Publication: Sanar eserlerinin yapay sinir ağları ile özgünlük tespiti ve ayrıştırılması
Abstract
Son yıllarda bilgisayar teknolojileri alanındaki gelişmeler Dalgacık Dönüşümü (Wavelet), YSA olarak da adlandırılan Yapay Sinir Ağları (Artificial Neural Networks) ve DVM olarak adlandırılan Destek Vektör Makineleri (Support Vector Machine) gibi akıllı sistemlerin farklı disiplinlerde yoğun olarak kullanılmasına olanak sağlamıştır. Jeofizik, tıp ve bio-medikal gibi birçok alanda kullanılan akıllı sistemler, yüksek oranda başarılı sonuçlar vermektedirler. Bu araştırmada ise Dalgacık Dönüşümü ile YSA ve Dalgacık Dönüşümü ile DVM uygulamaları sanat eserlerinin özgünlük tespitinde ve ayrıştırılmasında kullanılmış, böylelikle resimlerin ayırt edilmesinde daha nesnel ve tatmin edici sonuçlar alınması hedeflenmiştir. Buna göre; ilk aşamada aynı dönem sanatçıların eserlerinin birbirlerinden ayırt edilmesi hedeflenmiş ve bu bağlamda Pablo Picasso ile Georges Braque’ın eserleri ayrıştırmaya veri olarak kullanılmıştır. Tezin ikinci aşamasında ise sanat tarihinden önemli bir usta sanatçının özgün sanat eserlerinin farklı ressamlar tarafından yapılmış sahtelerinden ayırt edilmesi hedeflenmiştir. Bu problemin çözümünde ise Matisse’in özgün eserleri ile farklı ressamlar tarafından yapılmış sahtelerinden yararlanılmıştır. Yapay Sinir Ağları, Destek Vektör Makineleri, Dalgacık Dönüşümü, Sanat Eseri, Sahte ve sahtecilik, Pablo Picasso, Georges Braque ve Henri Matisse.
In recent years, with the latest developments in computer technology intelligent systems which are being widely used in different fields such as geophysics, medicine and bio-medical have recently successfully high results. In this research, wavelets-ANN (Artificial Neural Networks) and wavelets-SVM (Support Vector Machine) processes have been used in the authentication and separation of masterpieces. Thus, achieving more objective and satisfactory results are targeted. In this thesis, in order to solve the problems of the authentication and separation of masterpieces, two methods have been applied. According to these two methods, firstly the separation of the masterpieces of the same period artist have been given priority and this context, the works of Pablo Picasso and Georges Braque have been used as data for the separation process. In the second stage of the thesis, separation between authentic works of a master artist of the art history and his fakes which have been painted by different painters is aimed at. In solving this problem, Matisse’s authentic works and his fakes have been used. Key Words: Artificial Neural Networks, Support Vector Machines, Wavelet Transform Masterpieces, Fake, Pablo Picasso, Georges Braque and Henri Matisse.
In recent years, with the latest developments in computer technology intelligent systems which are being widely used in different fields such as geophysics, medicine and bio-medical have recently successfully high results. In this research, wavelets-ANN (Artificial Neural Networks) and wavelets-SVM (Support Vector Machine) processes have been used in the authentication and separation of masterpieces. Thus, achieving more objective and satisfactory results are targeted. In this thesis, in order to solve the problems of the authentication and separation of masterpieces, two methods have been applied. According to these two methods, firstly the separation of the masterpieces of the same period artist have been given priority and this context, the works of Pablo Picasso and Georges Braque have been used as data for the separation process. In the second stage of the thesis, separation between authentic works of a master artist of the art history and his fakes which have been painted by different painters is aimed at. In solving this problem, Matisse’s authentic works and his fakes have been used. Key Words: Artificial Neural Networks, Support Vector Machines, Wavelet Transform Masterpieces, Fake, Pablo Picasso, Georges Braque and Henri Matisse.