A DL based Predictive Model using Optimization Techniques for Diabetes

A method for implementing Gaussian low-pass filter by using switched-current technique has been presented in this paper. The stable rational transfer function for Gaussian-shaped low-pass filter is approximated by the multiplication of the follower-integrator functions based on central limit theorem. An SI filter structure constructed from the inverse follow-the-leader feedback configuration has been proposed to synthesize the obtained transfer function, in which the switched-current bilinear integrator is employed as the building cell. Explicit design formulas are also derived. The 8th-order Gaussian low-pass filter is selected as the example to elaborate the design details. Simulation result shows that the proposed design method has high approximation precision and low magnitude sensitivity.

ABSTRACTThe pan-sharpening scheme combines high-resolution panchromatic imagery (HRPI) data and low-resolution multispectral imagery (LRMI) data to get a single merged high-resolution multispectral image (HRMI). The pan-sharpened image has extensive information that will promote the efficiency of image analysis methods. Pan-sharpening technique is considered as a pixel-level fusion scheme utilized for enhancing LRMI using HRPI while keeping LRMI spectral information. In this article, an efficient optimized integrated adaptive principal component analysis (APCA) and high-pass modulation (HPM) pan-sharpening method is proposed to get excellent spatial resolution within fused image with minimal spectral distortion. The proposed method is adjusted with multi-objective optimizationto determine the optimal window size and σfor the Gaussian low-pass filter (GLPF) and gain factor utilized for adding the high-pass details extracted from the HRPI to the LRMI principlecomponent of maximum correlation. Optimization results show that if the spatial resolution ratio of HRPI to LRMI is 0.50, then a GLPF of 5 × 5 window size and σ = 1.640 yields HRMI with low spectral distortion and high spatial quality. If the HRPI/LRMI spatial resolution ratio is 0.25, then a GLPF of 7 × 7 window size and σ = 1.686 yields HRMI with low spectral distortion and high spatial quality. Simulation tests demonstrated that the proposed optimized APCA–HPM fusion scheme gives adjustment between spectral quality and spatial quality and has small computational and memory complexity.

Enhancing underwater image via color correction and Bi-interval contrast enhancement

The comprehensive use of high-resolution remote sensing (HRS) images and deep learning (DL) methods can be used to further accurate urban green space (UGS) mapping. However, in the process of UGS segmentation, most of the current DL methods focus on the improvement of the model structure and ignore the spectral information of HRS images. In this paper, a multiscale attention feature aggregation network (MAFANet) incorporating feature engineering was proposed to achieve segmentation of UGS from HRS images (GaoFen-2, GF-2). By constructing a new decoder block, a bilateral feature extraction module, and a multiscale pooling attention module, MAFANet enhanced the edge feature extraction of UGS and improved segmentation accuracy. By incorporating feature engineering, including false color image and the Normalized Difference Vegetation Index (NDVI), MAFANet further distinguished UGS boundaries. The UGS labeled datasets, i.e., UGS-1 and UGS-2, were built using GF-2. Meanwhile, comparison experiments with other DL methods are conducted on UGS-1 and UGS-2 to test the robustness of the MAFANet network. We found the mean Intersection over Union (MIOU) of the MAFANet network on the UGS-1 and UGS-2 datasets was 72.15% and 74.64%, respectively; outperforming other existing DL methods. In addition, by incorporating false color image in UGS-1, the MIOU of MAFANet was improved from 72.15% to 74.64%; by incorporating vegetation index (NDVI) in UGS-1, the MIOU of MAFANet was improved from 72.15% to 74.09%; and by incorporating false color image and the vegetation index (NDVI) in UGS-1, the MIOU of MAFANet was improved from 72.15% to 74.73%. Our experimental results demonstrated that the proposed MAFANet incorporating feature engineering (false color image and NDVI) outperforms the state-of-the-art (SOTA) methods in UGS segmentation, and the false color image feature is better than the vegetation index (NDVI) for enhancing green space information representation. This study provided a practical solution for UGS segmentation and promoted UGS mapping.

In this article, it is aimed to classify healthy and four different plant diseases from Cassava plant leaf images. For this purpose, the “Cassava-Leaf-Disease-Classification” data set, which is an up-to-date and difficult data set published in 2020, was used. The used data set includes a total of 21,397 images consisting of healthy and four different diseases. In the study, in the MATLAB environment, the images were first subjected to the Chan-Vese (CV) Segmentation method and the area of interest was determined. Features were extracted with the ResNet 50 and MobileNetV2 deep learning architectures from the detected areas. Extracted features are classified by Support Vector Machine and K-Nearest Neighbor algorithms. The images are divided into two as training and testing according to the K-fold 5 value. The average highest success rates in training and test data were achieved by using the ResNet50 architecture and SVM classifier together, as a result of segmentation. As a result of training and testing processes, 85.4% and 84.4% success rates were obtained, respectively. At the end of the test process of the study, a trained network was obtained according to ResNet50, where the highest success rates were obtained, and MobileNetV2, another deep learning architecture used in the study. It has been compiled with MATLAB Builder NE in order to run these two networks in the web environment. The methods obtained as a result of the compilation are integrated into the ASP.NET MVC5 programming language. Finally, it has been made available to manufacturers with a web-based embedded interface. Thus, a deep learning-based decision support system has been developed that can be easily used by all manufacturers in the web environment.

Over the past few decades, convolutional neural networks (CNNs) have achieved outstanding results in addressing a broad scope of computer vision problems. Despite these improvements, fully convolutional Siamese neural networks (FCSNN) still hardly adapt to complex scenes, such as appearance change, scale change, similar objects interference, etc. The present study focuses on an enhanced FCSNN based on convolutional block hyperparameters optimization, a new activation function (ModReLU) and Gaussian low pass filter. The optimization of hyperparameters is an important task, as it has a crucial ascendancy on the tracking process performance, especially when it comes to the initialization of weights and bias. They have to work efficiently with the following activation function layer. Inadequate initialization can result in vanishing or exploding gradients. In the first method, we propose an optimization strategy for initializing weights and bias in the convolutional block to ameliorate the learning of features so that each neuron learns as much as possible. Next, the activation function normalizes the output. We implement the convolutional block hyperparameters optimization by setting the convolutional weights initialization to constant, the bias initialization to zero and the Leaky ReLU activation function at the output. In the second method, we propose a new activation, ModReLU, in the activation layer of CNN. Additionally, we also introduce a Gaussian low pass filter to minimize image noise and improve the structures of images at distinct scales. Moreover, we add a pixel-domain-based color adjustment implementation to enhance the capacity of the proposed strategies. The proposed implementations handle better rotation, moving, occlusion and appearance change problems and improve tracking speed. Our experimental results clearly show a significant improvement in the overall performance compared to the original SiamFC tracker. The first proposed technique of this work surpasses the original fully convolutional Siamese networks (SiamFC) on the VOT 2016 dataset with an increase of 15.42% in precision, 16.79% in AUPC and 15.93% in IOU compared to the original SiamFC. Our second proposed technique also reveals remarkable advances over the original SiamFC with 18.07% precision increment, 17.01% AUPC improvement and an increase of 15.87% in IOU. We evaluate our methods on the Visual Object Tracking (VOT) Challenge 2016 dataset, and they both outperform the original SiamFC tracker performance and many other top performers.

<b><sc>Abstract.</sc></b> Deep learning has achieved great success in computer vision. However, most of the existing methods are designed and tested on public datasets without severe occlusions, which may lead to low recall or even algorithm failure in real-world animal industry where occlusions commonly exist. Therefore, whether deep learning methods are still applicable in scenes with frequently occurring occlusions remains to be verified. This study assessed the applicability and capacity limit of deep learning methods on scenarios of pigs in farrowing pens as a case study. We collected 1,100 images from a farrowing pen where farrowing crate was adopted, and 94.8% of images have at least one piglet under occlusion. Six state-of-the-art deep learning methods in three major computer vision tasks, including object detection (Faster RCNN and Yolov4), semantic segmentation (FCN and Unet), and instance segmentation (Mask RCNN and Yolact++) were evaluated, respectively. To further challenge these methods, we randomly and gradually added virtual occlusions in ten cases on the images and tested their capacity limitation under extreme occlusion. The result showed that all the six deep learning methods achieved satisfactory performance on the original baseline case while the performance fell when occlusion became heavier. Object detection methods were less affected by occlusion than instance segmentation methods due to their rougher output. Semantic segmentation method Unet demonstrated the least effect by occlusion due to its pixel-wise classification characters, while the instance segmentation method Mask RCNN was most affected by occlusion. We suggest further study on segmentation method in animal industry.

The aim of this study is to develop a image segmentation system for Histopathological images by using Deep Learning Methods. In today's world cancer is a world wide problem for a lot of people from all around the world. Over several years in scientific studies showed us that scientist tried to find a way to cure cancer or reduce the damage or increase the chance of survival from cancer. As studies shows that early diagnosis of the cancer plays a huge part on the treatment process and highly increases the chance of survival of the cancer patients. Most of the cancer diagnosis system uses a medical image analyze. However most of the cancer types requires is a large amount of test and examination on medical cell images. It is a hard and time consuming process. As a result of this, there is a need has arisen for image segmentation. With the development on computer science, computers started to play a big role for making this process much more fast and accurate. Recent development on Machine Learning and Deep Learning methods can provide efficient image segmentation for large amount of data in a agreeable time line. There was a lot of studies which focused on traditional methods for image segmentation but in this study we aim on the recent development on Deep Learning methods which is used for image segmentation. First of all, an introduction about image segmentation for Histopathological images by Deep Learning Methods is given. After that, literature review about recent studies which is covered our main topic is given. Then, our feasibility studies and system analysis is addressed. Then, implementation and performance studies is given. Finally, an assessment of this study on image segmentation by Deep Learning methods into clinical images are given. The main topic as Deep Learning methods we targeted U-Net, Mask R-CNN, UNet++ and YOLO. We implemented those models to the medical histopathologial image data. We provided the process, result and analyzed outcome and the summary for all the models that we used in this paper.

Introduction: One of the leading causes of death among people is brain tumors. Accurate tumor classification leads to appropriate decision-making and providing the most efficient treatment to the patients. This study aims to optimize of the brain tumor MR images classification accuracy using the optimal threshold, PCA and training Adaptive Neuro Fuzzy Inference System (ANFIS) with different repetitions. Materials and Methods: This procedure used in this study consists of five steps: (1) T1, T2 weighted images collection, (2) tumor separation with different threshold levels, (3) feature extraction, (4) presence and absence of feature reduction applying principal component analysis (PCA) and (5) ANFIS classification with 0, 20, and 200 training repetitions. Results: ANFIS accuracy was 40%, 80% and 97% for all features and 97%, 98.5% and 100% for the 6 selected features by PCA in 0, 20 and 200 training repetitions, respectively. Conclusion: The findings of the present study showed that accuracy can be raised up to 100% by using an optimized threshold method, PCA and increasing training repetitions.

In this proposed work, a dual image watermarking algorithm is used to protect the data against copyright violations. In this work, the DICOM image is used as a host image. Two watermark images used are the MNNIT logo and the personal data of the patient. This method utilizes the advantages of Schur decomposition, lifting wavelet transform (LWT), discrete cosine transform (DCT) and singular value decomposition (SVD). The scaling factor is a vital parameter of watermarking technique. The firefly optimization technique is used to get the optimized scaling factor. The Speeded-up robust features (SURF) are used for watermarking authentication. To evaluate the performance of the proposed algorithm, peak signal-to-noise ratio (PSNR), normalized correlation coefficient (NCC), and structural similarity index measurement (SSIM) are used. The proposed method is tested against various attacks such as Salt and Pepper noise, Gaussian noise, Gaussian low pass filter, Average filter, Median filter, Histogram equalization, Sharpening, Rotation and Region of interest filtering. The proposed algorithm shows a high level of robustness and imperceptibility. It is found that the features of the input host image and the watermarked image are matching correctly on applying the SURF technique.

Early detection of Alzheimer’s disease (AD) has become one of the major research topics nowadays. The utilization of the computerized system may help medical experts to better understand and analyse the magnetic resonance imaging (MRI) images of AD patients for early detection. One of the commonly steps taken for the analysis of the image is image denoising using certain filters. However, finding shows that previous researchers use different approaches. This study aims to analyse the effect of different denoising techniques towards detection of Alzheimer’s disease. Data of two different groups (AD patients and Normal Control) were collected from Alzheimer's Disease Neuroimaging Initiative (ADNI). Then brain extraction and skull stripping were performed. Several image denoising techniques were implemented for both groups namely median filter, Wiener filter, histogram equalization filter and Gaussian low pass filter. After that, all images underwent texture feature extraction process and analysis were made to see the effect of those denoising techniques towards the features of Gray-Level Co-occurrence Matrix (GLCM) extracted which are the contract, correlation, energy and homogeneity features. The result shows that the use of mentioned denoising filters do not give effect to the extracted features. All values of contrast, correlation, energy and homogeneity cannot clearly distinguish between AD and NC groups. Without any filters on the other hand, contrast feature gives the best output in distinguishing between AD and NC groups with the normalized value of 0.1. The result from this study may help in thorough investigation of other features or hybrid features that could be used for the purpose of detection and classification of AD.

The extraction of rotation invariant representation is important for many signal processing tasks such as image analysis, computer vision, pattern recognition and so forth. In this paper, we show that, under certain conditions, the Two-Dimensional Fractional Fourier Transform (2D-FRFT) possesses this attractive property through mathematical analysis and extensive computer simulations. Based on this property, we propose a novel digital image watermarking method which combines 2D chirp signal with the addition and rotation invariant properties of 2D-FRFT in order to improve robustness and security of digital image watermark in 2D-FRFT domain. Experimental results show that the proposed method is robust against numerous watermarking attacks including rotation geometrical transform, JPEG compression, crop, median filtering, histogram equalization, salt & pepper noise, Gaussian lowpass filter, shift, dithering and Gaussian noise, and secures against unauthorized information copying and redistribution.

This paper proposes a color filter array interpolation method, which expands the preceding gradient based scheme by adopting Gaussian low-pass filter. It starts from the assumption that high frequency components of the green plane in an image are almost the same as those of the red or blue planes in the same image. We identify that the conventional gradient method can make use of this idea and can be expanded by Gaussian low-pass filter to enhance its performance. To determine the reasonable mask size and the appropriate Gaussian sigma value, various values are tested with sample images, and used later to interpolate the green plane for other test images in the proposed method. Afterward, red and blue plane interpolations are performed, and directional median processing is applied along the flat edge lines. This process can effectively suppress false colors introduced during the interpolation of color planes. The proposed method provides a reasonable visual quality and generally higher peak-signal-to-noise-ratio compared to existing methods.

Active contour methods are widely used for medical image segmentation. Using level set algorithms the applications of active contour methods have become flexible and convenient. This paper describes the evaluation of the performance of the active contour models using performance metrics and statistical analysis. We have implemented five different methods for segmenting the synovial region in arthritis affected ultrasound image. A comparative analysis between the methods of segmentation was performed and the best segmentation method was identified using similarity criteria, standard error, and F-test. For further analysis, classification of the segmentation techniques using support vector machine (SVM) classifier is performed to determine the absolute method for synovial region detection. With these results, localized region based active contour named Lankton method is defined to be the best segmentation method.

In the passive containment cooling system (PCCS), water distribution tests are essential to verify the water distribution devices performance. Without regular boundaries and homogeneous intensities, images of water film acquired from these tests are hard to be detected by conventional approaches. We propose an improved segmentation method to identify the water film areas from the complex background. Considering the gray distortion resulted from asymmetric illumination, the method combines the modified motion segmentation and optimal threshold method. Detection results show that this method is hardly affected by the illumination change, and also insensitive to noise.