Image Processing Domain Research Articles

Dual Attention Convolutional Neural Network Based on Adaptive Parametric ReLU for Denoising ECG Signals with Strong Noise.

Electrocardiogram (ECG) signal is one of the most important methods for diagnosing cardiovascular diseases but is usually affected by noises. Denoising is therefore necessary before further analysis. Deep learning-related methods have been applied to image processing and other domains with great success but are rarely used for denoising ECG signals. This paper proposes an effective and simple model of encoder-decoder structure for denoising ECG signals (APR-CNN). Specifically, Adaptive Parametric ReLU (APReLU) and Dual Attention Module (DAM) are introduced in the model. Rectified Linear Unit (ReLU) is replaced with the APReLU for better negative information retainment. The DAM is an attention-based module consisting of a channel attention module and spatial attention module, through which the inter-spatial and inter-channel relationship of the input data are exploited. We tested our model on the MIT-BIH dataset, and the results show that the APR-CNN can handle ECG signals with a different signal-to-noise ratio (SNR). The comparative experiment proves our model is better than other deep learning and traditional methods.Clinical Relevance- This paper proposed a method capable of denoising ECG signals with strong noise to alleviate difficulties for further medical analysis.

SiteForge: Detecting and localizing forged images on microblogging platforms using deep convolutional neural network

Microblogging applications are currently used to disseminate information with concise text and images. Nevertheless, they are also the largest platform for circulating forged images. Forged images are digital photographs that have been modified to deceive or distort the information they communicate. These manipulated images posted on microblogging apps like Twitter often create biased user emotions leading to harmful consequences like religious feuds or riots. Microblogging platforms and fact-checking industry are investing in artificial intelligence solutions to detect these forged images on time. Many image forensic techniques are proposed earlier, but their effectiveness falls short on real-world images shared over microblogging sites. As forged images shared over these platforms are typically altered using multiple manipulation techniques, it is hard for forensic techniques to detect them. This paper proposes a customized convolutional neural network with an attention mechanism to spot fake images shared over microblogging platforms. Deep learning convolutional networks learn the intrinsic feature set of images and can detect the forged images. To handle multiple manipulations in an image, the applied attention mechanism focuses on the most relevant image region to learn the inherent feature sets. The model utilizes High-pass filters from the image processing domain to initialize kernel weights of the neural network. This supports the proposed model to converge faster and achieve better accuracy. The pooling layers are designed to specifically handle images from microblogging sites. The solution is universal and can detect complex tampering scenarios like text-editing, face-swapping, copy-move, splicing and mirroring. Local Interpretable Model-agnostic Explanations (LIME) is utilized to localize the manipulated region in a forged image. LIME also adds interpretability and confidence to the proposed model, a common concern in deep learning models. The model is verified against the publicly available CASIA 2.0 dataset. An accuracy score of 94.7% is achieved, which is better than the previous state-of-art papers in fake image detection. In order to test the model on real-world images published on Twitter, a recent dataset is built from an Indian viewpoint. The model achieves a modest accuracy of 83.2% over the real-world Twitter dataset. The experiment proves that the proposed model can accurately detect the forged images over social platforms. It can be utilized in the fact-checking field to improve manual efforts. It will also support manual fact-checkers in swift decision making.

AHP validated literature review of forgery type dependent passive image forgery detection with explainable AI

Nowadays, a lot of significance is given to what we read today: newspapers, magazines, news channels, and internet media, such as leading social networking sites like Facebook, Instagram, and Twitter. These are the primary wellsprings of phony news and are frequently utilized in malignant manners, for example, for horde incitement. In the recent decade, a tremendous increase in image information generation is happening due to the massive use of social networking services. Various image editing software like Skylum Luminar, Corel PaintShop Pro, Adobe Photoshop, and many others are used to create, modify the images and videos, are significant concerns. A lot of earlier work of forgery detection was focused on traditional methods to solve the forgery detection. Recently, Deep learning algorithms have accomplished high-performance accuracies in the image processing domain, such as image classification and face recognition. Experts have applied deep learning techniques to detect a forgery in the image too. However, there is a real need to explain why the image is categorized under forged to understand the algorithm’s validity; this explanation helps in mission-critical applications like forensic. Explainable AI (XAI) algorithms have been used to interpret a black box’s decision in various cases. This paper contributes a survey on image forgery detection with deep learning approaches. It also focuses on the survey of explainable AI for images.

An Analysis of 3D Steganography Techniques

Steganography has become a preferred technique these days to successfully hide secret messages. Various research has been done in the past to justify and analyze suitable types of cover file, such as images, audio, videos, etc. Advancement in the image-processing domain has opened various possibilities of using three-dimensional (3D) images as cover files. In this paper, a systematic study of the research work done on 3D steganography in the last fifteen years has been carried out. The study is divided into different sections based on the types of algorithms used, additional security features, evaluation parameters, etc. Moreover, certain steganalysis techniques that are applicable for 3D steganography are also discussed.

Image Restoration using Recent Techniques: A Survey

Image Restoration is one of the challenging and essential milestones in the image processing domain. Digital image processing is a technique for manipulating digital images using a variety of computer algorithms. The process of transforming the degraded or damaged image to the original image can be known as Image Restoration. The image restoration process improves image quality by converting the degraded image into the original clean image. The techniques for image restoration are comprised of predefined parameters through which digital image gets processed for refinements. The purpose of restoration is to start with the acquired image and then estimate the original image as accurately as possible. A degraded image can be contaminated by any of a blur or noise or both. Many factors can contribute to image degradation, including poor capture, poor lighting, and poor eyesight. Medical science, defensive sensor systems, forensic detections, and astrology all rely on image restoration for accuracy. This paper discusses various image restoration techniques using recent trends for performance improvements.

Lossy Image Compression Techniques

In recent years, domain image processing plays a most significant role in realtime applications. Due to reserved bandwidth and capacity, images are needed to be compressed, enhance, and noise-free to reduce the storage space as well as the transmission time. Compression overcomes the problem such as insufficient bandwidth of network and storage of memory device. Nowadays, many techniques are used to compress an image with a good quality. In this paper, I briefly discuss the concept of image compression and review the techniques which are used in loss image compression. This technique losses some data by using transform codes. Two codes are used to compress an image such as wavelet and fractal. The techniques scalar and vector quantization are used to quantize an input and partition into sub-images. I briefly discussed about the 2D-Discrete wavelet transform codes with examples which were done by Mat lab software.

Hybrid Machine Learning Approach to Detect the Changes in SAR Images for Salvation of Spectral Constriction Problem

For implementing change detection approaches in image processing domain, spectral limitations in remotely sensed images are remaining as an unresolved challenge. Recently, many algorithms have been developed to detect spectral, spatial, and temporal constraints to detect digital change from the synthetic aperture radar (SAR) images. The unsupervised method is used to detect the appropriate changes in the digital images, which are taken between two different consecutive periods at the same scene. Many of the algorithms are identifying the changes in the image by utilizing a similarity index-based approach. Therefore, it fails to detect the original changes in the images due to the recurring spectral effects. This necessitated the need to initiate more research for suppressing the spectral effects in the SAR images. This research article strongly believes that the unsupervised learning approach can solve the spectral issues to correct in the appropriate scene. The convolutional neural network has been implemented here to extract the image features and classification, which will be done through a SVM classifier to detect the changes in the remote sensing images. This fusion type algorithm provides better accuracy to detect the relevant changes between different temporal images. In the feature extraction, the semantic segmentation procedure will be performed to extract the flattened image features. Due to this procedure, the spectral problem in the image will be subsided successfully. The CNN is generating feature map information and trained by various spectral images in the dataset. The proposed hybrid technique has developed an unsupervised method to segment, train, and classify the given input images by using a pre-trained semantic segmentation approach. It demonstrates a high level of accuracy in identifying the changes in images.

Designing a grey wolf optimization based hyper-parameter optimized convolutional neural network classifier for skin cancer detection

In recent history, Convolutional Neural Networks have attained breakthroughs in addressing many intractable problems in the domain of image processing. But its performance builds upon its chosen hyper parameters and it is a tedious job to manually fine tune these hyper parameters. Hence, in this research, an Automated Hyper-parameter Optimized Convolution Neural Net is proposed which is further applied to uncover the class of skin cancer. The method has utilized a Grey Wolf Optimization algorithm for optimizing the hyper parameters of CNN, by adopting a proper encoding scheme. The effectiveness of the model is verified by comparing it with the performance of Particle Swarm Optimization and Genetic algorithm based hyper-parameter optimized CNN applied on the International Skin Imaging Collaboration skin lesion multi class data set. Simulation results infer that the proposed model is able to produce a testing accuracy up to 98.33% which is around 4% and 1% more compared to PSO and GA based models respectively. Similarly with the proposed model, the testing loss realized is around 0.17% which is 39.2% and 15% less compared to PSO and GA based models respectively. The experimental results clearly demonstrate that the proposed model performs competitively compared to other reported models.

Grey wolf optimizer and other metaheuristic optimization techniques with image processing as their applications: a review

Image processing is an evolutionary field in the domain of computer vision that currently has a comprehensive spectrum of applications. It is being employed in image segmentation, image classification, medical imaging, image compression, etc. A lot of real-world prominent issues are tackling employed through these application techniques. These techniques can be employed by means of various algorithms; however, these offered immensely dominant outcomes with existing and modified optimization algorithms. Some of the metaheuristic optimization algorithms applied during the above techniques include Ant Colony Optimization (ACO), Genetic Algorithm (GA), Bat Algorithm (BA), Grey Wolf Optimizer (GWO), Evolutionary Strategy (ES), Particle Swarm Optimization (PSO), Genetic Programming (GP), and so forth. Hence, this manuscript’s main objective is that the study of several applied optimization algorithms and their variants thus lead to the various domain of image processing concludes it work more efficiently and robustly.

An Efficient Dimension Reduction based Fusion of CNN and SVM Model for Detection of Abnormal Incident in Video Surveillance

Performing dimensionality reduction in the camera captured images without any loss is remaining as a big challenge in image processing domain. Generally, camera surveillance system is consuming more volume to store video files in the memory. The normally used video stream will not be sufficient for all the sectors. The abnormal conditions should be analyzed carefully for identifying any crime or mistakes in any type of industries, companies, shops, etc. In order to make it comfortable to analyze the video surveillance within a short time period, the storage of abnormal conditions of the video pictures plays a very significant role. Searching unusual events in a day can be incorporated into the existing model, which will be considered as a supreme benefit of the proposed model. The massive video stream is compressed in preprocessing the proposed learning method is the key of our proposed algorithm. The proposed efficient deep learning framework is based on intelligent anomaly detection in video surveillance in a continuous manner and it is used to reduce the time complexity. The dimensionality reduction of the video captured images has been done by preprocessing the learning process. The proposed pre-trained model is used to reduce the dimension of the extracted image features in a sequence of video frames that remain as the valuable and anomalous events in the frame. The selection of special features from each frame of the video and background subtraction process can reduce the dimension in the framework. The proposed method is a combination of CNN and SVM architecture for the detection of abnormal conditions at video surveillance with the help of an image classification procedure. This research article compares various methods such as background subtraction (BS), temporal feature extraction (TFE), and single classifier classification methods.

Colour Detection Based Robotic Cart Using Raspberry Pi

One of the most critical subjects of embedded vision is color tracking in real time. Many computer vision applications begin by detecting and tracking moving objects in video scenes. Customers arriving at hypermarkets may benefit from this concept. A color detection algorithm locates pixels in an image that fit a predetermined color scheme. To differentiate detected pixels from the rest of the image, the color of the detected pixels can be modified. The robot is programmed to track objects by turning left and right to keep the target in view and driving forward and backward to keep the distance between the robot and the object steady. By maintaining a surrounding distance, detection of other objects of the same color pattern is ignored. By keeping a safe distance between the user and the robot, other objects of the same color pattern are not detected. The camera on an ARM11 Raspberry Pi computer attached to the robot is used to capture images. Using inbuilt python files, the acquired image is processed to locate the color using RGB varying pattern methodology. To make the product work smarter, this system also includes automatic billing via RFID reader and tag. The new concept of image processing domain is based on this device theory.

Molecular Modulation by Lentivirus-Delivered Specific shRNAs in Endoplasmic Reticulum Stressed Neurons.

The accumulation of unfolded proteins within the endoplasmic reticulum (ER), caused by any stress condition, triggers the unfolded protein response (UPR) through the activation of specialized sensors. UPR attempts first to restore homeostasis; but if damage persists the signaling induces apoptosis. There is increasing evidence that sustained and unresolved ER stress contributes to many pathological conditions including neurodegenerative diseases. Because the UPR controls cell fate by switching between cytoprotective and apoptotic processes, it is essential to understand the events defining this transition, as well as the elements involved in its modulation. Recently, we demonstrated that abnormal GM2 ganglioside accumulation causes depletion of ER Ca2+ content, which in turn activates PERK (PKR-like-ER kinase), one of the UPR sensors. Furthermore, PERK signaling participates in the neurite atrophy and apoptosis induced by GM2 accumulation. In this respect, we have established an experimental system that allows us to molecularly modulate the expression of downstream PERK components and thus change vulnerability of neurons to undergo neuritic atrophy. We performed knockdown of calcineurin (cytoprotective) and CHOP (pro-apoptotic) expression in rat cortical neuronal cultures. Cells were infected with lentivirus-delivered specific shRNA and then treated with GM2 at different times, fixed and immunostained with anti-MAP2 (microtube-associated protein 2) antibody. Later, cell images were recorded using a fluorescence microscope and total neurite outgrowth was evaluated by using the public domain image processing software ImageJ. The inhibition of expression of those PERK signaling components clearly made it possible to either accelerate or delay the neuritic atrophy induced by ER stress. This approach might be used in cell system models of ER stress to evaluate the vulnerability of neurons to neurite atrophy.

Band selection strategies for hyperspectral image classification based on machine learning and artificial intelligent techniques –Survey

As the hyperspectral image consists of hundreds of highly correlated spectral bands, the selection of informative and highly discriminative bands is necessary for hyperspectral image classification. The recent growth of machine learning and artificial intelligence techniques play a major role in various domains of hyperspectral image processing. In this paper, a comprehensive survey of machine learning and artificial intelligence technique-based band selection strategies for hyperspectral image classification is given. As per the outcome of this study, we have identified the research challenges and research for future directions in band selection strategies for hyperspectral image classification.

Identification of copy-move and splicing based forgeries using advanced SURF and revised template matching

Various image tampering detection approaches are used to find the variations or inconsistencies in statistical image features. But still these techniques lack behind to identify copy-move and splicing based manipulations. The manipulation in digital data encourages the crimes, particularly in the domain of image processing and computer vision-based applications. Therefore, to find image forgeries, new method needs to be designed so that originality of data is authenticated in the court of law or jurisdiction. To achieve, a pixel based forgery detection framework for copy-move and splicing based forgeries is suggested in this paper. Initially, pre-processing over image data is performed to enhance the textural information. The proposed system estimates various features using enhanced SURF and template matching for the identification of fake image regions. The relevant key parameters are estimated and compared with the calculated threshold value. The evaluation is carried out using CASIA forged image dataset. The results are evaluated and compared with other existing methods through a comprehensive set of experiments. The enhanced SURF method produces a forgery detection accuracy of 97%, while template matching gives 100% forgery detection. As a whole system, the accuracy is 97.5%. Thus, the demonstrated result shows that the proposed framework attains considerably more detection accuracy compared to other state-of-art techniques.

Hyperspectral Image Denoising via Clustering-Based Latent Variable in Variational Bayesian Framework

The hyperspectral-image (HSI) noise-reduction step is a very significant preprocessing phase of data-quality enhancement. It has been attracting immense research attention in the remote sensing and image processing domains. Many methods have been developed for HSI restoration, the goal of which is to remove noise from the whole HSI cube simultaneously without considering the spectral–spatial similarity. When a noise-removal algorithm is used globally to the entire data set, it would not eliminate all levels of noise, effectively. Furthermore, most of the existing methods remove independent and identically distributed (i.i.d.) Gaussian noise. The real scenarios are much more complicated than this assumption. The complexity created by natural noise that has a non-i.i.d. structure leads to inefficient methods containing underestimation and invalid performance. In this article, we calculated the spatial–spectral similarity criteria by defining a set of clustering-based latent variables (CLVs) in a Bayesian framework to improve the robustness. These criteria can be extracted using the clustering operators. Then, by applying the CLV to the variational Bayesian model, we investigated a new low-rank matrix factorization denoising approach based on the proposed clustering-based latent variable (CLV-LRMF) to remove noise with the non-i.i.d. mixture of Gaussian structures. Finally, we switched to the GPU for MATLAB implementation to reduce the runtime. The experimental results show that the performance has been improved by applying the proposed CLV and demonstrate the effectiveness of the proposed CLV-LRMF over other state-of-the-art methods.

NDVI COMPUTATION OF LISS III IMAGES USING QGIS

Feature Selection and Extraction is a very significant and mandatory part in the domain of image processing. After the relevant preprocessing operations, the relevant features have to be extracted using suitable algorithms. In multispectral imagery, the features are identified and extracted based on the applications and objectives of the analysis such as color, texture, brightness, intensity etc. Some of the prominent algorithms used for feature extraction are mean shift algorithm, Principal Component transformation, Wavelet based Transformation, Local Binary Patterns etc. Texture based feature detection and extraction is the most prominent method adopted which involves multispectral images. With respect to hyperspectral images, dimensionality is a critical issue to be dealt appropriately.

Lossy Image Compression Techniques

In recent years, domain image processing plays a most significant role in realtime applications. Due to reserved bandwidth and capacity, images are needed to be compressed, enhance, and noise-free to reduce the storage space as well as the transmission time. Compression overcomes the problem such as insufficient bandwidth of network and storage of memory device. Nowadays, many techniques are used to compress an image with a good quality. In this paper, I briefly discuss the concept of image compression and review the techniques which are used in loss image compression. This technique losses some data by using transform codes. Two codes are used to compress an image such as wavelet and fractal. The techniques scalar and vector quantization are used to quantize an input and partition into sub-images. I briefly discussed about the 2D-Discrete wavelet transform codes with examples which were done by Mat lab software.

SSDMNV2: A real time DNN-based face mask detection system using single shot multibox detector and MobileNetV2

Face mask detection had seen significant progress in the domains of Image processing and Computer vision, since the rise of the Covid-19 pandemic. Many face detection models have been created using several algorithms and techniques. The proposed approach in this paper uses deep learning, TensorFlow, Keras, and OpenCV to detect face masks. This model can be used for safety purposes since it is very resource efficient to deploy. The SSDMNV2 approach uses Single Shot Multibox Detector as a face detector and MobilenetV2 architecture as a framework for the classifier, which is very lightweight and can even be used in embedded devices (like NVIDIA Jetson Nano, Raspberry pi) to perform real-time mask detection. The technique deployed in this paper gives us an accuracy score of 0.9264 and an F1 score of 0.93. The dataset provided in this paper, was collected from various sources, can be used by other researchers for further advanced models such as those of face recognition, facial landmarks, and facial part detection process.

A survey on local binary pattern and gabor filter as texture descriptors of smart profiling systems

With the dramatic expansion of image information nowadays, image processing and computer visions are playing a significant role in terms of several applications such as image classification, image segmentation, pattern recognition, and image retrieval. One of the important features that have been used in many image applications is texture. The texture is the characteristic of a set of pixels that formed the image. Therefore, analyzing such texture would have a significant impact on segmenting the image or detecting important portions of such image. This paper aims to overview the feature extraction and description techniques depicted in the literature to characterize regions for images. In particular, two of popular descriptors will be examined including Local Binary Pattern (LBP) and Gabor Filter. The key characteristic behind such investigation lies in how the features of an image would contribute toward the process of recognition and image classification. In this regard, an extensive discussion is provided on both LBP and Gabor descriptors along with the efforts that have been intended to combine them. The reason behind investigating these descriptors is that they are considered the most common local and global descriptors used in the literature. The overall aim of this survey is to show current trends on using, modifying and adapting these descriptors in the domain of image processing.

Efficient image structural similarity quality assessment method using image regularised feature

Image regularised features play a critical role in image processing domain, by integrating regularised feature and structural similarity, a new full-reference image assessment method (IRF_SSIM) is proposed in this study. As well known, the gradient operator always be used to capture the edge information of the image, while the total variational regularised features can be adopted to calculate the detailed change information of image contrast and texture, as well as noise removal and edge retention. Therefore, the IRF_SSIM method extends the gradient features into the image regularised features to measure the structural changes in the image. In addition, image quality is also affected by variations of luminance and contrast. For a more comprehensive image quality assessment, the IRF_SSIM method considers the changes in structure, luminance and contrast simultaneously. In other words, the total image quality is estimated by structural similarity calculated by integrating the effects of image structure, luminance and contrast changes. Comparing with the representative methods, the experimental results illustrate that the IRF_SSIM method is highly consistent with the subjective assessment results.