Digitala Vetenskapliga Arkivet

Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Performance Evaluation of Computer-Generated Holography Algorithms
KTH, School of Electrical Engineering and Computer Science (EECS).
2025 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesisAlternative title
Prestandautvärdering av Algoritmer för Datorgenererad Holografi (Swedish)
Abstract [en]

This thesis investigates the performance of various computer-generated holography algorithms. Computer-generated holography is a promising technology for rendering three-dimensional images through holograms without requiring physical objects. However, its computational complexity and hologram generation quality remain challenging despite the rapid advancements in ML algorithms especially in high-resolution display applications like augmented and virtual reality. This research compares the computational efficiency and hologram quality between iterative algorithms and contemporary deep learning techniques. Using a quantitative approach, algorithms are evaluated based on accuracy, memory usage, and processing time, and they are implemented on standard computational platforms. The research compares the performance of iterative algorithms, such as the Gerchberg-Saxton (GS) method, with advanced machine learning approaches including Convolutional Neural Networks (CNNs) and Transformers. The results indicate significant differences in performance between the two methods. The iterative algorithm, while robust, often requires longer processing times and higher computational resources. ML algorithms, particularly CNNs and Transformers, offer significant improvements in processing speed and scalability, which are critical for real-time CGH applications. However, these benefits come with challenges in model complexity and the need for extensive training data sets. The study also highlights the potential of Transformers in handling complex spatial relationships in holography. This study contributes to the field of digital holography by providing a detailed analysis of CGH algorithms. The findings suggest that while ML offers substantial benefits in speed and scalability, there is a crucial need for further research and comparison to improve the time consumption and enhance the quality of the holograms produced. This work lays the groundwork for future explorations into CGH algorithms.

Abstract [sv]

Denna avhandling undersöker prestandan hos olika algoritmer för dator- genererad holografi. Datorgenererad holografi är en lovande teknik för att rendera tredimensionella bilder genom hologram utan att kräva fysiska prototyper. Dock kvarstår utmaningar gällande dess beräkningskomplexitet och kvaliteten på hologramgenerering, särskilt med de snabba framstegen inom ML-algoritmer. Denna forskning jämför beräkningsmässig effektivitet och hologramkvalitet mellan iterativa algoritmer och moderna djupinlär- ningstekniker. Genom en kvantitativ metod utvärderas algoritmerna baserat på noggrannhet, minnesanvändning och bearbetningstid, implementerat på standardberäkningsplattformar. Forskningen jämför prestandan hos iterativa algoritmer, såsom Gerchberg- Saxton (GS) metoden, med avancerade maskininlärningsmetoder inklusive faltning neurala nätverk (CNNs) och transformatorer. Resultaten visar betydande skillnader i prestanda mellan de två metoderna. Den iterativa algoritmen, som är robust, kräver ofta längre bearbetningstider och större beräkningsresurser. ML-algoritmer, särskilt CNNs och transformatorer, er- bjuder betydande förbättringar i bearbetningshastighet och skalbarhet, vilket är avgörande för realtidsapplikationer av CGH. Dessa fördelar medför dock utmaningar i modellkomplexitet och behovet av omfattande träningsdata. Studien belyser även potentialen hos transformatorer att hantera komplexa spatiala relationer i holografi. Denna studie bidrar till fältet för digital holografi genom att tillhandahålla en detaljerad analys av CGH-algoritmer. Resultaten tyder på att medan ML erbjuder betydande fördelar i hastighet och skalbarhet, finns det ett avgörande behov av vidare forskning för att förbättra tidsåtgången och höja kvaliteten på de producerade hologrammen. Detta arbete lägger grunden för framtida utforskningar av CGH-algoritmer.

Place, publisher, year, edition, pages
2025. , p. 67
Series
TRITA-EECS-EX ; 2025:35
Keywords [en]
Computer-Generated Holography, Machine Learning, Gerchberg-Saxton Algorithm, Convolutional Neural Networks, Transformers
Keywords [sv]
Datorgenererad Holografi, Maskininlärning, Gerchberg-Saxton Algoritmen, Faltning Neurala Nätverk, Transformatorer
National Category
Computer and Information Sciences
Identifiers
URN: urn:nbn:se:kth:diva-361840OAI: oai:DiVA.org:kth-361840DiVA, id: diva2:1948914
External cooperation
Ericsson AB
Supervisors
Examiners
Available from: 2025-04-03 Created: 2025-04-01 Last updated: 2025-04-03Bibliographically approved

Open Access in DiVA

fulltext(7736 kB)54 downloads
File information
File name FULLTEXT02.pdfFile size 7736 kBChecksum SHA-512
f12d724f4791d4adaca18a07895b3356145616e9aafed138653e38b1feaff0738b4f9d1ce58eec757421923037179483add1ea9957b61de686de7ed3bbd170d2
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Computer and Information Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 55 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 245 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf