Morphing techniques for facial images

MIYALKOVSKA, ELENA (author), Šajn, Luka (mentor)

Image morphing is an image processing technique where one image gradually transforms into another through a smooth transition over a period time. Multiple morphing techniques exist including cross-dissolving, mesh warping, energy minimization, and others. In the scope of this thesis, two morphing algorithms were discussed and implemented. The first one uses only the cross-dissolving technique. The second one uses facial landmarks auto-detection, space triangulation, warping, and cross-dissolving. A software solution in the programming language Python was prepared for morphing between source and target facial images. Using the software, we have produced some very smooth facial image transitions including morphing between the same and different faces. We have also successfully performed mood changing with the help of morphing between different facial expressions.

• Language: English
• Keywords: face morphing, image transformations, cross-dissolving, image warping, mesh warping, field morphing
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2022
• PID: 20.500.12556/RUL-135682
• COBISS.SI-ID: 102853379
• Publication date in RUL: 25.03.2022

Secondary language

• Language: Slovenian
• Title: Tehnike za preoblikovanje obraznih slik

Abstract

Preoblikovanje slike (angl. morphing) je tehnika za obdelavo slik, pri kateri se ena slika postopoma preoblikuje v drugo z gladkim prehodom v določenem časovnem obdobju. Obstaja več tehnik preoblikovanja, in sicer povprečje dveh slik, deformacija mreže, minimizacija energije in drugi. V okviru diplomske naloge sta bila obravnavana in implementirana dva algoritma preoblikovanja. Prvi uporablja samo tehniko povprečje dveh slik. Drugi uporablja samodejno zaznavanje in označevanje obraznih točk, triangulacijo prostora, deformacijo in povpreče dveh slik. Pripravljena je bila programska rešitev v programskem jeziku Python za preoblikovanje med izvorno in ciljno sliko obraza. S programsko rešitvijo smo izdelali nekaj zelo gladkih prehodov med enakimi in različnimi obrazi. Uspešno smo izvedli tudi spreminjanje razpoloženja posameznika s pomočjo preoblikovanja med različnimi obraznimi izrazi.

• Keywords: preoblikovanje obraznih slik, transformacija slik, povprečje dveh slik, deformacija slik, deformacija mreže