Thogarachetti Hitendra Sarma
Research Subjects
Research Interests
Thogarachetti Hitendra Sarma has a total of 22 co-authors. Here are the top co-authors by number of shared publications:
- Pulabaigari Viswanath
IIIT-SriCity,Andhra Pradesh,India
7 shared publications - Boddi Eswara Reddy
JNTUA College of Engineering
4 shared publications - Neeraj Kumar Nadipelli
Indian Institute of Space Science and Technology
3 shared publications K. Mrudula
Department of Mathematics, GNITS, Hyderabad, India
2 shared publications- Murali Kanthi
Department of Data Science, CMR Technical Campus, Hyderabad, India
2 shared publications R. Dharma Reddy
Vasavi College of Engineering
2 shared publications
Latest Publications
- 10 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 22 (1545598X) , 1-5
Efficient Hyperspectral Band Selection Using GA-SR-NMI-VI: A Hybrid Similarity and Evolutionary-Based Approach
Hyperspectral image (HSI) analysis requires effective band selection techniques to enhance classification accuracy while maintaining computational efficiency. Similarity-based ranking normalized mutual information and variation of information (SR-NMI-VI) is a recent SR approach that leverages NMI and VI to compute band rankings. This letter presents an extended version of SR-NMI-VI called genetic algorithm (GA)-SR-NMI-VI, an advanced feature selection approach that integrates GA with similarity-based ranking. The proposed GA-SR-NMI-VI is a two-step approach, where SR-NMI-VI is first used for band ranking, followed by a GA-based optimization to determine the optimal number of bands. For comparative evaluation, two additional methods are considered: GA-SR-structural similarity index (SSIM)-NMI-VI, which adds SSIM to enhance spatial–spectral ranking, and graph-regularized fast and robust principal component analysis (GR-FRPCA), an unsupervised low-rank clustering-based approach. To validate the proposed approaches, an experimental study has been conducted on various HSI datasets covering a diverse range of classes from 2 to 16 and covering different classification scenarios, including Oil Spill, Cubert Drone, WHU-Hi-LongKou, and WHU-Hi-HanChuan. Empirically, it is shown that the proposed GA-SR-NMI-VI and its SSIM-enhanced variant consistently achieve higher accuracy and kappa scores across various machine learning models. GA-SR-NMI-VI achieves a 70%–80% reduction in the number of bands while improving classification accuracy by 2%–5%. Notably, traditional classifiers such as random forest and support vector machine (SVM) perform comparably to deep learning models while benefiting from lower computational costs, highlighting the effectiveness of GA-based methods in scenarios where deep learning may be computationally expensive or infeasible. The details of the experimental setup and the reproducible code are available at the following link: https://github.com/neerajkumarnadipelli/GA-SR-NMI-VI
- 10 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 22 (1545598X) , 1-5
Efficient Hyperspectral Band Selection Using GA-SR-NMI-VI: A Hybrid Similarity and Evolutionary-Based Approach
Hyperspectral image (HSI) analysis requires effective band selection techniques to enhance classification accuracy while maintaining computational efficiency. Similarity-based ranking normalized mutual information and variation of information (SR-NMI-VI) is a recent SR approach that leverages NMI and VI to compute band rankings. This letter presents an extended version of SR-NMI-VI called genetic algorithm (GA)-SR-NMI-VI, an advanced feature selection approach that integrates GA with similarity-based ranking. The proposed GA-SR-NMI-VI is a two-step approach, where SR-NMI-VI is first used for band ranking, followed by a GA-based optimization to determine the optimal number of bands. For comparative evaluation, two additional methods are considered: GA-SR-structural similarity index (SSIM)-NMI-VI, which adds SSIM to enhance spatial–spectral ranking, and graph-regularized fast and robust principal component analysis (GR-FRPCA), an unsupervised low-rank clustering-based approach. To validate the proposed approaches, an experimental study has been conducted on various HSI datasets covering a diverse range of classes from 2 to 16 and covering different classification scenarios, including Oil Spill, Cubert Drone, WHU-Hi-LongKou, and WHU-Hi-HanChuan. Empirically, it is shown that the proposed GA-SR-NMI-VI and its SSIM-enhanced variant consistently achieve higher accuracy and kappa scores across various machine learning models. GA-SR-NMI-VI achieves a 70%–80% reduction in the number of bands while improving classification accuracy by 2%–5%. Notably, traditional classifiers such as random forest and support vector machine (SVM) perform comparably to deep learning models while benefiting from lower computational costs, highlighting the effectiveness of GA-based methods in scenarios where deep learning may be computationally expensive or infeasible. The details of the experimental setup and the reproducible code are available at the following link: https://github.com/neerajkumarnadipelli/GA-SR-NMI-VI
- Open Access
- 1 November 2024
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 36 (2) , 1309-1318
Optimizing hyperspectral classification: spectral similarity-based band selection with chord k-means
Band selection is crucial for achieving high classification accuracy in hyperspectral image (HSI) analysis, especially when ground truth data are limited. While unsupervised algorithms are preferred in such scenarios, the effectiveness of k-means clustering depends heavily on the choice of similarity measure. This article presents a novel two-level clustering approach for band selection. In the first level, bands are clustered using k-means with various similarity measures such as Euclidean distance, spectral angle mapper (SAM), and spectral information divergence (SID). Subsequently, the second level leverages the chord metric k-means clustering to form clusters of HSI scenes upon optimal band clusters from the first level. This initial band selection reduces dimensionality and guides subsequent k-means clustering. The proposed chord-based clustering method, utilizing the chord metric, outperforms standard k-means variants, demonstrating significant improvements in accuracy. Experimental results on publicly available hyperspectral datasets confirm the effectiveness of the proposed approach as an alternative to traditional k-means algorithms, showcasing significant improvements in accuracy.
- 1 November 2024
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 36 (2) , 1309-1318
Open AccessOptimizing hyperspectral classification: spectral similarity-based band selection with chord k-means
Band selection is crucial for achieving high classification accuracy in hyperspectral image (HSI) analysis, especially when ground truth data are limited. While unsupervised algorithms are preferred in such scenarios, the effectiveness of k-means clustering depends heavily on the choice of similarity measure. This article presents a novel two-level clustering approach for band selection. In the first level, bands are clustered using k-means with various similarity measures such as Euclidean distance, spectral angle mapper (SAM), and spectral information divergence (SID). Subsequently, the second level leverages the chord metric k-means clustering to form clusters of HSI scenes upon optimal band clusters from the first level. This initial band selection reduces dimensionality and guides subsequent k-means clustering. The proposed chord-based clustering method, utilizing the chord metric, outperforms standard k-means variants, demonstrating significant improvements in accuracy. Experimental results on publicly available hyperspectral datasets confirm the effectiveness of the proposed approach as an alternative to traditional k-means algorithms, showcasing significant improvements in accuracy.
- 6 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 12 (1) , 190-194
International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing 2023: A Brief Report [Conference Reports]
Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.
- 6 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 12 (1) , 190-194
International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing 2023: A Brief Report [Conference Reports]
Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.
- Open Access
- 20 December 2021
- Published by ENGG Journals Publications in Indian Journal of Computer Science and Engineering
- Vol. 12 (6) , 1559-1567
HYBRID LEARNING APPROACH FOR FEATURE EXTRACTION AND CLASSIFICATION IN HYPERSPECTRAL IMAGES
- 20 December 2021
- Published by ENGG Journals Publications in Indian Journal of Computer Science and Engineering
- Vol. 12 (6) , 1559-1567
Open AccessHYBRID LEARNING APPROACH FOR FEATURE EXTRACTION AND CLASSIFICATION IN HYPERSPECTRAL IMAGES
- 23 November 2021
- Published by Springer Nature
- p. 773-782
A New CNN for Pixel Classification in Hyperspectral Images
- 23 November 2021
- Published by Springer Nature
- p. 773-782
A New CNN for Pixel Classification in Hyperspectral Images
- 30 April 2020
- Published by Springer Nature
- p. 101-107
Dimensionality Reduction in Hyperspectral Images Using Auto-encoders
- 30 April 2020
- Published by Springer Nature
- p. 101-107
Dimensionality Reduction in Hyperspectral Images Using Auto-encoders
- 1 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
A Prototype based Hybrid Approach to speed-up Kernel FCM-K
Kernel versions of FCM have been proved to be better than FCM in identifying overlapping and linearly inseparable clusters in the input space. Kernel FCM-F (KFCM-F) and Kernel FCM-K (KFCM-K) are the two kernel versions of FCM. In KFCM-F the cluster centers are considered in the feature space, where as in KFCM-K the cluster centers are identified in the kernel space. KFCM-K is superior than KFCM-F w.r.t the clustering quality, but it is not applicable on large data sets because of its quadratic time complexity i.e., $O(n^{2})$ where $n$ is the size of the data set. This paper propose a new prototype based hybrid technique to speed-up KFCM-K for large data sets. The proposed method initially identifies some representative data items from the given data, say $l$ where $l < < n$ , in linear time. The conventional kernel FCM-K is then applied over these representatives. As $l < < n$ the clustering time is reduced to $O(n+l^{2})$ from $O(n^{2})$ . Experimental study on several benchmark data sets shows that the proposed method converges in less time when compare to conventional KFCM-K, but with a negligible deviation in the clustering quality.- 1 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
A Prototype based Hybrid Approach to speed-up Kernel FCM-K
Kernel versions of FCM have been proved to be better than FCM in identifying overlapping and linearly inseparable clusters in the input space. Kernel FCM-F (KFCM-F) and Kernel FCM-K (KFCM-K) are the two kernel versions of FCM. In KFCM-F the cluster centers are considered in the feature space, where as in KFCM-K the cluster centers are identified in the kernel space. KFCM-K is superior than KFCM-F w.r.t the clustering quality, but it is not applicable on large data sets because of its quadratic time complexity i.e., $O(n^{2})$ where $n$ is the size of the data set. This paper propose a new prototype based hybrid technique to speed-up KFCM-K for large data sets. The proposed method initially identifies some representative data items from the given data, say $l$ where $l < < n$ , in linear time. The conventional kernel FCM-K is then applied over these representatives. As $l < < n$ the clustering time is reduced to $O(n+l^{2})$ from $O(n^{2})$ . Experimental study on several benchmark data sets shows that the proposed method converges in less time when compare to conventional KFCM-K, but with a negligible deviation in the clustering quality.- 25 September 2019
- Published by Springer Nature
- p. 125-133
Intelligent Computing for Air Pollution Monitoring Using GIS, Remote Sensing and Machine Learning
- 25 September 2019
- Published by Springer Nature
- p. 125-133
Intelligent Computing for Air Pollution Monitoring Using GIS, Remote Sensing and Machine Learning