Recovery of image and video based on compressive sensing via tensor approximation and Spatio-temporal correlation
- PDF / 1,148,214 Bytes
- 18 Pages / 439.37 x 666.142 pts Page_size
- 67 Downloads / 196 Views
Recovery of image and video based on compressive sensing via tensor approximation and Spatio-temporal correlation Yuanhong Zhong 1 Qiang Li 3
1
2
1
2
& Jing Zhang & Zhaokun Zhou & Xinyu Cheng & Guan Huang &
Received: 31 January 2020 / Revised: 29 July 2020 / Accepted: 16 September 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
In recent years, block-based compressive sensing (BCS) has been extensively studied because it can reduce computational complexity and data storage by dividing the image into smaller patches, but the performance of the reconstruction algorithm is not satisfactory. In this paper, a new reconstruction model for image and video is proposed. The model makes full use of spatio-temporal correlation and utilizes low-rank tensor approximation to improve the quality of the reconstructed image and video. For image recovery, the proposed model obtains a lowrank approximation of a tensor formed by non-local similar patches, and improves the reconstruction quality from a spatial perspective by combining non-local similarity and lowrank property. For video recovery, the reconstruction process is divided into two phases. In the first phase, each frame of the video sequence is regarded as an independent image to be reconstructed by taking advantage of spatial property. The second phase performs tensor approximation through searching similar patches within frames near the target frame, to achieve reconstruction by putting the spatio-temporal correlation into full play. The resulting model is solved by an efficient Alternating Direction Method of Multipliers (ADMM) algorithm. A series of experiments show that the quality of the proposed model is comparable to the current state-of-the-art recovery methods. Keywords Block-based compressive sensing . Image and video recovery . Low-rank tensor approximation . High ordersingular valuedecomposition (HOSVD) . Spatio-temporal correlation
1 Introduction Traditional image and video sampling and reconstruction are constrained by the Nyquist sampling theory, which requires a sampling rate of no less than twice a signal’s bandwidth. Due to the * Yuanhong Zhong [email protected] Extended author information available on the last page of the article
Multimedia Tools and Applications
redundancy in the signal, it’s always followed by the computational complexity problem [4]. In order to efficiently remove transmission redundancy and decrease computation, the technology of compressive sensing (CS) [4] breaks the limitation of Nyquist sampling theory and has drawn great interests. CS requires only a small number of measurements by projecting the signal onto a random basis to achieve the reconstruction. More clearly, according to CS theory, a signal with a sparse representation in some domain can be recovered with high probability from these measurements [8, 10]. Video signals have more redundancy in both time and space, therefore, the CS theory has great prospects for sparse representation of video [35]. In past several years, comp
Data Loading...
