Lord Amoah
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Lord Amoah has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
Ampofo Twumasi Kwabena
1 shared publication- Di Yang
University of Florida, Department 4008, 1000 E. University Avenue, Gainesville, FL, United States [US
1 shared publication Goma Tshivetta Christian Fersein Jorvialom
Engineering Research Center of Digital Forensics, Ministry of Education, Nanjing University of Information Science & Technology, Nanjing, 210044, China
1 shared publication- Guangjie Liu
Nanjing University of Information Science and Technology
1 shared publication Guangyong Gao
1 shared publication- Guo Cao
Nanjing University of Science and Technology
1 shared publication
Latest Publications
- 15 January 2025
- Published by Elsevier in Computers & Security
- Vol. 151
Investigating the impact of feature selection on adversarial transferability in intrusion detection system
- 15 January 2025
- Published by Elsevier in Computers & Security
- Vol. 151
Investigating the impact of feature selection on adversarial transferability in intrusion detection system
- Open Access
- 28 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 17 (19391404) , 3091-3107
A Metareinforcement-Learning-Based Hyperspectral Image Classification With a Small Sample Set
The fine spectral information contained in hyperspectral images (HSI) is combined with rich spatial features to provide feature qualities that serve as distinguishing variables for efficient classification performance. The task's objective is to correctly identify and categorize several object categories in the HSI, such as the ground, flora, water, and buildings, based on their spectral characteristics beneficial for a variety of applications, including mapping minerals, analyzing vegetation, and mapping urban land-use. The difficulty of learning new task-specific knowledge from a limited data sample that encourages less training has not been overcome by deep learning models. The capacity of current models to generalize to new tasks on small data sets is still lacking. By learning features that are transferable to facilitate adaptation to novel tasks on small samples, meta-reinforcement learning shows promise in overcoming such difficulties. We proposed a meta-reinforcement learning (Meta-RL) model that decouples task inference to improve meta-training, and accelerate meta-learning with small HSI labeled samples for efficient classification. The model employs a Capsule network for effective cooperation between spectra-spatial bands. To minimize the temporal difference error, the Apex-X Deep Q network parameter update is used to meta-train our model. The proposed model obtains an overall accuracy between 95.85% and 96.78% with computational time between 3207.9s and 7487.9s for training and validation as well as between 21.57s and 32.98s for testing. The experimental results prove the competitiveness of the proposed model to existing traditional deep learning, meta-learning, and reinforcement learning methods in both classification accuracy and computational cost.
- 28 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 17 (19391404) , 3091-3107
Open AccessA Metareinforcement-Learning-Based Hyperspectral Image Classification With a Small Sample Set
The fine spectral information contained in hyperspectral images (HSI) is combined with rich spatial features to provide feature qualities that serve as distinguishing variables for efficient classification performance. The task's objective is to correctly identify and categorize several object categories in the HSI, such as the ground, flora, water, and buildings, based on their spectral characteristics beneficial for a variety of applications, including mapping minerals, analyzing vegetation, and mapping urban land-use. The difficulty of learning new task-specific knowledge from a limited data sample that encourages less training has not been overcome by deep learning models. The capacity of current models to generalize to new tasks on small data sets is still lacking. By learning features that are transferable to facilitate adaptation to novel tasks on small samples, meta-reinforcement learning shows promise in overcoming such difficulties. We proposed a meta-reinforcement learning (Meta-RL) model that decouples task inference to improve meta-training, and accelerate meta-learning with small HSI labeled samples for efficient classification. The model employs a Capsule network for effective cooperation between spectra-spatial bands. To minimize the temporal difference error, the Apex-X Deep Q network parameter update is used to meta-train our model. The proposed model obtains an overall accuracy between 95.85% and 96.78% with computational time between 3207.9s and 7487.9s for training and validation as well as between 21.57s and 32.98s for testing. The experimental results prove the competitiveness of the proposed model to existing traditional deep learning, meta-learning, and reinforcement learning methods in both classification accuracy and computational cost.
- 11 June 2022
- Published by Elsevier in Journal of Information Security and Applications
- Vol. 68
Automatic contrast enhancement with reversible data hiding using bi-histogram shifting
- 11 June 2022
- Published by Elsevier in Journal of Information Security and Applications
- Vol. 68
Automatic contrast enhancement with reversible data hiding using bi-histogram shifting
- Open Access
- 1 January 2022
- Published by Tech Science Press in Journal of Quantum Computing
- Vol. 4 (3) , 183-197
Reversible Data Hiding with Contrast Enhancement Using Bi-histogram Shifting and Image Adjustment for Color Images
Prior versions of reversible data hiding with contrast enhancement (RDHCE) algorithms strongly focused on enhancing the contrast of grayscale images. However, RDHCE has recently witnessed a rise in contrast enhancement algorithms concentrating on color images. This paper implies a method for color images that uses the RGB (red, green, and blue) color model and is based on bi-histogram shifting and image adjustment. Bi-histogram shifting is used to embed data and image adjustment to achieve contrast enhancement by adjusting the images resulting from each channel of the color images before combining them to generate the final enhanced image. Images are first divided into three channels—R, G, and B—and the Max, Med, and Min channels are then determined from these. Before histogram shifting, some calculations are done to determine how many iterations there will be for each channel. The images are adjusted to improve visual quality in the enhanced images after data has been embedded in each channel. The experimental results show that the enhanced images produced by the proposed method are qualitatively and aesthetically superior to those produced by some earlier methods, and their quality was assessed using PSNR, SSIM, RCE, RMBE, and CIEDE2000. The embedding rate obtained by the suggested method is acceptable.
- 1 January 2022
- Published by Tech Science Press in Journal of Quantum Computing
- Vol. 4 (3) , 183-197
Open AccessReversible Data Hiding with Contrast Enhancement Using Bi-histogram Shifting and Image Adjustment for Color Images
Prior versions of reversible data hiding with contrast enhancement (RDHCE) algorithms strongly focused on enhancing the contrast of grayscale images. However, RDHCE has recently witnessed a rise in contrast enhancement algorithms concentrating on color images. This paper implies a method for color images that uses the RGB (red, green, and blue) color model and is based on bi-histogram shifting and image adjustment. Bi-histogram shifting is used to embed data and image adjustment to achieve contrast enhancement by adjusting the images resulting from each channel of the color images before combining them to generate the final enhanced image. Images are first divided into three channels—R, G, and B—and the Max, Med, and Min channels are then determined from these. Before histogram shifting, some calculations are done to determine how many iterations there will be for each channel. The images are adjusted to improve visual quality in the enhanced images after data has been embedded in each channel. The experimental results show that the enhanced images produced by the proposed method are qualitatively and aesthetically superior to those produced by some earlier methods, and their quality was assessed using PSNR, SSIM, RCE, RMBE, and CIEDE2000. The embedding rate obtained by the suggested method is acceptable.
- Open Access
- 1 January 2021
- Published by Tech Science Press in Journal of Quantum Computing
- Vol. 3 (2) , 65-77
A Bi-Histogram Shifting Contrast Enhancement for Color Images
Recent contrast enhancement (CE) methods, with a few exceptions, predominantly focus on enhancing gray-scale images. This paper proposes a bihistogram shifting contrast enhancement for color images based on the RGB (red, green, and blue) color model. The proposed method selects the two highest bins and two lowest bins from the image histogram, performs an equalized number of bidirectional histogram shifting repetitions on each RGB channel while embedding secret data into marked images. The proposed method simultaneously performs both right histogram shifting (RHS) and left histogram shifting (LHS) in each histogram shifting repetition to embed and split the highest bins while combining the lowest bins with their neighbors to achieve histogram equalization (HE). The least maximum number of histograms shifting repetitions among the three RGB channels is used as the default number of histograms shifting repetitions performed to enhance original images. Compared to an existing contrast enhancement method for color images and evaluated with PSNR, SSIM, RCE, and RMBE quality assessment metrics, the experimental results show that the proposed method's enhanced images are visually and qualitatively superior with a more evenly distributed histogram. The proposed method achieves higher embedding capacities and embedding rates in all images, with an average increase in embedding capacity of 52.1%.
- 1 January 2021
- Published by Tech Science Press in Journal of Quantum Computing
- Vol. 3 (2) , 65-77
Open AccessA Bi-Histogram Shifting Contrast Enhancement for Color Images
Recent contrast enhancement (CE) methods, with a few exceptions, predominantly focus on enhancing gray-scale images. This paper proposes a bihistogram shifting contrast enhancement for color images based on the RGB (red, green, and blue) color model. The proposed method selects the two highest bins and two lowest bins from the image histogram, performs an equalized number of bidirectional histogram shifting repetitions on each RGB channel while embedding secret data into marked images. The proposed method simultaneously performs both right histogram shifting (RHS) and left histogram shifting (LHS) in each histogram shifting repetition to embed and split the highest bins while combining the lowest bins with their neighbors to achieve histogram equalization (HE). The least maximum number of histograms shifting repetitions among the three RGB channels is used as the default number of histograms shifting repetitions performed to enhance original images. Compared to an existing contrast enhancement method for color images and evaluated with PSNR, SSIM, RCE, and RMBE quality assessment metrics, the experimental results show that the proposed method's enhanced images are visually and qualitatively superior with a more evenly distributed histogram. The proposed method achieves higher embedding capacities and embedding rates in all images, with an average increase in embedding capacity of 52.1%.