A dual autoencoder employing separable convolutional layers for image denoising and deblurring is represented. Combining two autoencoders is presented to gain higher accuracy and simultaneously reduce the complexity of neural network parameters by using separable convolutional layers. In the proposed structure of the dual autoencoder, the first autoencoder aims to denoise the image, while the second one aims to enhance the quality of the denoised image. The research includes Gaussian noise (Gaussian blur), Poisson noise, speckle noise, and random impulse noise. The advantages of the proposed neural network are the number reduction in the trainable parameters and the increase in the similarity between the denoised or deblurred image and the original one. The similarity is increased by decreasing the main square error and increasing the structural similarity index. The advantages of a dual autoencoder network with separable convolutional layers are demonstrated by a comparison of the proposed network with a convolutional autoencoder and dual convolutional autoencoder.

/pdf/dual-autoencoder-network-with-separable-convolutional-layers-rxrwoiha.pdf

Dual Autoencoder Network with Separable Convolutional Layers for Denoising and Deblurring Images

Underwater imaging poses significant challenges as water alters the behavior of light in comparison to air or vacuum. Therefore, it is crucial to effectively utilize the unique characteristics of unclear edges in hazy underwater images to achieve high-performance results in real-time applications. In this paper, we exploit such features as edges and visual perception in underwater haze images. To achieve this, we estimate the true transmission of the image by enhancing the visibility of discontinuous edges using the reverse gamma correction based on the Generalized Autoregressive Conditional Heteroscedasticity (GARCH) model. The parameters of the GARCH model are defined by the local and global pixel dynamics in adjacent neighborhoods, which improves the color orientation of the image while preserving image details. Additionally, we perform deseasoning to separate pixels while maintaining the natural contours of interactions between them. By considering the volatility of the RGB color channels, we predict the variance of the pixels based on the difference of the deseasoned pixels, which improves pixel intensity and allows for scene depth estimation. While estimating gamma correction and global ambient light, we deseasonalize image pixels based on their colors, enhancing the color of the final dehazed images. Using the greedy algorithm with a Convolutional Neural Network (CNN), our proposed method outperforms commonly used state-of-the-art methods. 

Reverse gamma correction based GARCH model for underwater image dehazing and detail exposure

Sound waves’ usage gives rise to ultrasound imagery as it is an indispensable tool for identification and diagnosis of the inner structure of the body for the detection of eventual sickness or anomalous tissues. Ultrasound imaging is a secure, reliable and economical diagnostic reading test. However, ultrasound imagery has an intrinsic property, i.e., speckle noise which diminishes the contrast and resolution of any imagery as it abolishes adequate details or boundaries which are mandatory to be well-preserved. In recent times, many researchers have contemplated distinct filtering schemes for mitigating noisy speckles to achieve good estimation in biomedical operations which include filtration schemes that transform the multiplicative noise into additive with the usage of algorithmic operations. This paper examines and compiles several schemes that are mostly utilized for the eradication of speckling from ultrasound imageries, and a comparison has been done for all the procedures studied based on experiments to illustrate the advantages of every technique. Since genuine ultrasound imageries are already degraded and actual noiseless imageries do not exist. So, synthetic imageries have been formed for testing the procedures. Nevertheless, the scheme has been offered for despeckling in real imageries. 

Despeckling of Ultrasound Imagery with Qualitative Filtering Techniques

CactiViT: Image-based smartphone application and transformer network for diagnosis of cactus cochineal

In the 21st century when the world is developing at a rapid speed, there is a dire need of having high resolution images as extent of human activities are growing at a rapid pace. However, changes in environmental condition makes a difficult task for making an image completely denoisable for the researchers. With the noise in the image, it is difficult to get a clear image or video quality. Due to this aspect, image denoising plays a vital role as it not only produces appealing images but however recovers the degraded values of pixels. Image denoising is a technique that is used to remove the noises which cause blur image/ distorted image. This research article attempts to study about the image denoising and various algorithms/techniques of image denoising and its application in various domains.

Image Denoising: Pre-processing for enhanced subsequent CAD Analysis

Agriculture was influential since ages in the development of human civilization. It is the backbone of our economic system and relies heavily on horticulture. To increase their productivity, various diseases may affect the plants that are to be handled by the farmers within time. For better yields and gains, the diseases in plants need to be controlled and eradicated to ensure the inhibition of the spread of the disease. In view of automating the process of disease detection, many researchers are developing image processing based solutions. The images captured through cameras are utilized to detect and distinguish the disease type using the features extracted from the image of the leaf or fruit. The different set of diseases are defined by the lesion properties and this enables the system to make decisions regarding the presence or absence of disease. This paper presents an approach towards optimized feature extraction and selection stage for Citrus greening disease. The resultant feature vector will train the machine learning model to distinguish between healthy and unhealthy images. The testing of different Machine learning models is done in this article for greening disease and an optimized solution is selected for the system design.

A Genetic Algorithm based Feature Optimization method for Citrus HLB Disease Detection using Machine Learning

Underwater image enhancement is necessary to study aquatic flora and fauna. However, due to light absorption and scattering, acquired subaqueous images are gravely hazed and degraded. This results in low contrast of the underwater image. In literature, many algorithms aim to dehaze and enhance an image’s quality. The NLP-GRM method aims to design an efficient algorithm that carries out superior results under a different environmental condition in terms of visual analysis and objective evaluation. Our approach integrates the gradient residual minimization (GRM) and then on local prior to a new method called NLP-GRM. Initially, an underwater image is processed through a nonlocal prior to dehaze the image by color assumption. The nonlocal prior (NLP) output works with robust GRM, reinforcing the edge strength and detail in the image post removal of underwater haze. The execution of the NLP-GRM algorithm has been observed by quantitative metrics and subjectively as well. The experimental results demonstrate that the NLP-GRM is superior to existing underwater enhancement methods.

Image Enhancement Using Nonlocal Prior and Gradient Residual Minimization for Improved Visualization of Deep Underwater Image

Internet speed plays a critical role today in the business and public services arena. High speed internet has taken industry to the new heights specifically in the field of information technology, online education, financial services, entertainment, health care and many more. Currently wireless fidelity (Wi-Fi) is commonly used at homes, organizations, airports and institutions to access the high speed internet. Researchers felt that enhancing the existing speed to the next level will further unlock a plethora of opportunities for the betterment of human life. This technology can be used to enhance the defense technique by replacing it with the radio frequency mechanism and can help to have a secure transmission of data. This paper will help readers to make aware of the new technology being explored called Light Fidelity (Li-Fi). Li-Fi is the next level of technology based on wireless optical communication which uses visible light for data transmission whereas presently Wi-Fi uses radio waves.

Light Fidelity: A future of wireless communication

An automated system for citrus disease detection relies upon using computer-aided tools. By minimizing manual error in disease detection and diagnosis, the computer-aided diagnostic (CAD) tools have acquired a major role in the resources needed to boost and increase the efficiency in production yield. The CAD tools have the properties of automatically speeding up the system’s decision-making capability on the complex disease data. Since these plant diseases can be devastating in terms of economic loss and loss of nutrition, the need for disease detection is growing at an early stage in order to meet human nutritional needs. The use of these computer-aided methods has enabled reducing the loss of production and early disease detection. This article proposes a machine learning-based methodology to automate the process of disease detection thereby reducing the manual efforts. While designing the complete system, the optimization and improvement at every substage are considered in this article so that the complete system possesses high levels of accuracy and thereby can equip farmers to deploy disease detection in a better way.

Ayush Dogra

Papers

Image Denoising: Pre-processing for enhanced subsequent CAD Analysis

A Genetic Algorithm based Feature Optimization method for Citrus HLB Disease Detection using Machine Learning

Image Enhancement Using Nonlocal Prior and Gradient Residual Minimization for Improved Visualization of Deep Underwater Image

Light Fidelity: A future of wireless communication

An Automated Citrus Disease Detection System Using Hybrid Feature Descriptor