International Journal of Signal and Imaging Systems Engineering (7 papers in press)

Regular Issues

• Lung Disease Classification using NF-RVFL Optimised by IROA with HybridGNet Segmentation and DIAG Augmented Data in chest X-Ray Analysis  
  by C.H. Ashok Babu, J. Pandu, Ravi Shankar Reddy Gosula 
  Abstract: Lung diseases like lung opacity, pneumonia, and COVID-19 pose major health challenges, creating a need for more advanced diagnostic solutions. Existing methods suffer from high space requirements, data imbalance, long processing times, and deep learning (DL) limitations like vanishing gradients and increased misclassification. To solve this, a novel lung disease classification framework is proposed that integrates data augmentation, segmentation, and classification. The diffusion-based In-Distribution Anomaly Generation (DIAG) pipeline uses multimodal latent diffusion models to generate realistic anomalous images, effectively resolving data imbalance. For segmentation, HybridGNet model combines convolutional neural networks (CNNs) with graph convolutional neural networks to accurately decode anatomical structures. Classification is performed using neuro-fuzzy random vector functional link (NF-RVFL) model, which employs an interpretable IF-THEN decision structure. Its parameters are optimised using Improved Remora Optimization Algorithm, reducing complexity and enhancing performance. Experimental results show an accuracy of 0.989, outperforming existing methods and demonstrating strong potential for precise lung disease classification.
  Keywords: Classification; segmentation; Data augmentation; Chest X-Ray; Lung disease.
  DOI: 10.1504/IJSISE.2025.10074986
   
  
• A Combined use of Wavelet Packet Decomposition and Convolutional Neural Network for Robust Bearing Fault Diagnosis  
  by Adel Boulassel, Hocine Bendjama, Mohamed Larbi Mihoub 
  Abstract: Rotating machines play a vital role in industrial systems, where unplanned downtime can lead to significant financial and operational losses. Rolling bearings, essential for supporting rotational motion, are prone to faults such as misalignment, imbalance, and wear, making their monitoring crucial for predictive maintenance. This paper proposes a novel fault diagnosis method combining wavelet packet decomposition (WPD) with a one-dimensional convolutional neural network (1D-CNN). WPD decomposes vibration signals to extract timefrequency fault features, from which statistical indicators such as kurtosis, energy, and crest factor are calculated. A new fault-sensitive indicator, defined as the product of these metrics, selects the most informative component for fault detection. The reconstructed signal is then classified using a 1D-CNN, achieving excellent diagnostic accuracy under varying loads. Compared with Random Forest (RF) model, the proposed method better captures non-stationary features and enhances classification performance, offering a robust solution for rotating machinery fault diagnosis.
  Keywords: Bearing faults diagnosis; wavelet packet decomposition; convolutional neural network; wind turbine; rotating machines.
  DOI: 10.1504/IJSISE.2025.10075044
   
  
• Robust Cotton Leaf Disease Classification Using Hybrid Deep Learning Approach with Multi- Loss Function  
  by Sailaja Madhu, V. Ravi Sankar 
  Abstract: This paper presents an advanced deep learning framework for the early detection of cotton leaf diseases. Pre-processing is performed using an adaptive median filter to remove noise from input images, enhancing clarity and quality. The improved images are then processed by a hybrid model combining the CN2-SE module and a modified Inception V3 network. The CN2-SE module extracts features, while the modified Inception V3 network also extracts complementary features. These features are merged into a unified set and passed through a multi-layer perceptron (MLP) for classification, which employs a hybrid multi-loss function to ensure improved accuracy while preventing overfitting. Tailored to address both sustainability challenges and global cotton demands, this approach achieves an outstanding classification accuracy of 99.41%, with strong performance in sensitivity, specificity, and recall. By advancing precision agriculture, this research provides an effective solution for automating cotton crop management and enhancing productivity and sustainability in cotton farming.
  Keywords: Robust Cotton leaf disease; CN2-SE; multi-layer perceptron with a multi-loss function; bacterial blight; leaf curl virus; and various fungal infections.
  DOI: 10.1504/IJSISE.2025.10075940
   
  
• Fault detection of industrial robotic arms based on joint tracking  
  by Yue Xu, Yinlong Zhang, Xin Wang, Chu Wang 
  Abstract: To ensure robotic system safety and reliability, accurate tracking of robotic arm and human joint positions and real-time monitoring of joint changes and early signs of wear or loosening are essential. Traditional methods suffer from limited accuracy and high maintenance costs in complex environments due to similar components, occlusion, and complicated backgrounds. To address these challenges, this paper proposes a fault detection method for robotic arms in complex scenarios that enhances joint identification and early warning capabilities. The approach integrates color detection, geometric perception, the Detect DBB module, and the integrated time processing (ITP) module. Leveraging deep feature learning, it enables real-time trajectory monitoring and precise identification of early wear and loosening. Experiments demonstrate that the proposed method improves detection accuracy, reduces unplanned downtime, extends the robotic arm's service life, and lowers maintenance costs, thus validating its effectiveness and practical value.
  Keywords: joint tracking; color detection; geometric perception; fault detection; joint tracking; color detection; geometric perception.
  DOI: 10.1504/IJSISE.2025.10074919
   
  
• A fusion framework for melanoma detection using SkCanNet model and hand-crafted features  
  by Apurva Shinde, Sangita Chaudhari 
  Abstract: Skin cancer remains a major global health concern, emphasising the need for early and accurate diagnosis. Computer-aided diagnosis (CAD) has been applied for skin cancer detection. A critical issue in skin cancer detection is the detection and removal of artefacts like hair and noise from lesion images. Existing hair removal techniques often fail to preserve lesion details, affecting classification accuracy. To enhance melanoma detection, an advanced pre-processing method is necessary while preserving lesion features with integration of deep learning with handcrafted features. This paper presents an advanced pre-processing technique using Modified E-shaver, Modified Dull Razor, and Adaptive Principal Curvature. SkCanNet, is proposed for feature extraction which is combined with standard classifiers like KNN, SVM, and logistic regression for classification. The proposed approach significantly enhances skin cancer detection accuracy, specificity, sensitivity, and F1-score. Compared to existing deep learning models proposed system presents a promising solution for early melanoma detection.
  Keywords: Melanoma skin cancer; grey-level cooccurrence matrix; local binary pattern; SkCanNet architecture.
  DOI: 10.1504/IJSISE.2026.10074889
   
  
• Elucidation of improved heuristic-assisted multi-dilated inception ResnetV2 with pyramidal attention for driver distraction detection  
  by Mohammed Sharfuddin Waseem, Shaik Munawar, Madugula Sujatha, Syed Abdul Moeed, Raghuram Bhukya 
  Abstract: Globally, driver distractions are required to be recognised since it becomes a major cause for traffic-related fatalities. In recent years, the models limit with examining the specific feature information of the input source. Hence the main purpose of this paper is to develop an automated model of detecting the driver behaviour. Further, the collected images are subjected as input to Adaptive Multi-dilated Inception ResnetV2 with Pyramidal Attention (AMIR-PA) for classifying the distracted behaviours. In order to further enhance the performance, the hyper-parameters are optimally selected using Renovated Position-based Crocodile Optimisation Algorithm (RP-COA). Finally, the proposed system is validated using different measures and compared among traditional approaches. After the validation, the proposed model acquires high accuracy value as 91.45% and 92.2% for ReLU and tanh activation function than existing models. Therefore, the findings reveal that the proposed system achieves higher detection results to evade the traffic accidents that occur in roadways.
  Keywords: driver distraction detection; Multi-dilated Inception ResnetV2 with Pyramidal Attention; RP-COA; Renovated Position-based Crocodile Optimisation Algorithm; adaptive multi-dilated inception ResnetV2 with pyramidal attention.
  DOI: 10.1504/IJSISE.2025.10075582
   
  
• Anomaly object detection from video with hybrid activation function-enabled optimised deep convolutional neural network  
  by Sangita Mahendra Rajput, M.D. Nikose 
  Abstract: The research proposed a novel DL model named Interactive Routing Algorithm Optimised Hybrid activation function enabled channel-wise deep convolutional neural network (IRA-C2NN) for accurate video anomaly detection. The model enhances the detection accuracy concerning the key parameters of the video input in hybridisation with the activation function. The selected frame from the video is subjected to moving object detection and tracking, the identified anomaly object is tracked in every frame using the YOLOv5. The combination of the pigeon and the particle swarm creates the IRA optimisation, which is hybridised with the classifier to enhance the detection rate of the anomalies, which minimises the error. The achievement of the research is shown as 96.03% with accuracy, 97% with sensitivity, 94.32% with specificity, and 96.33% with the F1 score.
  Keywords: anomaly detection; video surveillance; key frame selection; deep learning; convolutional neural network.
  DOI: 10.1504/IJSISE.2025.10075069