Kumari Pooja

Bio: Kumari Pooja is an academic researcher from Indian Institute of Space Science and Technology. The author has contributed to research in topics: Ground truth & Hyperspectral imaging. The author has an hindex of 1, co-authored 1 publications receiving 10 citations.

Multi-Scale Dilated Residual Convolutional Neural Network for Hyperspectral Image Classification

Abstract: Recently, deep Convolutional Neural Networks (CNNs) have been extensively studied for hyperspectral image classification. It has undergone significant improvement as compared to conventional classification methods. Yet, there are not much studies have been taken on sub-sampled ground truth dataset in CNN. This paper exploits CNN-based method along with multi-scale and dilated convolution with residual connection concepts for hyperspectral image classification on exclusive real time data set. Two raw and one standard full ground truth Pavia University datasets are used to characterize the performance. Out of raw exclusive datasets, one was taken over urban areas of Ahmedabad, India under ISRO-NASA joint initiative for HYperSpectral Imaging (HYSI) programme, and the other was collected using Hypersec VNIR integrated camera of our institute surroundings from the rooftop of the building.

MC-PDNet: Deep Unrolled Neural Network For Multi-Contrast Mr Image Reconstruction From Undersampled K-Space Data

Abstract: Multi-contrast (MC) MR images are similar in structure and can leverage anatomical structure to perform joint reconstruction especially from a limited number of k-space data in the Compressed Sensing (CS) setting. However CS-based multi-contrast image reconstruction has shown limited performance in these highly accelerated regimes due to the use of hand-crafted group sparsity priors. Deep learning can improve outcomes by learning the joint prior across multiple weighting contrasts. In this work, we extend the primal-dual neural network (PDNet) in the multi-contrast sense. We propose a MC-PDNet architecture which takes full advantage of multi-contrast information. Using an in-house database consisting of images from T2TSE, T2*GRE and FLAIR contrasts acquired in 66 healthy volunteers, we performed a retrospective study from 4-fold under-sampled data. It was shown that MC-PDNet improves image quality by at least 1dB in PSNR for each contrast individually in comparison with PD-Net, U-Net and DISN-5B architectures.

3-D Channel and Spatial Attention Based Multiscale Spatial–Spectral Residual Network for Hyperspectral Image Classification

Abstract: With the rapid development of aerospace and various remote sensing platforms, the amount of data related to remote sensing is increasing rapidly. To meet the application requirements of remote sensing big data, an increasing number of scholars are combining deep learning with remote sensing data. In recent years, based on the rapid development of deep learning methods, research in the field of hyperspectral image (HSI) classification has seen continuous breakthroughs. In order to fully extract the characteristics of HSIs and improve the accuracy of image classification, this article proposes a novel three-dimensional (3-D) channel and spatial attention-based multiscale spatial–spectral residual network (termed CSMS-SSRN). The CSMS-SSRN framework uses a three-layer parallel residual network structure by using different 3-D convolutional kernels to continuously learn spectral and spatial features from their respective residual blocks. Then, the extracted depth multiscale features are stacked and input into the 3-D attention module to enhance the expressiveness of the image features from the two aspects of channel and spatial domains, thereby improving the accuracy of classification. The CSMS-SSRN framework proposed in this article can achieve better classification performance on different HSI datasets.

Hierarchical Shrinkage Multiscale Network for Hyperspectral Image Classification With Hierarchical Feature Fusion

Abstract: Recently, deep learning (DL)-based hyperspectral image classification (HSIC) has attracted substantial attention. Many works based on the convolutional neural network (CNN) model have been certificated to be significantly successful for boosting the performance of HSIC. However, most of these methods extract features by using a fixed convolutional kernel and ignore multiscale features of the ground objects of hyperspectral images (HSIs). Although some recent methods have proposed multiscale feature extraction schemes, more computing and storage resources were consumed. Moreover, when using CNN to implement HSI classification, many methods only use the high-level semantic information extracted from the end of the network, ignoring the edge information extracted from shallow layers of the network. To settle the preceding two issues, a novel HSIC method based on hierarchical shrinkage multiscale network and the hierarchical feature fusion is proposed, with which the newly proposed classification framework can fuse features generated by both of multiscale receptive field and multiple levels. Specifically, multidepth and multiscale residual block (MDMSRB) is constructed by superposition dilated convolution to realize multiscale feature extraction. Furthermore, according to the change of feature size in different stages of the neural networks, we design a hierarchical shrinkage multiscale feature extraction network by pruning MDMSRB to reduce the redundancy of network structure. In addition, to make full use of the features extracted in each stage of the network, the proposed network hierarchically integrates low-level edge features and high-level semantic features effectively. Experimental results demonstrate that the proposed method achieves more competitive performance with a limited computational cost than other state-of-the-art methods.

Dynamic Data Augmentation Method for Hyperspectral Image Classification Based on Siamese Structure

Abstract: At present, deep learning classification researches of hyperspectral usually focus on optimizing the classification model. In essence, most of them did not take special measures for the characteristics of the small sample and imbalanced category distribution of hyperspectral itself. Aiming at the problems of small samples and imbalanced category distribution, we propose a dynamic data selection algorithm. For one thing, this algorithm can dynamically select the samples that need data augmentation most. For another, it can be nested in Stochastic gradient descent (SGD) and can be easily implemented. Furthermore, there will be differences between the original and the transformed sample because of data augmentation transformation, which obstructs trained models' performance. Aiming at the difference between the augmented sample and the original sample, we define the similarity score and introduce the Siamese training structure to obtain the similarity score by which we reduce the difference through the SGD algorithm. Experiments show that the method proposed in this article improves the classification results of the backbone training model when using data augmentation for training.

Multiscanning Strategy-Based Recurrent Neural Network for Hyperspectral Image Classification

Abstract: Most methods based on the convolutional neural network show satisfying performance for hyperspectral image (HSI) classification. However, the spatial dependence among different pixels is not well learned by CNNs. A recurrent neural network (RNN) can effectively establish the dependence of nonadjacent pixels and ensure that each feature activation in its output is an activation at the specific location concerning the whole image, in contrast to the usual local context window in the CNNs. However, recent limited conversion schemes in RNN-based methods for HSI classification cannot fully capture the complete spatial dependence of an HSI patch. In this study, a novel multiscanning strategy with RNN is proposed to feature the sequential character of the HSI pixel and fully consider the spatial dependence in the HSI patch. By investigating different scanning forms, eight scanning orders are considered spatially, which flattens one local HSI patch into eight neighboring continuous pixel sequences. Moreover, considering that eight scanning orders complement one local patch with correlative dependence, the concatenated features from all scanning orders are fed into the RNN again for complementarity. As a result, the network can achieve competitive classification performance on three publicly accessible datasets using fewer parameters than other state-of-the-art methods.

A Joint Spatial-Spectral Representation Based Capsule Network for Hyperspectral Image Classification

Abstract: Recently, capsule networks have shown excellent performance in hyperspectral image (HSI) classification. However, when the HSI contains complex topographic features, the classification performance of capsule networks will deteriorate. To address this issue, we propose a novel joint spatial-spectral representation based capsule network (JSSR-CapsNet), which deeply exploits spatial-spectral information to improve classification performance. Specifically, JSSR-CapsNet utilizes an attention module to highlight the validity of sensitive pixels from redundant spatial-spectral features. Then, we built a dilated convolutional pyramid (DCP) to take aggregation of pixels with the same class into account, and suppress the interference of noisy pixels to enhance the intra-class consistency. The quantitative results conducted on two real HSIs show that JSSR-CapsNet improves the overall accuracy by 1.86% and 2.07% respectively over state-of-the-art capsule networks.