Author

Yifei Pei

Date of Award

6-4-2020

Document Type

Thesis

Publisher

Santa Clara : Santa Clara University, 2020.

Degree Name

Master of Science (MS)

Departments

Computer Science and Engineering

First Advisor

Nam Ling

Second Advisor

Ying Liu

Abstract

Compressive sensing (CS) is a signal processing framework that effectively recovers a signal from a small number of samples. Traditional compressed sensing algorithms, such as basis pursuit (BP) and orthogonal matching pursuit (OMP) have several drawbacks, such as low reconstruction performance at small compressed sensing rates and high time complexity. Recently, researchers focus on deep learning to get compressive sensing matrix and reconstruction operations collectively. However, they failed to consider sparsity in their neural networks to compressive sensing recovery; thus, the reconstruction performances are still unsatisfied. In this thesis, we use 2D-discrete cosine transform and 2D-discrete wavelet transform to impose sparsity of recovered signals to deep learning in video frame compressive sensing. We find the reconstruction performance significantly enhanced. We also propose another compressive sensing framework utilizing the similarities between frame blocks to optimize deep learning training process for compressive video sensing by applying the Gaussian-mixture models(GMM). The results show our proposed framework further improves the quality of video reconstruction and stabilize the reconstruction across various video sequences. We also propose a potential research direction after this research.

The thesis's primary contribution are: (1) for the first moment it introduces the use of discrete cosine transformed images and discrete wavelet transformed images in deep learning for compressive sensing tasks; (2) we further apply GMMs to optimize our deep learning framework for compressive video sensing; (3) by combining discrete cosine transform and GMMs with deep learning, we propose our nal neural network architecture to achieve stronger reconstruction quality of compressed sensed video frames.

Share

COinS