International Journal of Image and Graphics

Copy move forgery in digital images became a common problem due to the wide accessibility of image processing algorithms and open-source editing software. The human visual system cannot identify the traces of forgery in the tampered image. The proliferation of such digital images through the internet and social media is possible with a finger touch. These tampered images have been used in news reports, judicial forensics, medical records, and financial statements. In this paper, a detailed review has been carried on various copy-move forgery detection (CMFD) and localization techniques. Further, challenges in the research are identified along with possible solutions.

Rapid system and hardware development of X-ray computed tomography (CT) technologies has been accompanied by equally exciting advances in image reconstruction algorithms. Of the two reconstruction algorithms, analytical and iterative, iterative reconstruction (IR) algorithms have become a clinically viable option in CT imaging. The first CT scanners in the early 1970s used IR algorithms, but lack of computation power prevented their clinical use. In 2009, the first IR algorithms became commercially available and replaced conventionally established analytical algorithms as filtered back projection. Since then, IR has played a vital role in the field of radiology. Although all available IR algorithms share the common mechanism of artifact reduction and/or potential for radiation dose reduction, the magnitude of these effects depends upon specific IR algorithms. IR reconstructs images by iteratively optimizing an objective function. The objective function typically consists of a data integrity term and a regularization term. Therefore, different regularization priors are used in IR algorithms. This paper will briefly look at the overall evolution of CT image reconstruction and the regularization priors used in IR algorithms. Finally, a discussion is presented based on the reality of various reconstruction methodologies at a glance to find the preferred one. Consequently, we will present anticipation towards future advancements in this domain.

To solve the problem of road recognition when the robot is driving on the sidewalk, a novel sidewalk detection algorithm from the first-person perspective is proposed, which is crucial for robot navigation. The algorithm starts from the illumination invariance graph of the sidewalk image, and the sidewalk “seeds” are selected dynamically to get the sidewalk features for unsupervised feature learning. The final sidewalk region will be extracted by multi-threshold adaptive segmentation and connectivity processing. The key innovations of the proposed algorithm are the method of illumination invariance based on PCA and the unsupervised feature learning for sidewalk detection. With the PCA-based illumination invariance, it can calculate the lighting invariance angle dynamically to remove the impact of illumination and different brick colors’ influence on sidewalk detection. Then the sidewalk features are selected dynamically using the parallel geometric structure of the sidewalk, and the confidence region of the sidewalk is obtained through unsupervised feature learning. The proposed method can effectively suppress the effects of shadows and different colored bricks in the sidewalk area. The experimental result proves that the F-measure of the proposed algorithm can reach 93.11% and is at least 7.7% higher than the existing algorithm.

The art of mimicking a human’s responses and behavior in a programming machine is called Artificial intelligence (AI). AI has been incorporated in games in such a way to make them interesting, especially in chess games. This paper proposes a hybrid optimization tuned neural network (NN) to establish a winning strategy in the chess game by generating the possible next moves in the game. Initially, the images from Portable Game Notation (PGN) file are used to train the NN classifier. The proposed Locust Mayfly algorithm is utilized to optimally tune the weights of the NN classifier. The proposed Locust Mayfly algorithm inherits the characteristic features of hybrid survival and social interacting search agents. The NN classifier involves in finding all the possible moves in the board, among which the best move is obtained using the mini-max algorithm. At last, the performance of the proposed Locust mayfly-based NN method is evaluated with help of the performance metrics, such as specificity, accuracy, and sensitivity. The proposed Locust mayfly-based NN method attained a specificity of 98%, accuracy of 98%, and a sensitivity of 98%, which demonstrates the productiveness of the proposed mayfly-based NN method in pruning.

Face detection is a computer technology being used in a variety of applications that identify human faces in digital images. In many face recognition challenges, Convolutional Neural Networks (CNNs) are regarded as a problem solver. Occlusion is determined as the most common challenge of face recognition in realistic applications. Several studies are undergoing to obtain face recognition without any challenges. However, the occurrence of noise and occlusion in the image reduces the achievement of face recognition. Hence, various researches and studies are carried out to solve the challenges involved with the occurrence of occlusion and noise in the image, and more clarification is needed to acquire high accuracy. Hence, a deep learning model is intended to be developed in this paper using the meta-heuristic approach. The proposed model covers four main steps: (a) data acquisition, (b) pre-processing, (c) pattern extraction and (d) classification. The benchmark datasets regarding the face image with occlusion are gathered from a public source. Further, the pre-processing of the images is performed by contrast enhancement and Gabor filtering. With these pre-processed images, pattern extraction is done by the optimal local mesh ternary pattern. Here, the hybrid Whale–Galactic Swarm Optimization (WGSO) algorithm is used for developing the optimal local mesh ternary pattern extraction. By inputting the pattern-extracted image, the new deep learning model namely “CNN with Gated Recurrent Unit (GRU)” network performs the recognition process to maximize the accuracy, and also it is used to enhance the face recognition model. From the results, in terms of accuracy, the proposed WGSO-$\mathrm{\text{CNN}}+\mathrm{\text{GRU}}$ model is better by 4.02%, 3.76% and 2.17% than the CNN, SVM and SRC, respectively. The experimental results are presented by performing their comparative analysis on a standard dataset, and they assure the efficiency of the proposed model. However, many challenging problems related to face recognition still exist, which offer excellent opportunities to face recognition researchers in the future.

Scientific studies of the elements that influence the box office performance of Indian films have generally concentrated on post-production elements, such as those discovered after a film has been completed or released, and notably for Bollywood films. Only fewer studies have looked at regional film industries and pre-production factors, which are elements that are known before a decision to greenlight a film is made. This study looked at Indian films using natural language processing and machine learning approaches to see if they would be profitable in the pre-production stage. We extract movie data and English subtitles (as an approximation to the screenplay) for the top five Indian regional film industries: Bollywood, Kollywood, Tollywood, Mollywood, and Sandalwood, as they make up a major portion of the Indian film industry’s revenue. Subtitle Vector (Sub2Vec), a Paragraph Vector model trained on English subtitles, was used to embed subtitle text into 50 and 100 dimensions. The proposed approach followed a two-stage pipeline. In the first stage, Return on Investment (ROI) was calculated using aggregated subtitle embeddings and associated movie data. Classification models used the ROI calculated in the first step to predicting a film’s verdict in the second step. The optimal regressor–classifier pair was determined by evaluating classification models using $F_1$-score and Cohen’s Kappa scores on various hyperparameters. When compared to benchmark methods, our proposed methodology forecasts box office success more accurately.

Due to the significance of aquatic robotics and marine engineering, the underwater video enhancement has gained huge attention. Thus, a video enhancement method, namely Manta Ray Foraging Lion Optimization-based fusion Convolutional Neural Network (MRFLO-based fusion CNN) algorithm is developed in this research for enhancing the quality of the underwater videos. The MRFLO is developed by merging the Lion Optimization Algorithm (LOA) and Manta Ray Foraging Optimization (MRFO). The blur in the input video frame is detected and estimated through the Laplacian’s variance method. The fusion CNN classifier is used for deblurring the frame by combining both the input frame and blur matrix. The fusion CNN classifier is tuned by the developed MRFLO algorithm. The pixel of the deblurred frame is enhanced using the Type II Fuzzy system and Cuckoo Search optimization algorithm filter (T2FCS filter). The developed MRFLO-based fusion CNN algorithm uses the metrics, Underwater Image Quality Measure (UIQM), Underwater Color Image Quality Evaluation (UCIQE), Structural Similarity Index Measure (SSIM), Mean Square Error (MSE), and Peak Signal-to-Noise Ratio (PSNR) for the evaluation by varying the blur intensity. The proposed MRFLO-based fusion CNN algorithm acquired a PSNR of 38.9118, SSIM of 0.9593, MSE of 3.2214, UIQM of 3.0041 and UCIQE of 0.7881.

Cardiomegaly is a radiographic abnormality, and it has significant prognosis importance in the population. Chest X-ray images can identify it. Early detection of cardiomegaly reduces the risk of congestive heart failure and systolic dysfunction. Due to the lack of radiologists, there is a demand for the artificial intelligence tool for the early detection of cardiomegaly. The cardiomegaly X-ray dataset is extracted from the cheXpert database. Totally, 46195 X-ray records with a different view such as AP view, PA views, and lateral views are used to train and validate the proposed model. The artificial intelligence app named CardioXpert is constructed based on deep neural network. The transfer learning approach is adopted to increase the prediction metrics, and an optimized training method called adaptive movement estimation is used. Three different transfer learning-based deep neural networks named APNET, PANET, and LateralNET are constructed for each view of X-ray images. Finally, certainty-based fusion is performed to enrich the prediction accuracy, and it is named CardioXpert. As the proposed method is based on the largest cardiomegaly dataset, hold-out validation is performed to verify the prediction accuracy of the proposed model. An unseen dataset validates the model. These deep neural networks, APNET, PANET, and LateralNET, are individually validated, and then the fused network CardioXpert is validated. The proposed model CardioXpert provides an accuracy of 93.6%, which is the highest at this time for this dataset. It also yields the highest sensitivity of 94.7% and a precision of 97.7%. These prediction metrics prove that the proposed model outperforms all the state-of-the-art deep transfer learning methods for diagnosing cardiomegaly thoracic disorder. The proposed deep learning neural network model is deployed as the web app. The cardiologist can use this prognostic app to predict cardiomegaly disease faster and more robust in the early state by using low-cost and chest X-ray images.

Twitter Spam has turned out to be a significant predicament of these days. Current works concern on exploiting the machine learning models to detect the spams in Twitter by determining the statistic features of the tweets. Even though these models result in better success, it is hard to sustain the performances attained by the supervised approaches. This paper intends to introduce a deep learning-assisted spam classification model on twitter. This classification is based on sentiments and topics modeled in it. The initial step is data collection. Subsequently, the collected data are preprocessed with “stop word removal, stemming and tokenization”. The next step is feature extraction, wherein, the post tagging, headwords, rule-based lexicon, word length, and weighted holoentropy features are extracted. Then, the proposed sentiment score extraction is carried out to analyze their variations in nonspam and spam information. At last, the diffusions of spam data on Twitter are classified into spam and nonspams. For this, an Optimized Deep Ensemble technique is introduced that encloses “neural network (NN), support vector machine (SVM), random forest (RF) and convolutional neural network (DNN)”. Particularly, the weights of DNN are optimally tuned by an arithmetic crossover-based cat swarm optimization (AC-CS) model. At last, the supremacy of the developed approach is examined via evaluation over extant techniques. Accordingly, the proposed AC-CS $+$ ensemble model attained better accuracy value when the learning percentage is 80, which is 18.1%, 14.89%, 11.7%, 12.77%, 10.64%, 6.38%, 6.38%, and 6.38% higher than SVM, DNN, RNN, DBN, MFO $+$ ensemble model, WOA $+$ ensemble model, EHO $+$ ensemble model and CSO $+$ ensemble model models.

One of the most significant challenges appears to be securing the Internet of Things (IoT) communication network. As a corollary, information security has become the basis for establishing trustworthiness in IoT network communication. Cryptography is one of the ways for securing information in this case. However, the majority of current approaches are static, making them subject to security threats. As a consequence, a new concept, dynamic encryption, is growing rapidly in IoT communication. In this paper, a dynamic encryption algorithm (DSIT) has been proposed to secure IoT communication. This algorithm is based on Feistel and Substitution–Permutation Network. DSIT is a block cipher that takes the 64-bit block of plaintext, 64-bit secret key, and a secret dynamic box (D-box) as input. It produces a 64-bit ciphertext by performing eight rounds of the DSIT algorithm. For each round, the key and D-box are updated. This dynamic effect provides high security to a dynamic IoT network. The proposed algorithm has been executed in IoT environment using Raspberry Pi 3 Model B$+$ and 50% average Avalanche effect has been achieved. The proposed algorithm efficiently encrypts the image data to secure the communication and high resistant to cryptanalysis attacks.

Nowadays, hyperspectral imaging (HSI) attracts the interest of many researchers in solving the remote sensing problems especially in various specific domains such as agriculture, snow/ice, object detection and environmental monitoring. In the previous literature, various attempts have been made to extract the critical information through hyperspectral imaging which is not possible through multispectral imaging (MSI). The classification in image processing is one of the important steps to categorize and label the pixels based on some specific rules. There are various supervised and unsupervised approaches which can be used for classification. Since the past decades, various classifiers have been developed and improved to meet the requirement of remote sensing researchers. However, each method has its own merits and demerits and is not applicable in all scenarios. Past literature also concluded that deep learning classifiers are more preferable as compared to machine learning classifiers due to various advantages such as lesser training time for model generation, handle complex data and lesser user intervention requirements. This paper aims to perform the review on various machine learning and deep learning-based classifiers for HSI classification along with challenges and remedial solution of deep learning with hyperspectral imaging. This work also highlights the various limitations of the classifiers which can be resolved with developments and incorporation of well-defined techniques.

The advent in graphic rendering software and technological progress in hardware can generate or modify photorealistic computer graphic (CG) images that are difficult to identify by human observers. Computer-generated images are used in magazines, film and advertisement industry, medical and insurance agencies, social media, and law agencies as an information carrier. The forged computer-generated image created by the malicious user may distort social stability and impacts on public opinion. Hence, the precise identification of computer graphic and photographic image (PG) is a significant and challenging task. In the last two decades, several researchers have proposed different algorithms with impressive accuracy rate, including a recent addition of deep learning methods. This comprehensive survey presents techniques dealing with CG and PG image classification using machine learning and deep learning. In the beginning, broad classification of all the methods in to five categories is discussed in addition to generalized framework of CG detection. Subsequently, all the significant works are surveyed and are grouped into five types: image statistics methods, acquisition device properties-based techniques, color, texture, and geometry-based methods, hybrid methods, and deep learning methods. The advantages and limitations of CG detection methods are also presented. Finally, major challenges and future trends in the CG and PG image identification field are discussed.