Autofocus criteria applications in holograms of marine plankton: laboratory and field observations in southeastern Brazil and the Gulf of Mexico

• Ano de defesa: 2024
• Autor(a) principal: Vega, Gelaysi Moreno
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: automatic classification; biophysical interactions; classificação automática; counting methods; detecção de foco; digital holography; distribuição de partículas; focus detection; holografia digital; holographic reconstruction; interações biofísicas; métodos de contagem; particle distributions; plâncton; plankton; reconstrução holográfica
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/21/21134/tde-25062024-105210/

In this thesis, fundamental concepts of holographic reconstruction are applied to enhance the imaging process of marine plankton. Digital holography is relevant in this context as it allows the refocusing of particles and organisms at different planes of the optical path, thereby increasing the imaged volume while maintaining detailed structures of interest for automatic classification. In the first section of this work, a focus metric called image quality (Q) is proposed and comparatively evaluated, estimated based on numerical simulation and experimental data obtained in the laboratory through the combination of three focus functions. A second component of this research assessed the applicability of the focus criterion (Q) in plankton studies. Samples prepared with different concentrations of live organisms were filmed using an in-line digital holography system. After segmentation and reconstruction using the Q index as the autofocus criterion, the number of organisms detected per holography was compared with counts under a stereomicroscope. The study showed a high linear agreement between the two methods, indicating that the use of digital holography in combination with the Q estimate is reliable for plankton quantification. A third study was conducted to determine the best focus criterion among a total of 18 functions, considering different taxonomic groups, organism morphology, and camera sensor characteristics. The definition function (D) performed well, but the Q function generated lower average error in all tested situations, standing out as the most efficient function for holographic reconstruction of planktonic organisms. The application of the Q function in in situ holograms was tested with data generated by a submersible digital holography system in the northern Gulf of Mexico. The holograms were segmented and reconstructed using the Q function for subsequent automatic classification. The concentration of particles and organisms obtained from the holograms was evaluated in relation to density, temperature, salinity, and chlorophyll along the water column to determine vertical distribution. Thin layers of phytoplankton were detected near the picnocline/halocline, in a region of high stratification and vertical stability. The phytoplankton was dominated by diatom species (Chaetoceros debilis, Chaetoceros socialis, and other centric diatoms). Colonies of C. socialis and chains of C. debilis showed multiple concentration peaks, with the absolute maximum coinciding with the thin layers. Zooplankton was dominated by Copepoda and gelatinous organisms (Chaetognatha and Appendicularia), observed in higher concentration usually above the picnocline. Heterogeneous zooplankton aggregations were recorded throughout the water column. The set of studies conducted in this work laid the groundwork for improving the hologram autofocus process and confirmed the potential application of digital holography techniques in marine plankton research.