SWAN: Self-supervised Wavelet Neural Network for Hyperspectral Image Unmixing

In this article, we present SWAN: a three-stage, self-supervised wavelet neural network for joint estimation of endmembers and abundances from hyperspectral imagery. The contiguous and overlapping hyperspectral band images are first expanded to Biorthogonal wavelet basis space that provides sparse, distributed, and multi-scale representations. The idea is to exploit latent symmetries from thus obtained invariant and covariant features using a self-supervised learning paradigm. The first stage, SWANencoder maps the input wavelet coefficients to a compact lower-dimensional latent space. The second stage, SWANdecoder uses the derived latent representation to reconstruct the input wavelet coefficients. Interestingly, the third stage SWANforward learns the underlying physics of the hyperspectral image. A three-stage combined loss function is formulated in the image acquisition domain that eliminates the need for ground truth and enables self-supervised training. Adam is employed for optimizing the proposed loss function, while Sigmoid with a dropout of 0.3 is incorporated to avoid possible overfitting. Kernel regularizers bound the magnitudes and preserve spatial variations in the estimated endmember coefficients. The output of SWANencoder represents estimated abundance maps during inference, while weights of SWANdecoder are retrieved to extract endmembers. Experiments are conducted on two benchmark synthetic data sets with different signal-to-noise ratios as well as on three real benchmark hyperspectral data sets while comparing the results with several state-of-the-art neural network-based unmixing methods. The qualitative, quantitative, and ablation results show performance enhancement by learning a resilient unmixing function as well as promoting self-supervision and compact network parameters for practical applications.

A Strictly Bounded Deep Network for Unpaired Cyclic Translation of Medical Images

Medical image translation is an ill-posed problem. Unlike existing paired unbounded unidirectional translation networks, in this paper, we consider unpaired medical images and provide a strictly bounded network that yields a stable bidirectional translation. We propose a patch-level concatenated cyclic conditional generative adversarial network (pCCGAN) embedded with adaptive dictionary learning. It consists of two cyclically connected CGANs of 47 layers each; where both generators (each of 32 layers) are conditioned with concatenation of alternate unpaired patches from input and target modality images (not ground truth) of the same organ. The key idea is to exploit cross-neighborhood contextual feature information that bounds the translation space and boosts generalization. The generators are further equipped with adaptive dictionaries learned from the contextual patches to reduce possible degradation. Discriminators are 15-layer deep networks that employ minimax function to validate the translated imagery. A combined loss function is formulated with adversarial, non-adversarial, forward-backward cyclic, and identity losses that further minimize the variance of the proposed learning machine. Qualitative, quantitative, and ablation analysis show superior results on real CT and MRI.

An unsupervised deep learning framework for medical image denoising

Medical image acquisition is often intervented by unwanted noise that corrupts the information content. This paper introduces an unsupervised medical image denoising technique that learns noise characteristics from the available images and constructs denoised images. It comprises of two blocks of data processing, viz., patch-based dictionaries that indirectly learn the noise and residual learning (RL) that directly learns the noise. The model is generalized to account for both 2D and 3D images considering different medical imaging instruments. The images are considered one-by-one from the stack of MRI/CT images as well as the entire stack is considered, and decomposed into overlapping image/volume patches. These patches are given to the patch-based dictionary learning to learn noise characteristics via sparse representation while given to the RL part to directly learn the noise properties. K-singular value decomposition (K-SVD) algorithm for sparse representation is used for training patch-based dictionaries. On the other hand, residue in the patches is trained using the proposed deep residue network. Iterating on these two parts, an optimum noise characterization for each image/volume patch is captured and in turn it is subtracted from the available respective image/volume patch. The obtained denoised image/volume patches are finally assembled to a denoised image or 3D stack. We provide an analysis of the proposed approach with other approaches. Experiments on MRI/CT datasets are run on a GPU-based supercomputer and the comparative results show that the proposed algorithm preserves the critical information in the images as well as improves the visual quality of the images.

Towards glass-box CNNs

Convolution neural networks (CNNs) are brain-inspired architectures popular for their ability to train and relearn visually complex tasks. It is incremental and scalable; however, CNN is mostly treated as black-box and involves multiple trial & error runs. We observe that CNN constructs powerful internal representations that help achieve state-of-the-art performance. Here we propose three layer glass-box (analytical) CNN for two-class image classifcation problems. First is a representation layer that encompasses both the class information (group invariant) and symmetric transformations (group equivariant) of input images. It is then passed through dimension reduction layer (PCA). Finally the compact yet complete representation is provided to a classifer. Analytical machine learning classifers and multilayer perceptrons are used to assess sensitivity. Proposed glass-box CNN is compared with equivariance of AlexNet (CNN) internal representation for better understanding and dissemination of results. In future, we would like to construct glass-box CNN for multiclass visually complex tasks.

Automatic Data Registration of Geostationary Payloads for Meteorological Applications at ISRO

The launch of KALPANA-1 satellite in the year 2002 heralded the establishment of an indigenous operational payload for meteorological predictions. This was further enhanced in the year 2003 with the launching of INSAT-3A satellite. The software for generating products from the data of these two satellites was taken up subsequently in the year 2004 and the same was installed at the Indian Meteorological Department, New Delhi in January 2006. Registration has been one of the most fundamental operations to generate almost all the data products from the remotely sensed data. Registration is a challenging task due to inevitable radiometric and geometric distortions during the acquisition process. Besides the presence of clouds makes the problem more complicated. In this paper, we present an algorithm for multitemporal and multiband registration. In addition, India facing reference boundaries for the CCD data of INSAT-3A have also been generated. The complete implementation is made up of the following steps: 1) automatic identification of the ground control points (GCPs) in the sensed data, 2) finding the optimal transformation model based on the match-points, and 3) resampling the transformed imagery to the reference coordinates. The proposed algorithm is demonstrated using the real datasets from KALPANA-1 and INSAT-3A. Both KALAPANA-1 and INSAT-3A have recently been decommissioned due to lack of fuel, however, the experience gained from them have given rise to a series of meteorological satellites and associated software; like INSAT-3D series which give continuous weather forecasting for the country. This paper is not so much focused on the theory (widely available in the literature) but concentrates on the implementation of operational software.