Temporal multiple moving objects recognition using shape-based descriptor matching

Simple item page
dc.contributor.authorZia, Juwairiah
dc.contributor.examiningcommitteeAshraf, Ahmed (Electrical and Computer Engineering)
dc.contributor.examiningcommitteeThomas, Gabriel (Electrical and Computer Engineering)
dc.contributor.supervisorPeters, James F.
dc.date.accessioned2023-08-09T19:39:38Z
dc.date.available2023-08-09T19:39:38Z
dc.date.issued2023-08-05
dc.date.submitted2023-08-05T06:33:19Zen_US
dc.date.submitted2023-08-09T02:58:26Zen_US
dc.degree.disciplineElectrical and Computer Engineeringen_US
dc.degree.levelMaster of Science (M.Sc.)
dc.description.abstractExtracting foregrounds from complex outdoor backgrounds in the temporal domain presents a significant challenge in object recognition. The primary reason is the changes in luminance observed in the moving foregrounds of video frames and the dynamic background leading to loss of minute changes undergone by foreground objects. This compromises the accuracy of object identification patterns and can lead to misclassifications in detection and recognition. These difficulties hinder the automatic analysis and effective classification of computer vision systems in real-world scenarios, making manual performance costly. Work has been done to address these challenges by employing computational geometry to estimate descriptive features of objects within images. In this thesis, we extend this notion by proposing an approach to estimating shape descriptors (features) in the temporal domain of moving foregrounds by analyzing shape's topological space properties for video processing. To achieve this, we introduce the concept of optical vortex nerves, which will be applied to various moving object videos. By constructing vortex nerves using vertex termed 0-simplices, edges termed 1-simplices, and triangles termed 2-simplices in the foregrounds, we can derive shape-based spatial-temporal descriptor sets for each moving object. Moreover, I then employed the shape descriptor sets for shape matching to assess the descriptive proximity between various vehicles and establish their descriptive nearness (similarity) for recognition. This practical implementation serves as an embodiment of the aforementioned concepts discussed throughout this document.
dc.description.noteOctober 2023
dc.identifier.urihttp://hdl.handle.net/1993/37447
dc.language.isoeng
dc.rightsopen accessen_US
dc.subjectimage similarity retrieval
dc.subjectmultiple object recognition
dc.subjecttopological space properties
dc.subjecttemporal domain
dc.subjectnear set
dc.subjectproximity
dc.subjectdescriptive proximity
dc.subjectdescriptive nearness
dc.subjectdescriptive apartness
dc.subjectdelaunay triangulation
dc.subjectmaximal nucleus cluster
dc.subjectshape
dc.titleTemporal multiple moving objects recognition using shape-based descriptor matching
dc.typemaster thesisen_US
local.subject.manitobano
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Zia_Juwairiah.pdf
Size:
13.14 MB
Format:
Adobe Portable Document Format
Description:
Download
License bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
license.txt
Size:
770 B
Format:
Item-specific license agreed to upon submission
Description:
Download
Collections
FGS - Electronic Theses and Practica