KURUMA PURNIMA

@sietk.org

Assistant Professor, Dept. of CSE
Siddhardha Institute of Science and Technology, Puttur

KURUMA PURNIMA


                        Download Compact PDF  
https://researchid.co/jpmpworks

RESEARCH, TEACHING, or OTHER INTERESTS

Signal Processing, Multidisciplinary, Computer Engineering, Computer Science
14

Scopus Publications

71

Scholar Citations

5

Scholar h-index

1

Scholar i10-index

Scopus Publications

  • IDCP-NET: An Improved Dark Channel Prior Network with Multi-Constraint Transmission Refinement for Image Dehazing
    Poornima M, M. A. Manivasagam, E Murali, B Himabindu, D Janani, Kuruma Purnima
    Proceedings of 2nd International Conference on Multi Agent Systems for Collaborative Intelligence Icmsci 2026, 2026
    Image degradation caused by haze and atmospheric scattering poses a significant challenge to outdoor computer vision systems, leading to reduced contrast, color distortion, and diminished scene visibility. Traditional haze removal methods, particularly those based on the Dark Channel Prior (DCP), often suffer from limitations, including color artifacts in bright regions and incomplete haze suppression under dense conditions. To address these issues, this paper proposes IDCP-Net, an advanced dehazing framework that integrates an improved DCP model with multi-constraint transmission estimation and context-aware refinement strategies. The proposed method utilizes polarimetric imaging cues for accurate atmospheric light estimation, adaptive multiscale DCP computation, guided filtering for edge preservation, and variance-based contextual refinement to enhance the transmission map. Comprehensive experiments on standard benchmark datasets demonstrate that IDCP-Net consistently outperforms state-of-the-art techniques in PSNR, SSIM, and perceptual similarity while achieving faster inference times, offering a reliable and efficient solution for real-world image dehazing applications.
  • Devising a comprehensive synthetic underwater image dataset
    Kuruma Purnima, C.Siva Kumar
    Journal of Visual Communication and Image Representation, 2025
  • An IoT-based Smart Wearable Safety Device using GSM, GPS, and ESP32-CAM
    Kuruma Purnima, Balagala Madhavi, Dollu Sukith, Katari Rajesh, Poli Sandeep Kumar Reddy, Bandi Venkata Kiran Kumar Reddy
    Proceedings of 3rd International Conference on Sustainable Computing and Data Communication Systems Icscds 2025, 2025
    This paper presents a smart wearable device aimed at enhancing personal safety, particularly in emergencies. The proposed system is built around an Arduino Nano microcontroller. It integrates a GPS module for real-time location tracking, a GSM module for sending SMS alerts, an ESP32-CAM for capturing and emailing images, a buzzer for audible alerts, and an LCD for real-time feedback. Upon pressing a single button, the device performs a sequence of actions that include sending the user’s location and images to pre-configured emergency contacts, while simultaneously activating a loud buzzer to attract nearby attention. Experimental evaluations demonstrate a rapid response time of 5–7 seconds from activation to alert delivery, independent of internet connectivity. The results confirm the system's reliability and efficiency compared to traditional alarms and smartphone-based safety apps. This multi-modal alert system offers a portable, effective, and practical solution for improving emergency response and ensuring user safety.
  • A Lightweight Framework for Underwater Image Enhancement with Focus on Color Cast Correction
    Venkata Lakshmi Keerthi K, B. Malakona Reddy, Aravabhumi Divya, Lokesh Raju V, R Priyadarshini, Kuruma Purnima
    Proceedings 2025 5th International Conference on Expert Clouds and Applications Icoeca 2025, 2025
    Underwater imaging plays a vital role in marine biology, underwater archaeology, and environmental monitoring applications. However, the quality of underwater images is often compromised due to color casts caused by light absorption and scattering in the water medium. These challenges hinder image analysis, object recognition, and visual interpretation. This paper presents a computationally efficient framework for enhancing underwater images affected by color casts. The proposed method employs multiscale decomposition and fusion techniques to restore color balance and improve image quality while preserving computational simplicity. Unlike traditional methods that often suffer from over-saturation, incomplete dehazing, or high computational requirements, this approach achieves superior results with minimal artifacts. Experimental evaluations, based on qualitative and quantitative metrics such as UIQM, MDM, and UIConM, demonstrate the method’s effectiveness in comparison with state-of-the-art techniques. The findings highlight the method’s potential for real-time applications in diverse underwater scenarios.
  • Adaptive Non-Subsampled Shearlet-PCNN Framework for Efficient MRI-PET Image Fusion
    Kuruma Purnima, Pudi Vasavi, S M Tamil Selvi, Shaik Sumiya, Shaik Suhel Ahamad, Krishna Kumar
    Proceedings of the International Conference on Research in Computational Intelligence and Communication Networks Icrcicn, 2025
    Medical image fusion aims to integrate complementary information from different imaging modalities into a single, informative image for improved clinical diagnosis. Despite the growing adoption of deep learning and generative models, these methods often require large datasets and substantial computational resources. This paper presents an efficient and interpretable MRI-PET image fusion technique using a non-subsampled shearlet transform combined with an adaptive pulse-coupled neural network (PCNN). The transform effectively captures anisotropic and directional features. At the same time, the adaptive PCNN governs the fusion logic through energy-attribute weighting for low-pass subbands and firingtime control for high-pass subbands. The proposed method is evaluated on the Whole Brain Atlas MRI-PET dataset and compared with recent state-of-the-art fusion models, including dual-attention CNN, parameter adaptive unit-linking PCNN, and coupled-GAN frameworks. Simulation results show that the proposed scheme achieves enhanced visual quality and competitive quantitative performance in terms of entropy, standard deviation, and structural similarity, while maintaining low computational complexity. This approach offers a practical balance between interpretability, efficiency, and fusion accuracy, making it suitable for clinical and diagnostic applications.
  • Deep Belief Networks for Ovarian Tumour Detection From Gene Data
    Venkata Lakshmi Keerthi. K, J Venkatagiri, Kuruma Purnima, Aravabhumi Divya, T.V.V. Satyanarayana, Rentamallu Ramaiah
    2025 6th International Conference on Data Intelligence and Cognitive Informatics Icdici 2025, 2025
    Ovarian cancer is one of the most lethal gynecological malignancies, with early and accurate diagnosis playing a critical role in improving patient survival rates. Traditional diagnostic techniques, such as imaging and biomarker analysis, have limitations in terms of specificity and sensitivity. This study uses a Deep Belief Network (DBN) to classify ovarian tumours based on microarray gene expression data. The dataset consists of gene expression profiles with over 15,000 features, representing continuous values corresponding to molecular markers of cancerous and normal ovarian tissues. The proposed DBN model is trained to extract deep hierarchical features, enabling the robust classification of tumour samples. Performance evaluation uses standard metrics, including accuracy, precision, recall, and F1-score. Experimental results demonstrate that the DBN-based approach achieves superior classification accuracy compared to conventional machine learning models. This research highlights the potential of deep learning in analyzing high-dimensional biological data, offering a promising direction for computational oncology and precision medicine.
  • Attention-Enhanced Deep Learning Framework for Driver Drowsiness Detection in Intelligent Transportation Systems
    Kuruma Purnima, Chandragiri Guruprasad, Kotte Chaitanya, Kanikanti Naga Sai Rama Krishna, A Naresh, Dasamandam Harsha Vardhan
    Proceedings of 6th International Conference on Iot Based Control Networks and Intelligent Systems Icicnis 2025, 2025
    Driver drowsiness is a major cause of road accidents, endangering lives and reducing transportation efficiency. To address this issue, this paper proposes an attention-enhanced deep learning framework for real-time driver drowsiness detection within intelligent transportation systems (ITS). The system integrates convolutional neural networks (CNNs) for spatial feature extraction, Long Short-Term Memory (LSTM) units for modeling temporal behavior, and an attention mechanism to prioritize fatigue-related cues such as eyelid closure, yawning, and gaze deviation. Attention maps highlight critical regions, improving interpretability and reducing false detections. The framework is optimized for low latency and can operate on embedded platforms for in-vehicle deployment. Experimental evaluations on benchmark and custom-acquired datasets demonstrate an accuracy exceeding 95%, with superior robustness under varying illumination and head pose conditions. The proposed approach provides a scalable, interpretable, and efficient solution for enhancing road safety and advancing real-time driver monitoring in modern ITS environments.
  • AI-based Helmet Detection and Alert System using YOLOv5 and Cloud Deployment
    Kuruma Purnima, Yepuru Siddamma, Ranganath Gari Rajesh, S K Premchandh, Thupakula Naresh, Jaya Krishna Sunkara
    Proceedings of 6th International Conference on Iot Based Control Networks and Intelligent Systems Icicnis 2025, 2025
    Road safety remains a critical global challenge, with a significant number of motorcycle-related fatalities caused by riders’ non-compliance with helmet regulations. This paper presents an AI-based helmet detection and alert system that leverages the YOLOv5 deep learning model and is integrated with cloud-based deployment for real-time monitoring and automated violation alerts. The proposed system utilizes image and video input from roadside or surveillance cameras to identify riders with and without helmets, employing optimized convolutional neural networks for rapid and accurate detection. A custom dataset was trained using Roboflow integration, and the YOLOv5 weights were fine-tuned to enhance model precision and recall. Experimental results demonstrate a mean average precision (mAP) of 94% and a recall of 90%, achieving high detection accuracy even under complex lighting and background conditions. Furthermore, the system’s cloud integration enables centralized data management, remote access, and real-time alert generation, offering scalability for smart city traffic enforcement. Comparative analysis with existing YOLO-based helmet detection models validates the effectiveness of the proposed framework in terms of accuracy, computational efficiency, and deployability in real-world traffic environments.
  • CSUID – Comprehensive synthetic underwater image dataset
    Kuruma Purnima, C. Siva Kumar
    Data in Brief, 2024
    The underwater environment is characterized by complex light traversal, encompassing effects such as color loss, contrast loss, water distortion, backscatter, light attenuation, and color cast, which vary depending on water purity, depth, and other factors. The dataset presented in this paper is prepared with 100 ground-truth images and 1,50,000 synthetic underwater images. This dataset approximates the effects of underwater environment with implementable combinations of color cast, blurring, low-light, and contrast reduction. These effects and their combinations, with different severity levels are applied to each ground-truth image to generate as many as 150 synthetic underwater images. In addition to the dataset of 1,50,100 images, a comprehensive set of 21 focus metrics, including the average contrast measure operator, Brenner's gradient-based metric, and many others, are calculated for each image.
  • Enhancement of Low-Light Images using Structure-Aware Illumination Mapping: A LIME Approach
    Kuruma Purnima, V. V. Satyanarayana Tallapragada, B Devi, M Sai Kumar, K Pavithra, T Greeshma Rao
    2024 15th International Conference on Computing Communication and Networking Technologies Icccnt 2024, 2024
    Low-light image enhancement is a fundamental task in computer vision with numerous real-world applications. This paper presents a novel approach to low-light image enhancement based on the estimation of the illumination map using a Retinex-based method. Unlike traditional Retinex-based approaches, which decompose an image into reflectance and illumination components, our method focuses solely on estimating the illumination, thereby reducing computational complexity. The proposed algorithm begins by constructing an initial illumination map based on the maximum intensity of each pixel across RGB channels. Subsequently, a structure-aware smoothing technique is applied to refine the illumination map, improving its accuracy. Two algorithms are developed to solve the refinement problem: an Augmented Lagrangian Multiplier (ALM) based exact solver and a sped-up solver for reduced computational load. Experimental results demonstrate the efficacy of the proposed method, showcasing significant improvements in low-light image quality compared to state-of-the-art methods. This approach offers a promising solution for enhancing low-light images in various scenarios, with potential applications in surveillance, photography, and medical imaging.
  • Efficient Non-Local Similarity-based Image Dehazing: A Pixel-Level Approach for Enhanced Performance and Robustness
    A. Swetha Rani, K Venkata Lakshmi Keerthi, M. V. Nageswara Rao, G.V. Pradeep Kumar, V. V. Satyanarayana Tallapragada, Kuruma Purnima
    8th International Conference on Electronics Communication and Aerospace Technology Iceca 2024 Proceedings, 2024
  • Detection of Parkinson's Disease using Machine Learning with Feature Analysis from Audio Signals
    Kuruma Purnima, M.A. Manivasagam, Dudekula Mohammed Kaif, M Vanitha, Nagella Rajesh, B Silambarasan, Shaik Suhail, G Ravi Kumar, V Hanumantha Rao
    4th IEEE International Conference on Mobile Networks and Wireless Communications Icmnwc 2024, 2024
  • Non-Gradient Based Design Metrics for Underwater Image Enhancement
    Kuruma Purnima, C. Siva Kumar
    International Conference on Self Sustainable Artificial Intelligence Systems Icssas 2023 Proceedings, 2023
  • Gradient-Based Design Metrics for Assessment of Underwater Image Enhancement
    Kuruma Purnima, C. Siva Kumar
    International Conference on Self Sustainable Artificial Intelligence Systems Icssas 2023 Proceedings, 2023

RECENT SCHOLAR PUBLICATIONS

  • IDCP-NET: An Improved Dark Channel Prior Network with Multi-Constraint Transmission Refinement for Image Dehazing
    M Poornima, MA Manivasagam, E Murali, B Himabindu, D Janani, ...
    2026 Second International Conference on Multi-Agent Systems for … , 2026
    2026
  • Adaptive Non-Subsampled Shearlet–PCNN Framework for Efficient MRI–PET Image Fusion
    K Purnima, P Vasavi, SMT Selvi, S Sumiya, SS Ahamad, K Kumar
    2025 Seventh International Conference on Research in Computational … , 2025
    2025
  • Attention-Enhanced Deep Learning Framework for Driver Drowsiness Detection in Intelligent Transportation Systems
    K Purnima, C Guruprasad, K Chaitanya, KNSR Krishna, A Naresh, ...
    2025 6th International Conference on IoT Based Control Networks and … , 2025
    2025
  • AI-based Helmet Detection and Alert System using YOLOv5 and Cloud Deployment
    K Purnima, Y Siddamma, RG Rajesh, SK Premchandh, T Naresh, ...
    2025 6th International Conference on IoT Based Control Networks and … , 2025
    2025
  • An IoT-based Smart Wearable Safety Device using GSM, GPS, and ESP32-CAM
    K Purnima, B Madhavi, D Sukith, K Rajesh, PSK Reddy, BVKK Reddy
    2025 3rd International Conference on Sustainable Computing and Data … , 2025
    2025
    Citations: 1
  • Deep Belief Networks for Ovarian Tumour Detection From Gene Data
    J Venkatagiri, K Purnima, A Divya, TVV Satyanarayana, R Ramaiah
    2025 6th International Conference on Data Intelligence and Cognitive … , 2025
    2025
  • A Lightweight Framework for Underwater Image Enhancement with Focus on Color Cast Correction
    BM Reddy, A Divya, R Priyadarshini, K Purnima
    2025 5th International Conference on Expert Clouds and Applications (ICOECA … , 2025
    2025
  • Devising a comprehensive synthetic underwater image dataset
    K Purnima, CS Kumar
    Journal of Visual Communication and Image Representation 107, 104386 , 2025
    2025
    Citations: 6
  • Detection of Parkinson's Disease using Machine Learning with Feature Analysis from Audio Signals
    K Purnima, MA Manivasagam, DM Kaif, M Vanitha, N Rajesh, ...
    2024 4th International Conference on Mobile Networks and Wireless … , 2024
    2024
    Citations: 5
  • Efficient non-local similarity-based image dehazing: A pixel-level approach for enhanced performance and robustness
    AS Rani, KVL Keerthi, MVN Rao, GVP Kumar, VVS Tallapragada, ...
    2024 8th International Conference on Electronics, Communication and … , 2024
    2024
    Citations: 3
  • CSUID–Comprehensive synthetic underwater image dataset
    K Purnima, CS Kumar
    Data in Brief 55, 110723 , 2024
    2024
    Citations: 7
  • Enhancing Library Resource Recommendations Using Collaborative Filtering Algorithms.
    BM Reddy, MA Manivasagam, K Purnima, A Divya, K Subramanyam
    Library of Progress-Library Science, Information Technology & Computer 44 (3) , 2024
    2024
  • Enhancement of Low-Light Images using Structure-Aware Illumination Mapping: A LIME Approach
    K Purnima, VVS Tallapragada, B Devi, MS Kumar, K Pavithra, TG Rao
    2024 15th International Conference on Computing Communication and Networking … , 2024
    2024
    Citations: 5
  • A comprehensive synthetic underwater image dataset
    K Purnima, CS Kumar
    Mendeley Data 3 , 2024
    2024
    Citations: 2
  • Non-gradient based design metrics for underwater image enhancement
    K Purnima, CS Kumar
    2023 International Conference on Self Sustainable Artificial Intelligence … , 2023
    2023
    Citations: 9
  • Gradient-based design metrics for assessment of underwater image enhancement
    K Purnima, CS Kumar
    2023 International Conference on Self Sustainable Artificial Intelligence … , 2023
    2023
    Citations: 14
  • Exploring the Potential of Invasive Weed Optimization: A Population-Based Metaheuristic for Optimization Problems
    K Sudha, M Suresh, RM Mallika, A Divya, K Purnima, E Sasikala
    Harbin Gongcheng Daxue Xuebao/Journal of Harbin Engineering University 44 (7 … , 2023
    2023
  • Underwater Image Enhancement Techniques–A Comprehensive Review
    K Purnima, CS Kumar
    Journal of Harbin Engineering University 44 (8), 159-173 , 2023
    2023
    Citations: 3
  • Optical Character Recognition with Geometric Attacks
    P Kuruma, ES Reddy
    Journal of Emerging Technologies and Innovative Research 7 (11), 547-554 , 2020
    2020
  • An Exploratory Study in Educating High School Children towards Maintaining an Eco-Friendly Environment
    K Purnima
    Strength for Today and Bright Hope for Tomorrow Volume 15: 3 March 2015 ISSN … , 2015
    2015
    Citations: 1

MOST CITED SCHOLAR PUBLICATIONS

  • Gradient-based design metrics for assessment of underwater image enhancement
    K Purnima, CS Kumar
    2023 International Conference on Self Sustainable Artificial Intelligence … , 2023
    2023.0
    Citations: 14
  • Non-gradient based design metrics for underwater image enhancement
    K Purnima, CS Kumar
    2023 International Conference on Self Sustainable Artificial Intelligence … , 2023
    2023.0
    Citations: 9
  • Super-resolution based image reconstruction
    JK Sunkara, K Purnima, S Muchakala, Y Ravisankariah
    International Journal of Computer Science and Technology 2 (3), 272-281 , 2011
    2011.0
    Citations: 8
  • CSUID–Comprehensive synthetic underwater image dataset
    K Purnima, CS Kumar
    Data in Brief 55, 110723 , 2024
    2024.0
    Citations: 7
  • Devising a comprehensive synthetic underwater image dataset
    K Purnima, CS Kumar
    Journal of Visual Communication and Image Representation 107, 104386 , 2025
    2025.0
    Citations: 6
  • Detection of Parkinson's Disease using Machine Learning with Feature Analysis from Audio Signals
    K Purnima, MA Manivasagam, DM Kaif, M Vanitha, N Rajesh, ...
    2024 4th International Conference on Mobile Networks and Wireless … , 2024
    2024.0
    Citations: 5
  • Enhancement of Low-Light Images using Structure-Aware Illumination Mapping: A LIME Approach
    K Purnima, VVS Tallapragada, B Devi, MS Kumar, K Pavithra, TG Rao
    2024 15th International Conference on Computing Communication and Networking … , 2024
    2024.0
    Citations: 5
  • A new accordion based video compression method
    JK Sunkara, K Purnima, EN Sagari, LR Subbareddy
    i-Manager's Journal on Electronics Engineering 1 (4), 14 , 2011
    2011.0
    Citations: 4
  • Efficient non-local similarity-based image dehazing: A pixel-level approach for enhanced performance and robustness
    AS Rani, KVL Keerthi, MVN Rao, GVP Kumar, VVS Tallapragada, ...
    2024 8th International Conference on Electronics, Communication and … , 2024
    2024.0
    Citations: 3
  • Underwater Image Enhancement Techniques–A Comprehensive Review
    K Purnima, CS Kumar
    Journal of Harbin Engineering University 44 (8), 159-173 , 2023
    2023.0
    Citations: 3
  • Ravisankariah Y,“
    JK Sunkara, K Purnima, S Muchakala
    Super-Resolution Based Image Reconstruction, 272-281 , 0
    Citations: 3
  • A comprehensive synthetic underwater image dataset
    K Purnima, CS Kumar
    Mendeley Data 3 , 2024
    2024.0
    Citations: 2
  • An IoT-based Smart Wearable Safety Device using GSM, GPS, and ESP32-CAM
    K Purnima, B Madhavi, D Sukith, K Rajesh, PSK Reddy, BVKK Reddy
    2025 3rd International Conference on Sustainable Computing and Data … , 2025
    2025.0
    Citations: 1
  • An Exploratory Study in Educating High School Children towards Maintaining an Eco-Friendly Environment
    K Purnima
    Strength for Today and Bright Hope for Tomorrow Volume 15: 3 March 2015 ISSN … , 2015
    2015.0
    Citations: 1
  • IDCP-NET: An Improved Dark Channel Prior Network with Multi-Constraint Transmission Refinement for Image Dehazing
    M Poornima, MA Manivasagam, E Murali, B Himabindu, D Janani, ...
    2026 Second International Conference on Multi-Agent Systems for … , 2026
    2026.0
  • Adaptive Non-Subsampled Shearlet–PCNN Framework for Efficient MRI–PET Image Fusion
    K Purnima, P Vasavi, SMT Selvi, S Sumiya, SS Ahamad, K Kumar
    2025 Seventh International Conference on Research in Computational … , 2025
    2025.0
  • Attention-Enhanced Deep Learning Framework for Driver Drowsiness Detection in Intelligent Transportation Systems
    K Purnima, C Guruprasad, K Chaitanya, KNSR Krishna, A Naresh, ...
    2025 6th International Conference on IoT Based Control Networks and … , 2025
    2025.0
  • AI-based Helmet Detection and Alert System using YOLOv5 and Cloud Deployment
    K Purnima, Y Siddamma, RG Rajesh, SK Premchandh, T Naresh, ...
    2025 6th International Conference on IoT Based Control Networks and … , 2025
    2025.0
  • Deep Belief Networks for Ovarian Tumour Detection From Gene Data
    J Venkatagiri, K Purnima, A Divya, TVV Satyanarayana, R Ramaiah
    2025 6th International Conference on Data Intelligence and Cognitive … , 2025
    2025.0
  • A Lightweight Framework for Underwater Image Enhancement with Focus on Color Cast Correction
    BM Reddy, A Divya, R Priyadarshini, K Purnima
    2025 5th International Conference on Expert Clouds and Applications (ICOECA … , 2025
    2025.0