Image restoration using truncated SVD filter bank based on an energy criterion

Zhang, Xiaoxin; Wang, S.

doi:10.1049/ip-vis:20045200

Cited by 7 publications

(4 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although a complete comparison of autoencoders with traditional orthogonal function decomposition is out of scope for this paper, we computed performance metrics for multiple PCA-based models to compare against the ABM presented in this work. The considered PCA variations taken into consideration in the analysis were PCA 49 , Sparse PCA 50 , and Truncated PCA 51 . Based on this analysis, we arrived at the following conclusions: For our dataset and models, (i) PCA has larger memory storage requirements than the decoder, 90 MB versus 0.211 MB for 32-bit floating point numbers.…”

Section: Resultsmentioning

confidence: 99%

Predicting wind-driven spatial deposition through simulated color images using deep autoencoders

Fernández-Godino

Lucas

Kong

2023

Sci Rep

View full text Add to dashboard Cite

For centuries, scientists have observed nature to understand the laws that govern the physical world. The traditional process of turning observations into physical understanding is slow. Imperfect models are constructed and tested to explain relationships in data. Powerful new algorithms can enable computers to learn physics by observing images and videos. Inspired by this idea, instead of training machine learning models using physical quantities, we used images, that is, pixel information. For this work, and as a proof of concept, the physics of interest are wind-driven spatial patterns. These phenomena include features in Aeolian dunes and volcanic ash deposition, wildfire smoke, and air pollution plumes. We use computer model simulations of spatial deposition patterns to approximate images from a hypothetical imaging device whose outputs are red, green, and blue (RGB) color images with channel values ranging from 0 to 255. In this paper, we explore deep convolutional neural network-based autoencoders to exploit relationships in wind-driven spatial patterns, which commonly occur in geosciences, and reduce their dimensionality. Reducing the data dimension size with an encoder enables training deep, fully connected neural network models linking geographic and meteorological scalar input quantities to the encoded space. Once this is achieved, full spatial patterns are reconstructed using the decoder. We demonstrate this approach on images of spatial deposition from a pollution source, where the encoder compresses the dimensionality to 0.02% of the original size, and the full predictive model performance on test data achieves a normalized root mean squared error of 8%, a figure of merit in space of 94% and a precision-recall area under the curve of 0.93.

show abstract

Section: Resultsmentioning

confidence: 99%

Predicting wind-driven spatial deposition through simulated color images using deep autoencoders

Fernández-Godino

Lucas

Kong

2023

Sci Rep

View full text Add to dashboard Cite

show abstract

“…In order to realize the angular super-resolution in scanning radar by using TSVD method, we first need to determine the truncation parameter in (9). The focus of this paper is to realize the angular super-resolution in scanning radar.…”

Section: Simulationsmentioning

confidence: 99%

“…Recently applications of TSVD method can be found in [8] for inverse scattering problem, [2] for improving the spatial resolution of radiometer data, and [9] for image restoration. But little work on TSVD method for improving the angular resolution in scanning radar has been reported.…”

Section: Introductionmentioning

confidence: 99%

A truncated singular value decomposition method for angular super-resolution in scanning radar

Huang

Zha

Yang

2015

2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS)

View full text Add to dashboard Cite

Angular super-resolution of scanning radar is an important problem in radar system. Some deconvolution methods are used to realize the angular super-resolution in scanning radar. However, the ill-posed nature of the deconvolution problem leads to the noise amplification in the angular super-resolution image. This phenomenon brings the difficulty in signal detection and tracking. In this paper, a deconvolution algorithm based on truncated singular value decomposition is proposed that achieves the angular super-resolution and noise suppression in scanning radar. To this end, we first convert the angular super-resolution task in scanning radar as an equivalent deconvolution problem. Then, the cause of noise amplification is analysed, which leads to the truncation singular value method for solving the deconvolution problem. Simulation results demonstrate that the proposed method is effective in achieving angular resolution with suppressing the noise amplification.

show abstract

“…Truncated singular value decomposition (TSVD) is a popular method for solving the deconvolution problem. Recently applications of TSVD method can be found in [ 35 ] for inverse scattering problem, [ 11 ] for improving the spatial resolution of radiometer data, and [ 36 ] for image restoration. But little work on TSVD method for improving the angular resolution in forward looking radar imaging has been reported.…”

Section: Introductionmentioning

confidence: 99%

Forward Looking Radar Imaging by Truncated Singular Value Decomposition and Its Application for Adverse Weather Aircraft Landing

Huang¹,

Zha²,

Wang³

et al. 2015

Sensors

View full text Add to dashboard Cite

The forward looking radar imaging task is a practical and challenging problem for adverse weather aircraft landing industry. Deconvolution method can realize the forward looking imaging but it often leads to the noise amplification in the radar image. In this paper, a forward looking radar imaging based on deconvolution method is presented for adverse weather aircraft landing. We first present the theoretical background of forward looking radar imaging task and its application for aircraft landing. Then, we convert the forward looking radar imaging task into a corresponding deconvolution problem, which is solved in the framework of algebraic theory using truncated singular decomposition method. The key issue regarding the selecting of the truncated parameter is addressed using generalized cross validation approach. Simulation and experimental results demonstrate that the proposed method is effective in achieving angular resolution enhancement with suppressing the noise amplification in forward looking radar imaging.

show abstract

Image restoration using truncated SVD filter bank based on an energy criterion

Cited by 7 publications

References 20 publications

Predicting wind-driven spatial deposition through simulated color images using deep autoencoders

Predicting wind-driven spatial deposition through simulated color images using deep autoencoders

A truncated singular value decomposition method for angular super-resolution in scanning radar

Forward Looking Radar Imaging by Truncated Singular Value Decomposition and Its Application for Adverse Weather Aircraft Landing

Contact Info

Product

Resources

About