Ghulam Farooque - Academia.edu (original) (raw)

Uploads

Papers by Ghulam Farooque

Research paper thumbnail of DHCAE: Deep Hybrid Convolutional Autoencoder Approach for Robust Supervised Hyperspectral Unmixing

Remote Sensing

Hyperspectral unmixing (HSU) is a crucial method to determine the fractional abundance of the mat... more Hyperspectral unmixing (HSU) is a crucial method to determine the fractional abundance of the material (endmembers) in each pixel. Most spectral unmixing methods are affected by low signal-to-noise ratios because of noisy pixels and bands simultaneously, requiring robust HSU techniques that exploit both 3D (spectral–spatial dimension) and 2D (spatial dimension) domains. In this paper, we present a new method for robust supervised HSU based on a deep hybrid (3D and 2D) convolutional autoencoder (DHCAE) network. Most HSU methods adopt the 2D model for simplicity, whereas the performance of HSU depends on spectral and spatial information. The DHCAE network exploits spectral and spatial information of the remote sensing images for abundance map estimation. In addition, DHCAE uses dropout to regularize the network for smooth learning and to avoid overfitting. Quantitative and qualitative results confirm that our proposed DHCAE network achieved better hyperspectral unmixing performance on...

Research paper thumbnail of Crack Detection and Images Inpainting Method for Thai Mural Painting Images

2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC), 2018

Thailand frescoes are an important art heritage in the world. However, the erosion of history has... more Thailand frescoes are an important art heritage in the world. However, the erosion of history has resulted in the color loss, stain and scratches of many mural paintings. How to repair the Thailand murals has become an urgent problem. It is an important scientific problem to use computer image inpainting technology to simulate and eliminate the missing pixels in the murals and obtain beautiful and intact murals. In this paper, a computer aided semi-automatic repair framework is proposed by combining a scratch detection procedure and a model optimization based inpainting procedure. To this end, we propose a scratch semi-automatic detection method. In this method, a small number of seed points are given by users, and the location of scratches is then computed by region growing method and morphological operation. After that, the pixel filling and color restoration in the missing region can be obtained by using different variational inpainting methods. The experiment shows that the proposed method is effective.

Research paper thumbnail of Hyperspectral Image Classification via a Novel Spectral–Spatial 3D ConvLSTM-CNN

Remote Sensing, 2021

In recent years, deep learning-based models have produced encouraging results for hyperspectral i... more In recent years, deep learning-based models have produced encouraging results for hyperspectral image (HSI) classification. Specifically, Convolutional Long Short-Term Memory (ConvLSTM) has shown good performance for learning valuable features and modeling long-term dependencies in spectral data. However, it is less effective for learning spatial features, which is an integral part of hyperspectral images. Alternatively, convolutional neural networks (CNNs) can learn spatial features, but they possess limitations in handling long-term dependencies due to the local feature extraction in these networks. Considering these factors, this paper proposes an end-to-end Spectral-Spatial 3D ConvLSTM-CNN based Residual Network (SSCRN), which combines 3D ConvLSTM and 3D CNN for handling both spectral and spatial information, respectively. The contribution of the proposed network is twofold. Firstly, it addresses the long-term dependencies of spectral dimension using 3D ConvLSTM to capture the i...

Research paper thumbnail of Coin Recognition with Reduced Feature Set SIFT Algorithm Using Neural Network

2016 International Conference on Frontiers of Information Technology (FIT), 2016

Coin recognition is one of the prime important activities for modern banking and currency process... more Coin recognition is one of the prime important activities for modern banking and currency processing systems. These systems are widely used for coin sorting, automatic counting, and vending machines. The technique at the heart of such systems is object recognition in a digital image. Object classification and recognition is still one of the challenging research areas because we put our cognitive capabilities in a computer system through an algorithm. The reliability of such systems mainly depends on feature selection and extraction mechanism. This paper presents a novel approach for coins recognition. The proposed method uses Scale Invariant Feature Transform (SIFT) algorithm to handle the issues of rotations, scaling and illumination in a digital image. This is followed by Principle Component Analysis (PCA) for reducing extracted features set. This reduced feature set is passed to feed forward back-propagation artificial neural network (ANN) for classification and recognition. The experimental results indicate that proposed approach achieves state-of-the-art results for Pakistani coin recognition.

Research paper thumbnail of DHCAE: Deep Hybrid Convolutional Autoencoder Approach for Robust Supervised Hyperspectral Unmixing

Remote Sensing

Hyperspectral unmixing (HSU) is a crucial method to determine the fractional abundance of the mat... more Hyperspectral unmixing (HSU) is a crucial method to determine the fractional abundance of the material (endmembers) in each pixel. Most spectral unmixing methods are affected by low signal-to-noise ratios because of noisy pixels and bands simultaneously, requiring robust HSU techniques that exploit both 3D (spectral–spatial dimension) and 2D (spatial dimension) domains. In this paper, we present a new method for robust supervised HSU based on a deep hybrid (3D and 2D) convolutional autoencoder (DHCAE) network. Most HSU methods adopt the 2D model for simplicity, whereas the performance of HSU depends on spectral and spatial information. The DHCAE network exploits spectral and spatial information of the remote sensing images for abundance map estimation. In addition, DHCAE uses dropout to regularize the network for smooth learning and to avoid overfitting. Quantitative and qualitative results confirm that our proposed DHCAE network achieved better hyperspectral unmixing performance on...

Research paper thumbnail of Crack Detection and Images Inpainting Method for Thai Mural Painting Images

2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC), 2018

Thailand frescoes are an important art heritage in the world. However, the erosion of history has... more Thailand frescoes are an important art heritage in the world. However, the erosion of history has resulted in the color loss, stain and scratches of many mural paintings. How to repair the Thailand murals has become an urgent problem. It is an important scientific problem to use computer image inpainting technology to simulate and eliminate the missing pixels in the murals and obtain beautiful and intact murals. In this paper, a computer aided semi-automatic repair framework is proposed by combining a scratch detection procedure and a model optimization based inpainting procedure. To this end, we propose a scratch semi-automatic detection method. In this method, a small number of seed points are given by users, and the location of scratches is then computed by region growing method and morphological operation. After that, the pixel filling and color restoration in the missing region can be obtained by using different variational inpainting methods. The experiment shows that the proposed method is effective.

Research paper thumbnail of Hyperspectral Image Classification via a Novel Spectral–Spatial 3D ConvLSTM-CNN

Remote Sensing, 2021

In recent years, deep learning-based models have produced encouraging results for hyperspectral i... more In recent years, deep learning-based models have produced encouraging results for hyperspectral image (HSI) classification. Specifically, Convolutional Long Short-Term Memory (ConvLSTM) has shown good performance for learning valuable features and modeling long-term dependencies in spectral data. However, it is less effective for learning spatial features, which is an integral part of hyperspectral images. Alternatively, convolutional neural networks (CNNs) can learn spatial features, but they possess limitations in handling long-term dependencies due to the local feature extraction in these networks. Considering these factors, this paper proposes an end-to-end Spectral-Spatial 3D ConvLSTM-CNN based Residual Network (SSCRN), which combines 3D ConvLSTM and 3D CNN for handling both spectral and spatial information, respectively. The contribution of the proposed network is twofold. Firstly, it addresses the long-term dependencies of spectral dimension using 3D ConvLSTM to capture the i...

Research paper thumbnail of Coin Recognition with Reduced Feature Set SIFT Algorithm Using Neural Network

2016 International Conference on Frontiers of Information Technology (FIT), 2016

Coin recognition is one of the prime important activities for modern banking and currency process... more Coin recognition is one of the prime important activities for modern banking and currency processing systems. These systems are widely used for coin sorting, automatic counting, and vending machines. The technique at the heart of such systems is object recognition in a digital image. Object classification and recognition is still one of the challenging research areas because we put our cognitive capabilities in a computer system through an algorithm. The reliability of such systems mainly depends on feature selection and extraction mechanism. This paper presents a novel approach for coins recognition. The proposed method uses Scale Invariant Feature Transform (SIFT) algorithm to handle the issues of rotations, scaling and illumination in a digital image. This is followed by Principle Component Analysis (PCA) for reducing extracted features set. This reduced feature set is passed to feed forward back-propagation artificial neural network (ANN) for classification and recognition. The experimental results indicate that proposed approach achieves state-of-the-art results for Pakistani coin recognition.