Использование метода главных компонент для кластеризации изображений лиц
Journal Title: Математичне моделювання - Year 2017, Vol 1, Issue 2
Abstract
USE PRINCIPAL COMPONENT ANALYSIS TO CLUSTERING IMAGES OF FACES Shumeiko A.A., Bratuta M. Abstract The range of tasks considered in image processing (Image Processing) is quite large, and among them an important role is played by problems related to the perception of human faces. This is finding on the image of a human face, identifying this person or analyzing the resulting image to assess the aggressiveness of the person, gender, age and other characteristics. One of the tasks is finding a person of the same type in the sample. What is the "similarity" of human faces. This question is not only and not so much mathematical as psychological, an example of this is Leopold Sondi's test. But, nevertheless, the task of "similarity" will always be of interest to a person, the eternal problem is who the child looks like. On the other hand, there are many years of observations that say that quite often in happy families, the spouses are similar to each other. What is primary - happy, because they are similar or similar, because they are happy? And it does not matter. What is important is that the task of likeness of individuals is meaningful and interesting in itself. Formulation of the problem. Let there be given a lot of images of human faces and an image for which one of the most similar ones should be selected and order them by rating (degree of similarity). In the paper, clustering faces images by the k-means method is suggested. The center of the cluster using the main component of the Principal Component Analysis clusters elements. As a criterion of proximity, the values of the corresponding coefficients are used. The application of the resulting clusterization to the solution of the problem of finding the persons most similar to the test image is considered. References [1] Schegoleva N.L. Application of algorithms for two-dimensional analysis of main components for image recognition problems of individuals / N.L. Schegoleva, G.A. Kukharev // Business Informatics .– 2011 .– №№4 (18) .– P.31–38. [2] Prokoshev V.G. The problem of automatic recognition of faces with one reference image / V.G. Prokoshev, M.M.Rozhkov, P.Shamin // Scientific and technical statements of SPbSPU .– 2010 .– №5 .– P.13–18. [3] Ligun A.O. Computer graphics (Image Processing and Compression Data)/ A.O. Ligun, O.O.Shumeiko .– Bila K.O., 2010.– 114 pp . [4] Shumeiko A.A. Intelligent Data Analysis (Introduction to Data Mining) / A.A. Shumeiko, S.L.Sotnik .– Dnepropetrovsk: Belaya E.A., 2012.– 212 pp.
Authors and Affiliations
А. А. Шумейко, М. Ю. Братута
Keywords
Related Articles
- EP ID EP276822
- DOI -
- Views 88
- Downloads 0
Обоснование соотношения площадей поперечного сечения грунтового теплообменника для геотермальной вентиляции
JUSTIFICATION FOR RATIO OF CROSS-SECTIONAL AREAS OF GROUND HEAT EXCHANGER FOR GEOTHERMAL VENTILATION Koviazin A. S. Abstract Effective management of the livestock industry is impossible without full feeding and create o...
Математическая модель тепло- и массопереноса в первом периоде сушки плоского слоя дисперсного материала
The analytical method for calculating the temperature field and the kinetics of evaporation of moisture during drying heat "massive" flat layer of moist particulate material.
ВИЗНАЧЕННЯ СИЛОВОГО ВПЛИВУ ЗМІННОГО СТРУМУ НА МЕТАЛ У ВАННІ УСТАНОВКИ КІВШ-ПІЧ
DETERMINATION OF THE POWER ACTION OF THE ALTERNATING CURRENT ON METAL IN THE POOL LADLE-FURNACE UNIT Krikent I.V. Abstract The work is devoted to theoretical research in the field of secondary steel processing. The expe...
РОЗРАХУНОК РОЗПОДІЛУ ГУСТИНИ МІЖАТОМНОГО ПОТЕНЦІАЛУ У СТРУКТУРНІЙ ОДИНИЦІ РЕЧОВИНИ
THE CALCULATION OF INTERATOMIC POTENTIAL DENSITY DISTRIBUTION IN THE STRUCTURAL UNITS OF MATTER Evfimko K.D., Mochalov A.A., Shapoval N.А., Koval S.S. Abstract Introduction.Theoretical study of the physical properties o...
Моделювання впливу похибок форми опорних поверхонь гідростатичного підшипника на статистичні характеристики траєкторій шпинделя
In this paper were researched laws formation of the statistical characteristics of the trajectories movement of the spindle mounted on hydrostatic bearings depending on the form errors, temperature deformations support s...