Current Search:   random field (x)

View All Items

  • CSV Spreadsheet
(1 - 4 of 4)
Gradient based MRF learning for image restoration and segmentation
Title
Gradient based MRF learning for image restoration and segmentation.
Creator
Samuel, Kegan, Tappen, Marshall, Da Vitoria Lobo, Niels, Foroosh, Hassan, Li, Xin, University of Central Florida
Abstract / Description
The undirected graphical model or Markov Random Field (MRF) is one of the more popular models used in computer vision and is the type of model with which this work is concerned. Models based on these methods have proven to be particularly useful in low-level vision systems and have led to state-of-the-art results for MRF-based systems. The research presented will describe a new discriminative training algorithm and its implementation.The MRF model will be trained by optimizing its parameters...
Show moreThe undirected graphical model or Markov Random Field (MRF) is one of the more popular models used in computer vision and is the type of model with which this work is concerned. Models based on these methods have proven to be particularly useful in low-level vision systems and have led to state-of-the-art results for MRF-based systems. The research presented will describe a new discriminative training algorithm and its implementation.The MRF model will be trained by optimizing its parameters so that the minimum energy solution of the model is as similar as possible to the ground-truth. While previous work has relied on time-consuming iterative approximations or stochastic approximations, this work will demonstrate how implicit differentiation can be used to analytically differentiate the overall training loss with respect to the MRF parameters. This framework leads to an efficient, flexible learning algorithm that can be applied to a number of different models.The effectiveness of the proposed learning method will then be demonstrated by learning the parameters of two related models applied to the task of denoising images. The experimental results will demonstrate that the proposed learning algorithm is comparable and, at times, better than previous training methods applied to the same tasks.A new segmentation model will also be introduced and trained using the proposed learning method. The proposed segmentation model is based on an energy minimization framework that is novel in how it incorporates priors on the size of the segments in a way that is straightforward to implement. While other methods, such as normalized cuts, tend to produce segmentations of similar sizes, this method is able to overcome that problem and produce more realistic segmentations.
Show less
Date Issued
2012
Identifier
CFE0004595, ucf:49207
Format
Document (PDF)
PURL
http://purl.flvc.org/ucf/fd/CFE0004595
CONTRIBUTIONS TO AUTOMATIC PARTICLE IDENTIFICATION IN ELECTRON MICROGRAPHS: ALGORITHMS, IMPLEMENTATION, AND APPLICATIONS
Title
CONTRIBUTIONS TO AUTOMATIC PARTICLE IDENTIFICATION IN ELECTRON MICROGRAPHS: ALGORITHMS, IMPLEMENTATION, AND APPLICATIONS.
Creator
Singh, Vivek, Marinescu, Dan, University of Central Florida
Abstract / Description
Three dimensional reconstruction of large macromolecules like viruses at resolutions below 8 \AA~ - 10 \AA~ requires a large set of projection images and the particle identification step becomes a bottleneck. Several automatic and semi-automatic particle detection algorithms have been developed along the years. We present a general technique designed to automatically identify the projection images of particles. The method utilizes Markov random field modelling of the projected images and...
Show moreThree dimensional reconstruction of large macromolecules like viruses at resolutions below 8 \AA~ - 10 \AA~ requires a large set of projection images and the particle identification step becomes a bottleneck. Several automatic and semi-automatic particle detection algorithms have been developed along the years. We present a general technique designed to automatically identify the projection images of particles. The method utilizes Markov random field modelling of the projected images and involves a preprocessing of electron micrographs followed by image segmentation and post processing for boxing of the particle projections. Due to the typically extensive computational requirements for extracting hundreds of thousands of particle projections, parallel processing becomes essential. We present parallel algorithms and load balancing schemes for our algorithms. The lack of a standard benchmark for relative performance analysis of particle identification algorithms has prompted us to develop a benchmark suite. Further, we present a collection of metrics for the relative performance analysis of particle identification algorithms on the micrograph images in the suite, and discuss the design of the benchmark suite.
Show less
Date Issued
2005
Identifier
CFE0000705, ucf:46610
Format
Document (PDF)
PURL
http://purl.flvc.org/ucf/fd/CFE0000705
COHERENCE PROPERTIES OF OPTICAL NEAR-FIELDS
Title
COHERENCE PROPERTIES OF OPTICAL NEAR-FIELDS.
Creator
Apostol, Adela, Dogariu, Aristide, University of Central Florida
Abstract / Description
Next generation photonics-based technologies will ultimately rely on novel materials and devices. For this purpose, phenomena at subwavelength scales are being studied to advance both fundamental knowledge and experimental capabilities. In this dissertation, concepts specific to near-field optics and experimental capabilities specific to near-field microscopy are used to investigate various aspects of the statistical properties of random electromagnetic fields in the vicinity of optically...
Show moreNext generation photonics-based technologies will ultimately rely on novel materials and devices. For this purpose, phenomena at subwavelength scales are being studied to advance both fundamental knowledge and experimental capabilities. In this dissertation, concepts specific to near-field optics and experimental capabilities specific to near-field microscopy are used to investigate various aspects of the statistical properties of random electromagnetic fields in the vicinity of optically inhomogeneous media which emit or scatter radiation. The properties of such fields are being characterized within the frame of the coherence theory. While successful in describing the far-field properties of optical fields, the fundamental results of the conventional coherence theory disregard the contribution of short-range evanescent waves. Nonetheless, the specific features of random fields at subwavelength distances from interfaces of real media are influenced by the presence of evanescent waves because, in this case, both propagating and nonpropagating components contribute to the detectable properties of the radiation. In our studies, we have fully accounted for both contributions and, as a result, different surface and subsurface characteristics of inhomogeneous media could be explored. We investigated different properties of random optical near-fields which exhibit either Gaussian or non-Gaussian statistics. We have demonstrated that characteristics of optical radiation such as first- and second-order statistics of intensity and the spectral density in the vicinity of random media are all determined by both evanescent waves contribution and the statistical properties of the physical interface. For instance, we quantified the subtle differences which exist between the near- and far-field spectra of radiation and we brought the first experimental evidence that, contrary to the predictions of the conventional coherence theory, the values of coherence length in the near field depend on the distance from the interface and, moreover, they can be smaller than the wavelength of light. The results included in this dissertation demonstrate that the statistical properties of the electromagnetic fields which exist in the close proximity of inhomogeneous media can be used to extract structural information. They also suggest the possibility to adjust the coherence properties of the emitted radiation by modifying the statistical properties of the interfaces. Understanding the random interference phenomena in the near-field could also lead to new possibilities for surface and subsurface diagnostics of inhomogeneous media. In addition, controlling the statistical properties of radiation at subwavelength scales should be of paramount importance in the design of miniaturized optical sources, detectors and sensors.
Show less
Date Issued
2005
Identifier
CFE0000408, ucf:46410
Format
Document (PDF)
PURL
http://purl.flvc.org/ucf/fd/CFE0000408
NEAR-FIELD OPTICAL INTERACTIONS AND APPLICATIONS
Title
NEAR-FIELD OPTICAL INTERACTIONS AND APPLICATIONS.
Creator
Haefner, David, Dogariu, Aristide, University of Central Florida
Abstract / Description
The propagation symmetry of electromagnetic fields is affected by encounters with material systems. The effects of such interactions, for example, modifications of intensity, phase, polarization, angular spectrum, frequency, etc. can be used to obtain information about the material system. However, the propagation of electromagnetic waves imposes a fundamental limit to the length scales over which the material properties can be observed. In the realm of near-field optics, this limitation is...
Show moreThe propagation symmetry of electromagnetic fields is affected by encounters with material systems. The effects of such interactions, for example, modifications of intensity, phase, polarization, angular spectrum, frequency, etc. can be used to obtain information about the material system. However, the propagation of electromagnetic waves imposes a fundamental limit to the length scales over which the material properties can be observed. In the realm of near-field optics, this limitation is overcome only through a secondary interaction that couples the high-spatial-frequency (but non-propagating) field components to propagating waves that can be detected. The available information depends intrinsically on this secondary interaction, which constitutes the topic of this study. Quantitative measurements of material properties can be performed only by controlling the subtle characteristics of these processes. This dissertation discusses situations where the effects of near-field interactions can be (i) neglected in certain passive testing techniques, (ii) exploited for active probing of static or dynamic systems, or (iii) statistically isolated when considering optically inhomogeneous materials. This dissertation presents novel theoretical developments, experimental measurements, and numerical results that elucidate the vectorial aspects of the interaction between light and nano-structured material for use in sensing applications.
Show less
Date Issued
2010
Identifier
CFE0003095, ucf:48318
Format
Document (PDF)
PURL
http://purl.flvc.org/ucf/fd/CFE0003095