance. Moreover, we argued that the division of the internal feature and external feature might not be suitable on the face similarity judgment.