In this bachelor thesis, innovative techniques of integrated architectural surveying, performed through laser scanning and aerophotogrammetry for cultural heritage’s knowledge and qualification, have been investigated and tested. The topic, of absolute relevance, has been approached with reference to the intricate case of study inherent to Venosa’s Most Holy Trinity Complex (PZ) – characterized by its deep historical roots dating back to the late-ancient era – by integrating different typologies of relief-techniques to generate a single numerical model of points. In particular, the measuring implemental operation was approached with the aim of reaching high levels of accuracy and resolution, which would consolidate in a detailed and reliable three-dimensional model of the architectural good. Many were the difficulties faced during the work, given the complexity and size of the site, which have made the integration of data acquired in successive and independent phases with the techniques “image-based” and “range-based” problematic. A further objective of the work was to investigate the chance of using artificial intelligence techniques (“Machine & Deep Learning”), applied on a portion of the point cloud belonging to the S-W outer wall of the Unfinished church, part of the abbey complex, in order to rate its pathological and degradation phenomena. Two approaches of Supervised Classification have been adopted, defined “Geometry & Color based” and “Texture based”, by using input-data sets catalogued on vector, geometric and radiometric features. The result was the extraction of semantic segmentation classes of the point cloud and of the three-dimensional model, highlighting their strengths and weaknesses. Thanks to the automation mechanism based on learning algorithms (Random Forest), interesting metric data related to the analysis of the degradation phenomena of the examined wall have been extracted. Later, the obtained results were overlayed on a specially developed HBIM model, through a methodology called “Scan to BIM”, also usable in the immersive and semi-immersive virtual reality online platform. The application of HBIM to the existing heritage offers many opportunities to improve and optimize the management, the protection and the valorization of the intended heritage, together with the monitoring and the potential types of operation to execute. The quality of the 3D reconstruction and the perfect overlaying of the virtual scene to reality, were the presuppositions of the executed project.
Nel seguente lavoro sviluppato nell’ambito della tesi, sono state approfondite e sperimentate tecniche innovative di rilievo architettonico integrato, eseguite tramite scansione laser e aerofotogrammetria, per la conoscenza e la qualificazione del patrimonio culturale. Il tema, di assoluta attualità, è stato affrontato con riferimento all’articolato caso di studio inerente al Complesso abbaziale della SS. Trinità di Venosa (PZ), dalle profonde radici storiche risalenti all’epoca tardo-antica, integrando tecniche di rilievo diverse per generare un unico modello numerico di punti. In particolare, è stata approcciata l’operazione mensoria strumentale con l’obiettivo di raggiugere alti livelli di accuratezza e risoluzione che si concretizzassero in un modello tridimensionale dettagliato e affidabile del bene architettonico. Non poche sono state le difficoltà affrontate durante il lavoro, data la complessità e le dimensioni del sito, che hanno reso problematica l’integrazione dei dati acquisiti in fasi successive e indipendenti con le tecniche “image-based” e “range -based”. Un ulteriore obiettivo del lavoro è stato quello di indagare la possibilità di utilizzo di tecniche di intelligenza artificiale (“Machine & Deep Learning”), applicate su una porzione della nuvola di punti inerente al paramento murario esterno S-O della chiesa Incompiuta, facente parte del complesso abbaziale, al fine di valutare i fenomeni patologici e di degrado presenti. Sono stati adottati due approcci di Classificazione supervisionata, definiti “Geometry & Color based” e “Texture based”, utilizzando set di dati input catalogati su caratteristiche vettoriali, geometriche e radiometriche. Risultato conseguito è stata l’estrazione di classi di segmentazione semantica della nuvola di punti e del modello tridimensionale, evidenziandone i punti di forza e di debolezza. Grazie al meccanismo di automazione basato su algoritmi di apprendimento (Random Forest), sono stati estratti dati metrici interessanti inerenti all’analisi dei fenomeni di degrado del paramento murario in esame. Successivamente i risultati ottenuti sono stati sovrapposti ad un modello HBIM appositamente elaborato, attraverso una metodologia denominata “Scan to BIM”, fruibile anche in realtà virtuale immersiva e semi-immersiva su piattaforma online. L'applicazione dell’HBIM al patrimonio esistente offre non poche opportunità di ottimizzare la gestione, la tutela, la valorizzazione dell’opera costruita, nonché il monitoraggio e gli eventuali tipi di intervento da eseguire. Qualità della ricostruzione 3D e perfetta sovrapposizione della scena virtuale alla realtà, sono stati i presupposti del lavoro eseguito.
TECNICHE INNOVATIVE DI RILIEVO ARCHITETTONICO INTEGRATO E DI SUPERVISED MACHINE LEARNING PER LA CONOSCENZA DEL PATRIMONIO CULTURALE. IL COMPLESSO ABBAZIALE DELLA SS. TRINITÀ DI VENOSA (PZ).
BULDO, MICHELE
2018/2019
Abstract
In this bachelor thesis, innovative techniques of integrated architectural surveying, performed through laser scanning and aerophotogrammetry for cultural heritage’s knowledge and qualification, have been investigated and tested. The topic, of absolute relevance, has been approached with reference to the intricate case of study inherent to Venosa’s Most Holy Trinity Complex (PZ) – characterized by its deep historical roots dating back to the late-ancient era – by integrating different typologies of relief-techniques to generate a single numerical model of points. In particular, the measuring implemental operation was approached with the aim of reaching high levels of accuracy and resolution, which would consolidate in a detailed and reliable three-dimensional model of the architectural good. Many were the difficulties faced during the work, given the complexity and size of the site, which have made the integration of data acquired in successive and independent phases with the techniques “image-based” and “range-based” problematic. A further objective of the work was to investigate the chance of using artificial intelligence techniques (“Machine & Deep Learning”), applied on a portion of the point cloud belonging to the S-W outer wall of the Unfinished church, part of the abbey complex, in order to rate its pathological and degradation phenomena. Two approaches of Supervised Classification have been adopted, defined “Geometry & Color based” and “Texture based”, by using input-data sets catalogued on vector, geometric and radiometric features. The result was the extraction of semantic segmentation classes of the point cloud and of the three-dimensional model, highlighting their strengths and weaknesses. Thanks to the automation mechanism based on learning algorithms (Random Forest), interesting metric data related to the analysis of the degradation phenomena of the examined wall have been extracted. Later, the obtained results were overlayed on a specially developed HBIM model, through a methodology called “Scan to BIM”, also usable in the immersive and semi-immersive virtual reality online platform. The application of HBIM to the existing heritage offers many opportunities to improve and optimize the management, the protection and the valorization of the intended heritage, together with the monitoring and the potential types of operation to execute. The quality of the 3D reconstruction and the perfect overlaying of the virtual scene to reality, were the presuppositions of the executed project.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi magistrale Buldo Michele 573235.pdf
File consultabile da postazioni biblioteca
Descrizione: Tecniche innovative di rilievo architettonico integrato e di Supervised Machine Learning per la conoscenza del patrimonio culturale. Il Complesso abbaziale della SS. Trinità di Venosa (PZ).
Dimensione
4.2 MB
Formato
Adobe PDF
|
4.2 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14253/65104