Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Gandharba Swain has a total of 61 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • A moment-preserving transform-domain robust medical image watermarking with extraction-side learning

    by Ranjan K. Senapati, G. Madhu, Sandeep Kautish, Gandharba Swain, Ali Wagdy Mohamed, Abdulaziz S. Almazyad

  • Open Access

    Federated proximal learning with data augmentation for brain tumor classification under heterogeneous data distributions

    by Swetha Ghanta, Venkata Sumanth Siddareddy, Prasanthi Boyapati, Sujit Biswas, Gandharba Swain, Ashok K. Pradhan

    The increasing use of electronic health records (EHRs) has transformed healthcare management, yet data sharing across institutions remains limited due to privacy concerns. Federated learning (FL) offers a privacy-preserving solution by enabling collaborative model training without centralized data sharing. However, non-independent and identically distributed (non-IID) data distributions, where the data across clients differ in class proportions and feature characteristics, pose a major challenge to achieving robust model performance. In this study, we propose a hybrid framework that combines the Federated Proximal (FedProx) algorithm with the ResNet50 architecture to address non-IID data issues. We artificially partitioned an IID brain tumor dataset into non-IID subsets to simulate real-world conditions and applied data augmentation techniques to balance class distributions. Global model performance is monitored across 100 training rounds with varying regularization parameters in FedProx. The proposed framework achieved an accuracy of 97.71% on IID data and 87.19% in extreme non-IID scenarios, with precision, recall, and F1-scores also demonstrating strong performance. These findings highlight the effectiveness of combining data augmentation with FedProx in mitigating data imbalance in FL, thereby supporting equitable and efficient training of privacy-preserving models for healthcare applications.


  • Open Access

    An AI-Based Method for Automated Breast Cancer Detection and Localization in Mammogram and Ultrasound Images

    by Pradeep Kumar, Ranjan K. Senapati, Prasanth Mankar, Santosh Kumar Choudhary, P. M. K. Prasad, Satish Muppidi, Gandharba Swain

    This research aims to improve the identification, classification, and segmentation of breast cancer using mammography and ultrasound images using a Refined Mask-RCNN framework. An OR operation merges the results, reducing misclassification and unnecessary biopsies. The system is deployed in the NVIDIA Jetson Nano developer kit to assist doctors and radiologists in the detection of cancerous cells. Compared to the original Mask-RCNN, the proposed method performs better in cancer identification and segmentation, showing improved metrics such as accuracy, precision, True Positive Rate (TPR), True Negative Rate (TNR), F-score, Balanced Classification Rate (BCR), Youden's index, Jaccard, and Dice coefficient, demonstrating the robustness and reliability of the combined approach in the clinical diagnosis of breast cancer.


  • Energy‐Aware Mobile Data Collection in Wireless Sensor Networks Using Heuristic Clustering and Path Optimization Techniques

    by Naween Kumar, Shailendra Pratap Singh, Gandharba Swain, Anand Nayyar

    Wireless sensor networks (WSNs) face critical energy constraints that directly affect network longevity, especially when scaled to large deployments. This paper uses a multi‐stage process to introduce an energy‐efficient data collection framework using mobile data collectors (MDCs). We employ a modified weighted set cover approach to identify a minimal set of strategically placed anchor points (APs), significantly reducing the number of required MDCs. After refining them with an optimized fuzzy C‐means clustering and a traffic load balancing technique, these APs ensure coverage for all areas and balance the load among each other. Using Euler graph analysis, the routing paths are determined through the minimum spanning tree and perfect matching approaches. Experimental results demonstrate that our suggested method reduces the amount of MDCs by 28%, the distance traveled by tours by 35%, the average delay in data collection by 10%, and increases data transmission efficiency by 16%. The reported results prove that our suggested method helps gather more data energy efficiently and prolong the lifetime of WSN nodes.


  • Open Access

    Image Watermarking for Tamper Correction Based on Hamming Error Correcting Code

    by Pinapati Lakshmana Rao, Reshma Sonar, Nanditha Boddu, Anita Pradhan, Ranjan K. Senapati, Gandharba Swain

    Watermarking techniques are designed to embed a watermark in a manner that enables precise detection of tampering at the receiver side. This article presents an image watermarking and tamper detection technique based on a modified Hamming code. The watermarking process is applied to each individual pixel of the gray image, with each pixel is represented using 8 bits. The 4 higher plane bits of a pixel are designated as data bits, and based on these, 3 redundant bits are computed. These 3 redundant bits are then modified using a Logistic Map (LM). Furthermore, based on the 4 data bits and 3 modified redundant bits, a parity bit is computed. The 3 modified redundant bits along with the parity bit form the Watermark Bits (WBs), which are embedded into the Least Significant Bit (LSB) positions of the pixel. This process results in a watermarked pixel. In the event that a single bit is tampered with, it can be easily identified and corrected at the receiver according to the detection and correction logic. The Hiding Capacity (HC) is 4 bits per pixel, and the distortion measure is 31 dB. Although the similarity index is reduced the accuracy of tamper detection is significantly improved.


  • Open Access

    Digital Image Watermarking for Image Integrity Verification and Tamper Correction

    by Anantha Rao Gottimukkala, Anita Pradhan, Naween Kumar, Ashok Kumar Pradhan, Ranjan K. Senapati, Gandharba Swain

    Images transmitted through internet can be easily tampered by the available image editing tools. This article proposes a Hamming code based watermarking approach for tamper localization and correction of images. The original image is divided into various blocks with 8 consecutive pixels. The 64 bits of the 8 pixels are arranged into an 8 × 8 matrix of bits. A modified (7,4) Hamming code (MHC) is applied on first 7 most significant bits (MSBs) of each row of the matrix. The first 4 MSBs are data bits. The next 3 bits are redundant bits. The watermark bits are calculated from the 4 MSBs and stored in 3 redundant bits. Furthermore, the column parity for the first 7 columns of the 8 × 8 matrix is computed and embedded in the least significant bits (LSBs) of the 7 rows. Thereafter the column parity of the first 7 bits of 8th column is stored in 8th bit location of 8th column. This technique can detect 1-bit error or 2-bit error if it occurs in one of the 8 pixels of the block. Experimental outcomes prove that this proposed scheme maintains 4.0 bits per pixel with 36.94 dB peak signal-to-noise ratio (PSNR) and 0.9781 structural similarity (SSIM).


  • Open Access

    Optimized Video Watermarking with Entropy-Aware Block Selection and Modified Hénon-map Encryption

    by Ranjan K. Senapati, Birendra Biswal, Sandeep Kautish, Gandharba Swain, Ayman Altameem, Abdulaziz S. Almazyad, Ali Wagdy Mohamed

    Video watermarking is a technique used to embed information into video content, ensuring privacy protection and preventing data misuse. However, watermarking schemes that rely on a single transform method are often vulnerable to noise and geometric distortions. To overcome these limitations, this paper introduces a hybrid approach that combines singular value decomposition with discrete wavelet transform and redundant discrete wavelet transform for more effective watermark embedding. The proposed method optimizes the balance between imperceptibility and robustness by fine-tuning the scale factor through a custom fitness function using fruit fly optimization. The watermark is adaptively embedded based on the video’s transform type, sub-band selection, attack types, and keyframes. Block selection within sub-bands is guided by entropy, edge entropy, and contrast. Watermark embedding and extraction are performed on selected blocks that meet these criteria. Experimental results evaluated using metrics such as normalized cross-correlation, image fidelity, and peak signal-to-noise ratio, demonstrate the superiority of our method over state-of-the-art methods.


  • Image watermarking based on remainder value differencing and extended Hamming code

    by Anantha Rao Gottimukkala, Naween Kumar, Jatindra Kumar Dash, Gandharba Swain

    Due to the availability of various photo editing tools, intruders can tamper with an image very easily. So, various watermarking and tamper detection approaches have been proposed by researchers. Basically, tamper detection techniques focus on embedding the watermark, extracting the water mark, and identifying the tampered regions. But it is very important that the tampered pixels should also be corrected. We bring forward an image watermarking technique for tamper detection and correction using remainder value differencing (RVD) and extended Hamming code (EHC). It operates on a pixel group of size 2 × 2. Watermark bits (WBs) are generated from four most significant bits of the pixels in a pixel group by EHC and concealed in four lower bit planes by the principle of RVD. The WBs are extracted at the receiver along with the identification of tampered pixels. The tampered pixels are corrected by the developed correction logic. As the principle of RVD is used, precautions are taken to avoid the fall-off boundary problem. The efficacy of this technique is accessed through various quality metrics. It is noted that it performs better than the existing techniques. The recorded peak signal-to-noise ratio value is 45.49 dB with structural similarity value 0.9889. The tampered pixels are identified and corrected.