Sparse representation of visual data for compression and compressed sensing

Abstract

The cover of this thesis depicts the sparse representation of a light field. From the back side to front, a light field is divided into small elements, then these elements (signals) are projected onto an ensemble of two 5D dictionaries to produce a sparse coefficient vector. I hope that the reader can deduce where the dimensionality of those Rubik's Cube looking matrices come from after reading Chapter 3 of this thesis. The light field used on the cover is a part of the Stanford Lego Gantry database (http://lightfield.stanford.edu/)

A Novel Efficient Secure and Error-Robust Scheme for Internet of Things Using Compressive Sensing

“…In CS, the reconstruction depends on the selection of a sparsifying basis of the signal. It has been shown that the reconstruction quality using an over-complete dictionary is better compared to one orthogonal transform dictionary [53]. Considering the IoT ecosystem, the reconstruction will be performed either on the edge or core cloud.…”

A Unified Framework for Compression and Compressed Sensing of Light Fields and Light Field Videos

“…Preliminary results of utilizing our framework for the compression of scene appearance data sets (i.e. spherical light fields), as well as compressive BRDF measurements for accelerating the capturing process of a gonioreflectometer are reported in [Miandji 2018]. With respect to future work, we believe that our theoretical results can be significantly improved by replacing the mutual coherence with another metric to tighten the bound on sparsity.…”

“…The trained ensemble enables a high degree of sparsity in the transformation domain, which has been shown to be an important factor for efficient compression [Mallat 2008;Miandji et al 2013;Zepeda et al 2011] and compressed sensing [Candès et al 2006;. While we only focus on light fields and light field videos, the framework can be readily applied for compression and compressed sensing of commonly used high dimensional data sets in graphics such as BTFs [Dana et al 1999], measured BRDFs, hyperspectral images and videos (see [Miandji 2018] for a few examples).…”

Multi-Mask Camera Model for Compressed Acquisition of Light Fields

“…a) Exposure time impact analysis: We note that all the acquisitions in this experiment are performed with the same amount of exposure time to conduct a fair comparison between them. Here, we consider three-shot acquisitions when the monochrome sensor is used, since only one monochrome captured image does not allow to recover color information as depicted in [62]. More precisely, given a total time t , the acquisition time of each color channel of full RGB images is t…”