International Journal of Image and Graphics

Nowadays, the image degradation field suffers from several challenges while processing underwater color images including color distortion and image blurring due to the scattering media. Moreover, to get appropriate multi-frame super-resolution images, there is essential for recovering a better quantity of images. Traditionally, the shift among images is directly evaluated when considering the under-sampled Low-Resolution (LR) images. On the other hand, the high-frequency LR image faces unreliability owing to the aliasing consequences of sub-sampling, but it will also degrade the recovery accuracy. This task design implements a novel image recovery model from the moving water surface by adopting the multi-objective adaptive higher-order spectral analysis. Image pre-processing, lucky region selection, and image recovery are the three main phases of this model. The bicoherence method and dice coefficient method are adopted for performing the lucky region selection. Finally, the adoption of the multi-objective adaptive bispectra method is used for performing the image recovery from the moving water surface. The improved Adaptive Fitness-oriented Random number-based Galactic Swarm Optimization (AFR-GSO) algorithm is used for optimizing the constraints of the bispectrum method. The experimental results verify the enrichment of image quality by the proposed model over the existing techniques.

Global food security can be influenced by the diseases in crop plants as several diseases straightforwardly influence the quality of the grains, vegetables, fruits, etc., which also results in affecting of agricultural productivity. Like other plants, the mango tree is also affected by several diseases, and also the identification of multi-disease classification with a single leaf is more complex, and also it is impossible to detect diseases with bare eyes. Based on the other plants, the mango tree is also affected by various diseases, which is more difficult to detect the disorders with bare eyes. It is error-prone, inconsistent, and unreliable. Here, the mango trees are affected during the production, and also affect the plant health regarding multi-diseases. When the plants are affected by the diseases, it may cause fewer amounts of productivity, as a result, impacting the economy. However, it is more critical to detect plant diseases with the large varieties of trees and plants. Various research tasks on deep learning approaches focus on identifying the diseases in plants including leaves and fruits. Thus, the main objective of this paper is to implement an effective and appropriate technique for diagnosing mango tree diseases and their symptoms through fruit and leaf images, and thus, there is a need for an appropriate system for cost-effective and early solutions to this problem. Hence, the main intention of this work is to implement an efficient and suitable technique for diagnosing mango tree diseases and also identify the symptoms through fruit and leaf images. Intending to overcome the existing challenges, there is a need for an appropriate system for achieving cost-effectiveness and also creating an early solution to resolve this problem. This paper intends to present novel deep learning models for mango tree multi-disease classification. Initially, the data collection is done for gathering the diseased parts of the mango tree in terms of leaf and fruit images. Then, the contrast enhancement of the images is performed by the “Contrast-Limited Adaptive Histogram Equalization (CLAHE)”. For the deep feature extraction of leaf images, and fruit images, Convolutional Neural Network (CNN) is employed, and the features from both inputs are concatenated for further processing. Further, the weighted feature selection is adopted for selecting the most significant features by the Adaptive Squirrel-Grey Wolf Search Optimization (AS-GWSO). Enhanced “Long Short Term Memory (LSTM)” is applied in the classification part with parameter optimization using the same AS-GWSO for enhancing classification accuracy. At last, the results of the designed system on various mango tree diseases verify that the designed approach has yielded the highest accuracy by evaluating conventional approaches. Therefore, it would also alleviate and treat the affected mango leaf diseases accurately.

Lung cancer is the leading cause of cancer-related death worldwide. Therefore, early diagnosis remains essential to allow access to appropriate curative treatment strategies. This paper presents a novel approach to assess the ability of Positron Emission Tomography/Computed Tomography (PET/CT) images for the classification of lung cancer in association with artificial intelligence techniques. We have built, in this work, a multi output Convolutional Neural Network (CNN) as a tool to assist the staging of patients with lung cancer. The TNM staging system as well as histologic subtypes classification were adopted as a reference. The VGG 16 network is applied to the PET/CT images to extract the most relevant features from images. The obtained features are then transmitted to a three-branch classifier to specify Nodal (N), Tumor (T) and histologic subtypes classification. Experimental results demonstrated that our CNN model achieves good results in TN staging and histology classification. The proposed architecture classified the tumor size with a high accuracy of 0.94 and the area under the curve (AUC) of 0.97 when tested on the Lung-PET-CT-Dx dataset. It also has yielded high performance for N staging with an accuracy of 0.98. Besides, our approach has achieved better accuracy than state-of-the-art methods in histologic classification.

Floods are the deadly and catastrophic disasters, causing loss of life and harm to assets, farmland, and infrastructure. To address this, it is necessary to devise and employ an effective flood management system that can immediately identify flood areas to initiate relief measures as soon as possible. Therefore, this research work develops an effective flood detection method, named Anti- Corona-Shuffled Shepherd Optimization Algorithm-based Deep Quantum Neural Network (ACSSOA-based Deep QNN) for identifying the flooded areas. Here, the segmentation process is performed using Fuzzy C-Means with Spatial Constraint Multi-Kernel Distance (MKFCM_S) wherein the Fuzzy C-Means (FCM) is modified with Spatial Constraints Based on Kernel-Induced Distance (KFCM_S). For flood detection, Deep QNN has been used wherein the training progression of Deep QNN is done using designed optimization algorithm, called ACSSOA. Besides, the designed ACSSOA is newly formed by the hybridization of Anti Corona Virus Optimization (ACVO) and Shuffled Shepherd Optimization Algorithm (SSOA). The devised method was evaluated using the Kerala Floods database, and it acquires the segmentation accuracy, testing accuracy, sensitivity, and specificity with highest values of 0.904, 0.914, 0.927, and 0.920, respectively.

Nowadays, the iris recognition system is one of the most widely used and most accurate biometric systems. The iris segmentation is the most crucial stage of iris recognition system. The accurate iris segmentation can improve the efficiency of iris recognition. The main objective of iris segmentation is to obtain the iris area. Recently, the iris segmentation methods based on convolutional neural networks (CNNs) have been grown, and they have improved the accuracy greatly. Nevertheless, their accuracy is decreased by low-quality images captured in uncontrolled conditions. Therefore, the existing methods cannot segment low-quality images precisely. To overcome the challenge, this paper proposes a robust convolutional neural network (R-Net) inspired by UNet for iris segmentation. R-Net is divided into two parts: encoder and decoder. In this network, several layers are added to ResNet-34, and used in the encoder path. In the decoder path, four convolutions are applied at each level. Both help to obtain suitable feature maps and increase the network accuracy. The proposed network has been tested on four datasets: UBIRIS v2 (UBIRIS), CASIA iris v4.0 (CASIA) distance, CASIA interval, and IIT Delhi v1.0 (IITD). UBIRIS is a dataset that is used for low-quality images. The error rate (NICE1) of proposed network is 0.0055 on UBIRIS, 0.0105 on CASIA interval, 0.0043 on CASIA distance, and 0.0154 on IITD. Results show better performance of the proposed network compared to other methods.

Over the past 10 years, deep learning has enabled significant advancements in the improvement of noisy speech. In an end-to-end speech enhancement, the deep neural networks transform a noisy speech signal to a clean speech signal in the time domain directly without any conversion or estimation of mask. Recently, the U-Net-based models achieved good enhancement performance. Despite this, some of them may neglect context-related information and detailed features of input speech in case of ordinary convolution. To address the above issues, recent studies have upgraded the performance of the model by adding various network modules such as attention mechanisms, long and short-term memory (LSTM). In this work, we propose a new U-Net-based speech enhancement model using a novel lightweight and efficient Shuffle Attention (SA), Gated Recurrent Unit (GRU), residual blocks with dilated convolutions. Residual block will be followed by a multi-scale convolution block (MSCB). The proposed hybrid structure enables the temporal context aggregation in time domain. The advantage of shuffle attention mechanism is that the channel and spatial attention are carried out simultaneously for each sub-feature in order to prevent potential noises while also highlighting the proper semantic feature areas by combining the same features from all locations. MSCB is employed for extracting rich temporal features. To represent the correlation between neighboring noisy speech frames, a two Layer GRU is added in the bottleneck of U-Net. The experimental findings demonstrate that the proposed model outperformed the other existing models in terms of short-time objective intelligibility (STOI), and perceptual evaluation of the speech quality (PESQ).

Studies deem that about 1 percent of the human population is affected by epileptic seizures on a global scale. It is characterized as an undue neuronal discharge in the brain and degrades the quality of life of the patients to a large extent. Children being unaware of a sudden onset of seizures could be affected by severe injury or even mortality. Machine-learning-based epileptic seizure detection from EEG (Electro-Encephalogram) signals have always been a hot area of research. However, the majority of the research works rely on correlated non-linear features extracted from the EEG signals, causing a high-computational overhead, and challenging their application in real-time clinical diagnosis.

This study proposes a robust seizure detection framework using Gaussian Deep Boltzmann Machine-based classifier and Symplectic Geometric Decomposition (SGD)-based features. The simplified eigenvalues derived through Symplectic Similarity Transform (SST) are employed as feature vectors for the classifier, eliminating the need for a deliberate feature extraction procedure. The study examines the transferability capability of the suggested framework in discriminating seizures in both neonates and pediatric subjects in unison, experimenting with classical annotated datasets. The model yielded a mean accuracy of about 97.91% and an F1 Score of 0.935 in pediatric seizure detection, and mean sensitivity and specificity of 99.05% and 98.28%, in neonatal seizure detection tasks, respectively. Thus, the model can be deemed comparable to the available state-of-the-art seizure detection frameworks.

The software is applied in various areas so the quality of the software is very important. The software defect prediction (SDP) is used to solve the software issues and enhance the quality. The robustness and reliability are the major concerns in the existing SDP approaches. Hence, in this paper, the hybrid optimization-based neural network (Optimized NN) is developed for the effective detection of the defects in the software. The two main steps involved in the Optimized NN-based SDP are feature selection and SDP utilizing Optimized NN. The data is fed forwarded to the feature selection module, where relief algorithm selects the significant features relating to the defect and no-defects. The features are fed to the SDP module, and the optimal tuning of NN classifier is obtained by the hybrid optimization developed by the integration of the social spider algorithm (SSA) and gray wolf optimizer (GWO). The comparative analysis of the developed prediction model reveals the effectiveness of the proposed method that attained the maximum accuracy of 93.64%, maximum sensitivity of 95.14%, maximum specificity of 99%, maximum $F_1$-score of 93.53%, and maximum precision of 99% by considering the K-fold.

Diabetes is a terrible health situation characterized by high-rise blood glucose levels. If it is not predicted at an early stage, then it generates the problems in the human body like kidney failure or premature death, and stroke. Controlling blood glucose levels provides patients with helpful dietary recommendations, which are critical components of diabetes management. In the past decades, diverse conventional approaches have been executed to predict the beginning stages of diabetes mellitus depending on physical and substance tests. Still, developing a new framework that can effectively diagnose diabetes mellitus-affected patients is required. To this end, the major target of this task is to predict diabetes mellitus with an advanced accuracy rate with the help of the Heuristic-based Ensemble Model Selection Strategy (H-EMSS). In the data collection phase, the Pima Indian Diabetes dataset (PID) is taken from the storage area of UCI. The data cleaning is performed in the pre-processing stage, which is the technique of removing or fixing, corrupted, incorrect, duplicate, incomplete data, or incorrectly formatted, inside a dataset. Then, the diabetes prediction is accomplished by the H-EMSS. Here, 10 base learners like Naive Bayes (NB), Convolutional Neural Network (CNN), Linear Regression (LR), Deep Neural Network (DNN), Support Vector Machine (SVM), Artificial Neural Network (ANN), Decision Tree (DT), Random Forest (RF), Auto Encoder (AE) and Recurrent Neural Network (RNN) are considered. From these, three classifiers are optimally selected by the Modified Scalar Factor-based Elephant Herding Optimization (MSF-EHO), so that the prediction rate will be high. The suggested methodology’s efficacy is also compared and analyzed, with the findings demonstrating the recommended model’s superiority. The overall evaluation is that the Root Mean Square Error (RMSE) of the designed Modified Scalar Factor-based Elephant Herding Optimization-Heuristic-based Ensemble Model Selection Strategy (MSF-EHO-H-EMSS) attains 4.601% and also the Mean Absolute Error (MAE) on the designed method achieves 0.99%. Thus, the given outcomes of the designed method revealed that it achieves elevated performance than the other existing techniques regarding diverse error metrics.

In this paper, we changed the methodology for pixel value differencing. The proposed method work on RGB color images improves the existing PVD technique in terms of embedding capacity and overcomes the issue of falling off boundaries in the traditional PVD technique, and provides security to the secret message from histogram quantization attack. Color images are composed of three different color channels (red, green, and blue), so we cannot apply the traditional pixel value differencing algorithm to them. Due to that, the proposed technique divides the RGB photograph in red, blue, and green channels. Following that the modified pixel value differencing algorithm is employed to all successive pixels of color channels. We get the total embedding capacity by adding the embedding capacities of each color component. After embedding the data, we concatenate the color channels to get the stegoimage. On a series of color images, we tested our pixel value differencing approach and found that the stego-picture’s visual excellence and payload capacity were reasonable. The variation in histogram between the stego and cover photographs was minor, making it resistant to histogram quantization attacks, and the suggested approach also solves the issue of falling off the boundary.

Cybersecurity has received greater attention in modern times due to the emergence of IoT (Internet-of-Things) and CNs (Computer Networks). Because of the massive increase in Internet access, various malicious malware have emerged and pose significant computer security threats. The numerous computing processes across the network have a high risk of being tampered with or exploited, which necessitates developing effective intrusion detection systems. Therefore, it is essential to build an effective cybersecurity model to detect the different anomalies or cyber-attacks in the network. This work introduces a new method known as Wavelet Deep Convolutional Neural Network (WDCNN) to classify cyber-attacks. The presented network combines WDCNN with Enhanced Rain Optimization Algorithm (EROA) to minimize the loss in the network. This proposed algorithm is designed to detect attacks in large-scale data and reduces the complexities of detection with maximum detection accuracy. The proposed method is implemented in PYTHON. The classification process is completed with the help of the two most famous datasets, KDD cup 1999 and CICMalDroid 2020. The performance of WDCNN_EROA can be assessed using parameters like specificity, accuracy, precision F-measure and recall. The results showed that the proposed method is about 98.72% accurate for the first dataset and 98.64% for the second dataset.

With the development of computer technology, image stylization has become one of the hottest technologies in image processing. To optimize the effect of artistic image style conversion, a method of artistic image style conversion optimized by attention mechanism is proposed. The CycleGAN network model is introduced, and then the generator is optimized by the attention mechanism. Finally, the application effect of the improved model is tested and analyzed. The results show that the improved model tends to be stable after 40 iterations, the loss value remains at 0.3, and the PSNR value can reach up to 15. From the perspective of the generated image effect, the model has a better visual effect than the CycleGAN model. In the subjective evaluation, 63 people expressed satisfaction with the converted artistic image. As a result, the cyclic generative adversarial network model optimized by the attention mechanism improves the clarity of the generated image, enhances the effect of blurring the target boundary contour, retains the detailed information of the image, optimizes the image stylization effect, and improves the image quality of the method and application value of the processing field.