Multi-Dataset Hyper-CNN for Hyperspectral Image Segmentation of Remote Sensing Images

Liu, Li; Awwad, Emad Mahrous; Ali, Yasser A.; Al-Razgan, Muna; Maarouf, Ali; Abualigah, Laith; Hoshyar, Azadeh Noori

doi:10.3390/pr11020435

Cited by 5 publications

(2 citation statements)

References 48 publications

(67 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Fast random forest (FRF) algorithm is employed to train a supervised machine learning model to identify and classify melanoma defects in skin lesion images The FRF algorithm was used to highlight anomalous areas of the skin that follow the Allen-Spitz criteria for melanoma. The results show that the FRF algorithm achieved a level of precision that could make it a convenient tool to aid dermatopathologists in the challenging analysis of malignant melanoma [21].…”

Section: Related Workmentioning

confidence: 96%

Genetically Optimized Neural Network for Early Detection of Glaucoma and Cardiovascular Disease Risk Prediction

Muthu,

Devadoss

2023

View full text Add to dashboard Cite

Glaucoma, a major ocular disease, often culminates in irreversible blindness if undiagnosed. Characterized by optic nerve fiber degeneration, it instigates structural changes to the optic nerve, thus precipitating vision loss. Despite being symptomless, early detection is imperative to prevent further visual impairment. The disease's inception is attributed to increased intraocular pressure, a condition influenced by blood pressure. Notably, the eye and heart share parallel characteristics, making glaucoma an early indicator of potential cardiac conditions. Increased blood pressure frequently accompanies Diabetes mellitus -a common complication exacerbating cardiac health and fostering the development of cardiovascular diseases. Leveraging computational technologies allows for the early-stage identification of glaucoma. The utilization of deep learning approaches and pruning techniques has yielded significant outcomes in detecting glaucoma-related abnormalities accurately. Pruning, a strategy implemented to eliminate redundant parameters while preserving optimal performance, is particularly beneficial. This study introduces a Genetically Optimized Neural Network (GONN) incorporating wavelet transformation for the detection of glaucoma, thereby assisting in diabetes and heart disease risk identification. Experimental results demonstrate that the GONN method outperforms conventional methods such as Artificial Neural Networks (ANN), Naï ve Bayes, multilayer perceptron, ensemble methods, K-Nearest Neighbor (KNN), and decision trees. Notably, the GONN technique achieves an impressive accuracy of 95%, an F1 score of 92%, and an Area Under the Curve (AUC) of 98.92%. This study's findings underscore the potential of the GONN technique in accurately identifying glaucoma, thereby aiding in early diabetes and heart disease risk prediction. The results demonstrate that the GONN approach is a viable tool for clinical practice, with potential implications for improved patient outcomes and healthcare efficiency.

show abstract

Section: Related Workmentioning

confidence: 96%

Genetically Optimized Neural Network for Early Detection of Glaucoma and Cardiovascular Disease Risk Prediction

Muthu,

Devadoss

2023

View full text Add to dashboard Cite

show abstract

“…There are various HSI segmentation methods presented in the literature [11] for instance deep learning approaches are widely used, support vector machines, Markov random fields etc. Some methods reduce the dimensionality and extract features by principal component analysis (PCA) or linear discriminant analysis (LDA).…”

Section: Introductionmentioning

confidence: 99%

Hyperspectral Image Segmentation and Entropy Calculation for Qualitative Analysis of Grassland

Oprisescu

Coliban

Ivanovici

2023

2023 International Symposium on Signals, Circuits and Systems (ISSCS)

View full text Add to dashboard Cite

Hyperspectral imaging provides high spectral resolution images of a scene in hundreds of narrow spectral bands. This technique proves to be very useful in Earth Observation applications such as land cover mapping, agriculture crop health assessment and other remote sensing tasks. The qualitative analysis of grassland areas is manually performed by computing the Shannon-Weaver biodiversity index. In this paper we propose a semi-automatic estimation of this biodiversity index in remotelysensed hyperspectral images. Starting from a spectral reflectance curve chosen to be representative for grassland, we perform the image segmentation based on histogram thresholding of spectral angle mapper (SAM) values. We then compute the entropy for the pixels belonging to the segmented grassland areas. In order to apply the classic Shannon definition of entropy, we perform a clustering for data dimensionality reduction. The evaluation of the proposed method is performed considering a manually generated ground truth.

show abstract

Effective segmentation of land-use and land-cover from hyperspectral remote sensing image

Yele,

Alegavi,

Sedamkar

2024

Int. j. inf. tecnol.

View full text Add to dashboard Cite

Multi-Dataset Hyper-CNN for Hyperspectral Image Segmentation of Remote Sensing Images

Cited by 5 publications

References 48 publications

Genetically Optimized Neural Network for Early Detection of Glaucoma and Cardiovascular Disease Risk Prediction

Genetically Optimized Neural Network for Early Detection of Glaucoma and Cardiovascular Disease Risk Prediction

Hyperspectral Image Segmentation and Entropy Calculation for Qualitative Analysis of Grassland

Effective segmentation of land-use and land-cover from hyperspectral remote sensing image

Contact Info

Product

Resources

About