Advanced search options

Sorted by: relevance · author · university · date | New search

You searched for `+publisher:"University of Michigan" +contributor:("Pappas, Thrasyvoulos N")`

.
Showing records 1 – 3 of
3 total matches.

▼ Search Limiters

University of Michigan

1. Zhai, Yuanhao. Perceptual Image Similarity Metrics and Applications.

Degree: PhD, Electrical Engineering: Systems, 2015, University of Michigan

URL: http://hdl.handle.net/2027.42/113586

This dissertation presents research in perceptual image similarity metrics and applications, e.g., content-based image retrieval, perceptual image compression, image similarity assessment and texture analysis.
The first part aims to design texture similarity metrics consistent with human perception. A new family of statistical texture similarity features, called Local Radius Index (LRI), and corresponding similarity metrics are proposed. Compared to state-of-the-art metrics in the STSIM family, LRI-based metrics achieve better texture retrieval performance with much less computation. When applied to the recently developed perceptual image coder, Matched Texture Coding (MTC), they enable similar performance while significantly accelerating encoding. Additionally, in photographic paper classification, LRI-based metrics also outperform pre-existing metrics. To fulfill the needs of texture classification and other applications, a rotation-invariant version of LRI, called Rotation-Invariant Local Radius Index (RI-LRI), is proposed. RI-LRI is also grayscale and illuminance insensitive. The corresponding similarity metric achieves texture classification accuracy comparable to state-of-the-art metrics. Moreover, its much lower dimensional feature vector requires substantially less computation and storage than other state-of-the-art texture features.
The second part of the dissertation focuses on bilevel images, which are images whose pixels are either black or white. The contributions include new objective similarity metrics intended to quantify similarity consistent with human perception, and a subjective experiment to obtain ground truth for judging the performance of objective metrics. Several similarity metrics are proposed that outperform existing ones in the sense of attaining significantly higher Pearson and Spearman-rank correlations with the ground truth. The new metrics include Adjusted Percentage Error, Bilevel Gradient Histogram, Connected Components Comparison and combinations of such.
Another portion of the dissertation focuses on the aforementioned MTC, which is a block-based image coder that uses texture similarity metrics to decide if blocks of the image can be encoded by pointing to perceptually similar ones in the already coded region. The key to its success is an effective texture similarity metric, such as an LRI-based metric, and an effective search strategy. Compared to traditional image compression algorithms, e.g., JPEG, MTC achieves similar coding rate with higher reconstruction quality. And the advantage of MTC becomes larger as coding rate decreases.
*Advisors/Committee Members: Neuhoff, David L. (committee member), Zhang, Jun (committee member), Pappas, Thrasyvoulos N. (committee member), Fessler, Jeffrey A. (committee member).*

Subjects/Keywords: Texture and image similarity metric; Texture and image retrieval; Texture and image classification; Perceptual image compression; Subjective experiment; Bilevel image similarity; Electrical Engineering; Engineering

Record Details Similar Records

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^{th} Edition):

Zhai, Y. (2015). Perceptual Image Similarity Metrics and Applications. (Doctoral Dissertation). University of Michigan. Retrieved from http://hdl.handle.net/2027.42/113586

Chicago Manual of Style (16^{th} Edition):

Zhai, Yuanhao. “Perceptual Image Similarity Metrics and Applications.” 2015. Doctoral Dissertation, University of Michigan. Accessed October 27, 2020. http://hdl.handle.net/2027.42/113586.

MLA Handbook (7^{th} Edition):

Zhai, Yuanhao. “Perceptual Image Similarity Metrics and Applications.” 2015. Web. 27 Oct 2020.

Vancouver:

Zhai Y. Perceptual Image Similarity Metrics and Applications. [Internet] [Doctoral dissertation]. University of Michigan; 2015. [cited 2020 Oct 27]. Available from: http://hdl.handle.net/2027.42/113586.

Council of Science Editors:

Zhai Y. Perceptual Image Similarity Metrics and Applications. [Doctoral Dissertation]. University of Michigan; 2015. Available from: http://hdl.handle.net/2027.42/113586

University of Michigan

2. Prelee, Matthew A. Manhattan Cutset Sampling and Sensor Networks.

Degree: PhD, Electrical Engineering: Systems, 2016, University of Michigan

URL: http://hdl.handle.net/2027.42/120876

Cutset sampling is a new approach to acquiring two-dimensional data, i.e., images, where values are recorded densely along straight lines. This type of sampling is motivated by physical scenarios where data must be taken along straight paths, such as a boat taking water samples. Additionally, it may be possible to better reconstruct image edges using the dense amount of data collected on lines. Finally, an advantage of cutset sampling is in the design of wireless sensor networks. If battery-powered sensors are placed densely along straight lines, then the transmission energy required for communication between sensors can be reduced, thereby extending the network lifetime.
A special case of cutset sampling is Manhattan sampling, where data is recorded along evenly-spaced rows and columns. This thesis examines Manhattan sampling in three contexts. First, we prove a sampling theorem demonstrating an image can be perfectly reconstructed from Manhattan samples when its spectrum is bandlimited to the union of two Nyquist regions corresponding to the two lattices forming the Manhattan grid. An efficient ``onion peeling'' reconstruction method is provided, and we show that the Landau bound is achieved. This theorem is generalized to dimensions higher than two, where again signals are reconstructable from a Manhattan set if they are bandlimited to a union of Nyquist regions. Second, for non-bandlimited images, we present several algorithms for reconstructing natural images from Manhattan samples. The Locally Orthogonal Orientation Penalization (LOOP) algorithm is the best of the proposed algorithms in both subjective quality and mean-squared error. The LOOP algorithm reconstructs images well in general, and outperforms competing algorithms for reconstruction from non-lattice samples. Finally, we study cutset networks, which are new placement topologies for wireless sensor networks. Assuming a power-law model for communication energy, we show that cutset networks offer reduced communication energy costs over lattice and random topologies. Additionally, when solving centralized and decentralized source localization problems, cutset networks offer reduced energy costs over other topologies for fixed sensor densities and localization accuracies. Finally, with the eventual goal of analyzing different cutset topologies, we analyze the energy per distance required for efficient long-distance communication in lattice networks.
*Advisors/Committee Members: Neuhoff, David L (committee member), Gilbert, Anna Catherine (committee member), Dick, Robert (committee member), Fessler, Jeffrey A (committee member), Pappas, Thrasyvoulos N (committee member).*

Subjects/Keywords: image sampling; sensor networks; Electrical Engineering; Engineering

Record Details Similar Records

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^{th} Edition):

Prelee, M. A. (2016). Manhattan Cutset Sampling and Sensor Networks. (Doctoral Dissertation). University of Michigan. Retrieved from http://hdl.handle.net/2027.42/120876

Chicago Manual of Style (16^{th} Edition):

Prelee, Matthew A. “Manhattan Cutset Sampling and Sensor Networks.” 2016. Doctoral Dissertation, University of Michigan. Accessed October 27, 2020. http://hdl.handle.net/2027.42/120876.

MLA Handbook (7^{th} Edition):

Prelee, Matthew A. “Manhattan Cutset Sampling and Sensor Networks.” 2016. Web. 27 Oct 2020.

Vancouver:

Prelee MA. Manhattan Cutset Sampling and Sensor Networks. [Internet] [Doctoral dissertation]. University of Michigan; 2016. [cited 2020 Oct 27]. Available from: http://hdl.handle.net/2027.42/120876.

Council of Science Editors:

Prelee MA. Manhattan Cutset Sampling and Sensor Networks. [Doctoral Dissertation]. University of Michigan; 2016. Available from: http://hdl.handle.net/2027.42/120876

3. Reyes, Matthew G. Cutset Based Processing and Compression of Markov Random Fields.

Degree: PhD, Electrical Engineering: Systems, 2011, University of Michigan

URL: http://hdl.handle.net/2027.42/84510

This thesis presents results related to the compression a Markov random field (MRF) bfX defined on a graph G=(V,E) by first losslessly compressing a cutset of sites U and then either losslessly compressing or estimating the remaining sites conditioned on the cutset. We present analytical solutions to the MAP estimate of a block conditioned on the commonly occurring boundaries with two or fewer runs of black, for both 4 pt. and 8 pt. grid graphs. Using these results we empirically demonstrate that Max-Product Loopy Belief Propagation converges to the correct results. We present a simple adaptive Arithmetic Encoding (AC) based method for losslessly compressing a square grid cutset of a binary image and, applying the Ising reconstruction results, show that the resulting lossy image coder is competitive compared to other methods. We present rigorous development of Local Conditioning for MRFs, algorithm for exact inference in cyclic graphs. We prove that the entropy of family of MRFs is monotone increasing in the associated exponential parameters and that the exponential parameters for the moment-matching reduced MRF induced by U for a subset of nodes are component-wise greater than the original exponential parameters within U. We also show that the divergence between an MRF induced by exponential parameter θ and another induced by θ' is monotone increasing in θ'. Furthermore, we prove that the divergence between the marginal distribution for bfX and reduced MRF follows a Pythagorean decomposition, providing reduced MRF analogue to well-known result in information geometry. We present efficient algorithms for optimal AC based lossless compression of acyclic and EASY cyclic MRFs, and use these for suboptimal lossless compression for HARD cyclic MRFs, called {em Reduced Cutset Coding}. Experiments with RCC on homogeneous Ising models both verify nearly optimal performance and provide estimates of upper and lower bounds to entropy.
*Advisors/Committee Members: Neuhoff, David L. (committee member), Blass, Andreas R. (committee member), Hero Iii, Alfred O. (committee member), Pappas, Thrasyvoulos N. (committee member), Sadanandarao, Sandeep P. (committee member).*

Subjects/Keywords: Markov Random Fields; Source Coding; Belief Propagation; Cutset; Ising Model; Monotonicity; Electrical Engineering; Engineering

Record Details Similar Records

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^{th} Edition):

Reyes, M. G. (2011). Cutset Based Processing and Compression of Markov Random Fields. (Doctoral Dissertation). University of Michigan. Retrieved from http://hdl.handle.net/2027.42/84510

Chicago Manual of Style (16^{th} Edition):

Reyes, Matthew G. “Cutset Based Processing and Compression of Markov Random Fields.” 2011. Doctoral Dissertation, University of Michigan. Accessed October 27, 2020. http://hdl.handle.net/2027.42/84510.

MLA Handbook (7^{th} Edition):

Reyes, Matthew G. “Cutset Based Processing and Compression of Markov Random Fields.” 2011. Web. 27 Oct 2020.

Vancouver:

Reyes MG. Cutset Based Processing and Compression of Markov Random Fields. [Internet] [Doctoral dissertation]. University of Michigan; 2011. [cited 2020 Oct 27]. Available from: http://hdl.handle.net/2027.42/84510.

Council of Science Editors:

Reyes MG. Cutset Based Processing and Compression of Markov Random Fields. [Doctoral Dissertation]. University of Michigan; 2011. Available from: http://hdl.handle.net/2027.42/84510