DeepBinaryMask: Learning a binary mask for video compressive sensing

Iliadis, Michael; Spinoulas, Leonidas; Katsaggelos, Aggelos K.

doi:10.1016/j.dsp.2019.102591

Cited by 63 publications

(46 citation statements)

References 34 publications

(61 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our approach is motivated by recent advances in hardware, autodifferentiation tools, and optimization algorithms for deep learning. While some recent work investigates joint optimization of either binary masks or color filter arrays with neural network post-processing for video compressed sensing or demosaicking [Chakrabarti 2016;Iliadis et al 2016], they do not consider the optimization of phase-modulating optical elements such as lenses and do not consider diffraction in their forward model. With this work, we thus take first steps towards utilizing the full potential of end-to-end optimization for computational camera design.…”

Section: Related Workmentioning

confidence: 99%

End-to-end optimization of optics and image processing for achromatic extended depth of field and super-resolution imaging

et al. 2018

View full text Add to dashboard Cite

Fig. 1. One of the applications of the proposed end-to-end computational camera design paradigm is achromatic extended depth of field. When capturing an image with a regular singlet lens (top left), out-of-focus regions are blurry and chromatic aberrations further degrade the image quality. With our framework, we optimize the profile of a refractive optical element that achieves both depth and chromatic invariance. This element is fabricated using diamond turning (right) or using photolithography. After processing an image recorded with this optical element using a simple Wiener deconvolution, we obtain an all-in-focus image with little chromatic aberrations (top center). Point spread functions for both the regular lens and the optimized optical element are shown in the bottom. In this paper, we explore several applications that demonstrate the efficacy of our novel approach to domain-specific computational camera design.

show abstract

Section: Related Workmentioning

confidence: 99%

End-to-end optimization of optics and image processing for achromatic extended depth of field and super-resolution imaging

et al. 2018

View full text Add to dashboard Cite

show abstract

“…For compressive video recovery, Iliadis etal. [23], [24] propose Deep Fully-Connected Network, where the encoder learns binary sensing mask and the decoder determines the reconstruction of the video. These approaches effectively avoid the expensive computation in traditional approaches and have achieved promising image/video reconstruction performance.…”

Section: Related Workmentioning

confidence: 99%

DR2-Net: Deep Residual Reconstruction Network for image compressive sensing

et al. 2019

View full text Add to dashboard Cite

Most traditional algorithms for compressive sensing image reconstruction suffer from the intensive computation. Recently, deep learning-based reconstruction algorithms have been reported, which dramatically reduce the time complexity than iterative reconstruction algorithms. In this paper, we propose a novel Deep Residual Reconstruction Network (DR 2 -Net) to reconstruct the image from its Compressively Sensed (CS) measurement. The DR 2 -Net is proposed based on two observations: 1) linear mapping could reconstruct a high-quality preliminary image, and 2) residual learning could further improve the reconstruction quality. Accordingly, DR 2 -Net consists of two components, i.e., linear mapping network and residual network, respectively. Specifically, the fully-connected layer in neural network implements the linear mapping network. We then expand the linear mapping network to DR 2 -Net by adding several residual learning blocks to enhance the preliminary image. Extensive experiments demonstrate that the DR 2 -Net outperforms traditional iterative methods and recent deep learning-based methods by large margins at measurement rates 0.01, 0.04, 0.1, and 0.25, respectively. The code of DR 2 -Net has been released on: https://github.com/coldrainyht/caffe dr2 33 33 33 33 33 33 64 32 1 Block CS Measurements Linear Mapping Residual Learning Block

show abstract

“…It permits the removal of small regions that are disjoint forms the larger objects without distorting the small features within the large objects. Neural networks have been proposed and a variety of works in binary reconstruction of grayscale images [45,46]. This technique is mainly used in X-ray images and MRI images and helps in the detection of cancer.…”

Section: Binary Reconstruction Of Gray Scale and Nonlinear Processingmentioning

confidence: 99%

Solving Image Processing Critical Problems Using Machine Learning

Sharma

Gupta

Jaiswal

2021

Studies in Big Data

View full text Add to dashboard Cite

The history of vision can go back around 543 BC. From fossil studies, researchers found out that no. of animal species floats out and soon develops the first animal with eyes. As evolution in bigger species around fifty percent of neurons in our body are involved in the visual processing. Vision allows human to perceive, comprehend, and understand surrounding them while computer vision aims to duplicate the impact of human vision electronically in the form of images. The history of making artificial vison starts around 1545, Gemma Frisius, and Encylopedie in eighteenth century. The foundation period of the concept lies from 1943 to 1956, during this period research is cantered towards understanding the concept of machine learning intelligence, artificial neural system, and automata theory computation. The first work in the field of Artificial intelligence in computer vision and image processing was done in MIT called "Summer Vision Project". The rise of artificial intelligence starts and it is used widely in every field of technology. As the technology advances machine learning is used in almost every field. This book chapter provides you some details how the machine learning is used in image processing and to solve the current problem and challenge in computer vison like: Denoising, Image compression, Image color enhancement, segmentation, etc. Machine learning algorithm like Neural Network, Support Vector Machine, Genetic Algorithm, Convolutional Neural Network, etc. were discussed in the book chapter.

show abstract

DeepBinaryMask: Learning a binary mask for video compressive sensing

Cited by 63 publications

References 34 publications

End-to-end optimization of optics and image processing for achromatic extended depth of field and super-resolution imaging

End-to-end optimization of optics and image processing for achromatic extended depth of field and super-resolution imaging

DR2-Net: Deep Residual Reconstruction Network for image compressive sensing

Solving Image Processing Critical Problems Using Machine Learning

Contact Info

Product

Resources

About