Sharad Ghimire - Academia.edu (original) (raw)

Papers by Sharad Ghimire

ICT with Intelligent Applications, 2021

Journal of the Institute of Engineering, 2019

Automatically extracting most conspicuous object from an image is useful and important for many ... more Automatically extracting most conspicuous object from an image is useful and important for many computer vision related tasks. Performance of several applications such as object segmentation, image classification based on salient object and content based image editing in computer vision can be improved using this technique. In this research work, performance of structured matrix decomposition with contour based spatial prior is analyzed for extracting salient object from the complex scene. To separate background and salient object, structured matrix decomposition model based on low rank matrix recovery theory is used along with two structural regularizations. Tree structured sparsity inducing regularization is used to capture image structure and to enforce the same object to assign similar saliency values. And, Laplacian regularization is used to enlarge the gap between background part and salient object part. In addition to structured matrix decomposition model, general high level...

The difficulty in machine learning-based image super-resolution is to generate high-frequency com... more The difficulty in machine learning-based image super-resolution is to generate high-frequency component in an image without introducing any artifacts. In this paper, Devnagari handwritten characters image using a generative adversarial network with a classifier is generated in high-resolution which is also classifiable. The generator architecture is modified by removing all batch normalization layers in generator architecture with a residual in residual dense block. Batch normalization is removed because it produces unwanted artifacts in the generated images. A Devnagari handwritten characters classifier is built using CNN. The classifier is used in the network to calculate the content loss. The adversarial loss is obtained from the GAN architecture and both of the losses are added to obtain total loss. Generated HR images is validated using six different evaluation metrics among which MSE, PSNR determines pixel-wise difference and SSIM compares images perceptually. Similarly, FID i...

Salient object detection is a useful and important technique in computer vision, which can scale ... more Salient object detection is a useful and important technique in computer vision, which can scale up the efficiency of several applications such as object detection, object segmentation and content based image editing. In this research work, an improved technique of salient object extraction using structured matrix decomposition and contour based spatial prior is implemented. In order to assist efficient background-salient objects separation, structured matrix decomposition model with two structural regularizations namely, tree-structured sparsity–inducing regularization and Laplacian regularization are used. Tree-structured sparsity-inducing regularization captures the image structure and enforces the same object to have similar saliency values and Laplacian regularization enlarges the gap between background and the salient object. In addition, integrating general high level priors and contour based spatial prior obtained from biologically inspired model is implemented to improve th...

In order to fulfill 4G requirements, deploying low power nodes (LPN) within a macro cell layout, ... more In order to fulfill 4G requirements, deploying low power nodes (LPN) within a macro cell layout, also known as heterogeneous networks (HetNets), is a promising solution to enhance overall system performance, performance at cell-edges and indoor coverage. LTE/LTE-Advanced system uses Orthogonal Frequency Division Multiple Access (OFDMA) method to cater services to multiple users simultaneously. Due to the orthogonal character of these sub-carriers, the interference between sub-carriers is eliminated within a cell but reuse of these sub-carriers in adjacent cells creates intercell interference. HetNets deployment includes new technical challenges related to increased interference level and throughput degradation; in particular, cell edge users of low power cell are vulnerable to strong interference from high power macro cells. Inter-cell interference coordination/enhanced intercell interference coordination (ICIC/eICIC) schemes in frequency, time and power domain can be used to overco...

Indoor localization and mapping is an important problem with many applications such as emergency ... more Indoor localization and mapping is an important problem with many applications such as emergency response, architectural modeling, and historical preservation. In this project, a metrically accurate, GPS-denied, indoor 3D static mapping system was developed using a moveable base coupled with three degree of freedom IMU. The experiment was performed by scanning a flat apartment which closely resembled the actual test environment. First the distance of various elements in the surrounding was collected using the LiDAR module along with the orientation of LiDAR Lite. This is termed as scanning mode of the system. After full scan of an area, the system was manually switched to Odometry mode wherein the system was moved to another location for new scan. The distance travelled as well as the direction of travel was noted. Secondly, the data is processed to obtain the position of different points in 3D space. The vast data set is stored in Point Cloud format. The point cloud is processed an...

Nowadays interest has grown in using Mobile Relay Station (MRS) system to provide cellular covera... more Nowadays interest has grown in using Mobile Relay Station (MRS) system to provide cellular coverage to the onboard users in public transport, particularly in High Speed Trains (HSTs) due to high penetration rate of portable electronic devices such as smart phones, tablets and laptops. In the railway scenarios, when the signal propagates into the train, it suffers through the high order of Vehicle Penetration Loss (VPL) due to modern construction materials and techniques used for the train. Beside that rapid temporal variations in the radio channel and Inter-Carrier Interference (ICI) issues due to Doppler shift effects produce a quick battery drain or increased call drop rates indeed. Therefore users inside train will suffer worse channel propagation conditions, similar to celledge users. Thus the coverage and capacity at railway scenario remain relatively small due to low Signal-to-Noise Ratio (SNR). The deployment of MRS system to transmit data between the Donor eNB (DeNB) and the...

ICT with Intelligent Applications, 2021

Journal of the Institute of Engineering, 2019

Automatically extracting most conspicuous object from an image is useful and important for many ... more Automatically extracting most conspicuous object from an image is useful and important for many computer vision related tasks. Performance of several applications such as object segmentation, image classification based on salient object and content based image editing in computer vision can be improved using this technique. In this research work, performance of structured matrix decomposition with contour based spatial prior is analyzed for extracting salient object from the complex scene. To separate background and salient object, structured matrix decomposition model based on low rank matrix recovery theory is used along with two structural regularizations. Tree structured sparsity inducing regularization is used to capture image structure and to enforce the same object to assign similar saliency values. And, Laplacian regularization is used to enlarge the gap between background part and salient object part. In addition to structured matrix decomposition model, general high level...

The difficulty in machine learning-based image super-resolution is to generate high-frequency com... more The difficulty in machine learning-based image super-resolution is to generate high-frequency component in an image without introducing any artifacts. In this paper, Devnagari handwritten characters image using a generative adversarial network with a classifier is generated in high-resolution which is also classifiable. The generator architecture is modified by removing all batch normalization layers in generator architecture with a residual in residual dense block. Batch normalization is removed because it produces unwanted artifacts in the generated images. A Devnagari handwritten characters classifier is built using CNN. The classifier is used in the network to calculate the content loss. The adversarial loss is obtained from the GAN architecture and both of the losses are added to obtain total loss. Generated HR images is validated using six different evaluation metrics among which MSE, PSNR determines pixel-wise difference and SSIM compares images perceptually. Similarly, FID i...

Salient object detection is a useful and important technique in computer vision, which can scale ... more Salient object detection is a useful and important technique in computer vision, which can scale up the efficiency of several applications such as object detection, object segmentation and content based image editing. In this research work, an improved technique of salient object extraction using structured matrix decomposition and contour based spatial prior is implemented. In order to assist efficient background-salient objects separation, structured matrix decomposition model with two structural regularizations namely, tree-structured sparsity–inducing regularization and Laplacian regularization are used. Tree-structured sparsity-inducing regularization captures the image structure and enforces the same object to have similar saliency values and Laplacian regularization enlarges the gap between background and the salient object. In addition, integrating general high level priors and contour based spatial prior obtained from biologically inspired model is implemented to improve th...

In order to fulfill 4G requirements, deploying low power nodes (LPN) within a macro cell layout, ... more In order to fulfill 4G requirements, deploying low power nodes (LPN) within a macro cell layout, also known as heterogeneous networks (HetNets), is a promising solution to enhance overall system performance, performance at cell-edges and indoor coverage. LTE/LTE-Advanced system uses Orthogonal Frequency Division Multiple Access (OFDMA) method to cater services to multiple users simultaneously. Due to the orthogonal character of these sub-carriers, the interference between sub-carriers is eliminated within a cell but reuse of these sub-carriers in adjacent cells creates intercell interference. HetNets deployment includes new technical challenges related to increased interference level and throughput degradation; in particular, cell edge users of low power cell are vulnerable to strong interference from high power macro cells. Inter-cell interference coordination/enhanced intercell interference coordination (ICIC/eICIC) schemes in frequency, time and power domain can be used to overco...

Indoor localization and mapping is an important problem with many applications such as emergency ... more Indoor localization and mapping is an important problem with many applications such as emergency response, architectural modeling, and historical preservation. In this project, a metrically accurate, GPS-denied, indoor 3D static mapping system was developed using a moveable base coupled with three degree of freedom IMU. The experiment was performed by scanning a flat apartment which closely resembled the actual test environment. First the distance of various elements in the surrounding was collected using the LiDAR module along with the orientation of LiDAR Lite. This is termed as scanning mode of the system. After full scan of an area, the system was manually switched to Odometry mode wherein the system was moved to another location for new scan. The distance travelled as well as the direction of travel was noted. Secondly, the data is processed to obtain the position of different points in 3D space. The vast data set is stored in Point Cloud format. The point cloud is processed an...

Nowadays interest has grown in using Mobile Relay Station (MRS) system to provide cellular covera... more Nowadays interest has grown in using Mobile Relay Station (MRS) system to provide cellular coverage to the onboard users in public transport, particularly in High Speed Trains (HSTs) due to high penetration rate of portable electronic devices such as smart phones, tablets and laptops. In the railway scenarios, when the signal propagates into the train, it suffers through the high order of Vehicle Penetration Loss (VPL) due to modern construction materials and techniques used for the train. Beside that rapid temporal variations in the radio channel and Inter-Carrier Interference (ICI) issues due to Doppler shift effects produce a quick battery drain or increased call drop rates indeed. Therefore users inside train will suffer worse channel propagation conditions, similar to celledge users. Thus the coverage and capacity at railway scenario remain relatively small due to low Signal-to-Noise Ratio (SNR). The deployment of MRS system to transmit data between the Donor eNB (DeNB) and the...