Moscow University
Anthropology
Bulletin
Anthropology
Bulletin
1) National Research Nuclear University “MEPhI”, Moscow; 2) Lomonosov Moscow State University, Anuchin’s Research Institute and Museum of Anthropology, Moscow
The purpose of this study is to evaluate the necessary and sufficient number of individual images to form a stable and recognizable visual image – a composite portrait of a particular sample. Data: 72 digital frontal photographs of Russian girls aged between 7-9 years. Method: software «faceONface» designed for composite photo portrait creation. Results: 10 composite portraits generated on the basis of randomized samples with different capacity (from 5 to 70 individuals); the Gonzalez’s method of «noise elimination» was applied to the data. The graph of variance in mean-square deviation as a function of the number of accumulated images, visualizes previously noticed empirical regularity of achieving a quantitative threshold in which the addition of new images does not significantly change the visual perception of the generalized portrait. A possible explanation of the phenomenon of «semantic stabilization» of an image when a certain threshold of accumulation is reached is proposed. When forming a generalized photo portrait, the individual characteristics of each person are suggested to be considered analogous to the digital «noise». When a threshold in 30-40 initial images is reached, the generalized portrait of a specific sample is freed from «personal noise» (personal morphological features of the face) and approaches a typologically uniform invariant image.
physical anthropology, composite photoportrait, individual features, elimination of digital noise, pixel, mean square deviation
Цит.: Savinetskiy A.B., Maurer A.M., Syroezhkin G.V. On the question of optimal number of initial images for obtaining an invariant composite photo portrait // Moscow University Anthropology Bulletin (Vestnik Moskovskogo Universiteta. Seria XXIII. Antropologia), 2017; 3/2017; с. 36-43
Download text