Search

Most people who get sick around the world have some kind of skin condition. Despite its frequency, it is very challenging to identify and requires substantial domain experience. This study explains how to identify various types of these disorders. We use a two-stage strategy that efficiently integrates image processing and deep learning based on molecular diagnostic histological features to reliably detect the disease. During the initial stage of the process, the image of the skin condition is processed in a variety of different ways before "feature extraction" is performed. In the second stage of this process, Machine Learning (ML) algorithms are used to diagnose diseases based on histological characteristics discovered through skin examination. These characteristics can be found by analysing the skin. After being trained and evaluated, the system could correctly name up to 95% of the six diseases. [ABSTRACT FROM AUTHOR]

So far, the identification of 16 types of tree damage is still following the guidelines stated in the Forest Health Monitoring method. The types of tree damage can be recognized by human vision, as well as by computers. Computer vision allows computers to identify things that humans can remember. In this case, the study can be realized with computer vision to make work easier. This study aimed to identify 16 types of damage in Forest Health Monitoring using image data or photos with computer vision. The stages of this research are image acquisition (image acquisition), image processing (image preprocessing), and feature extraction (feature extraction). The results showed that computer vision could identify images in JPG/JPEG (Joint Photographic Experts) format, assisted by the Convolutional Neural Network (CNN) algorithm. The percentage value of success using the CNN algorithm reaches 99.06%, with an error detection of 0.94%. However, several classes were incorrectly identified, indicating that the dataset needs improvement. Thus, it can be concluded that identifying 16 types of damage using image data or photos with computer vision has been successfully carried out. [ABSTRACT FROM AUTHOR]

Counterfeiting of quality rice was rife in Indonesia. This research was conducted to develop technology to identify differences in premium and non-premium rice quality based on pre-existing digital images. Artificial neural networks and digital image processing methods to identify premium and medium (non-premium) rice quality were applied in this research. Statistical analysis of this study used the SPSS program. This research is observation-type research. This research design uses an artificial neural network with uses 3 layers, namely the results of shape feature extraction on the metric, eccentricity, area, and perimeter parameters as input or input layers, hidden or hidden layers, and premium rice and non-premium (medium) rice as output or output layers. This research uses 52 images as training and 20 images as testing. The obtained image was taken at a distance of 25 cm. This research showed that the results of training using artificial neural networks (ANN) on 52 images obtained an accuracy of 92%. The test results using 20 images obtained 95% accuracy, 63.33% sensitivity, and 10% specificity. Based on statistical analysis using the Mann-Whitney test, it obtained the asymph value. Sig (2-tailed) < 0.05 indicates the difference between premium and non-premium rice using metric, eccentricity, perimeter, and area parameters. [ABSTRACT FROM AUTHOR]

Recognizing objects is an important task in many real-world applications, including security systems, surveillance, and robots, making it a hot topic in computer vision research. Because of the progress made in deep learning, real-time object identification systems have been created. These systems can instantly identify things in photos and movies. A major focus of computer vision research, object detection has found practical applications in areas as diverse as autonomous vehicles, robots, security cameras, and people counting. Although deep neural networks have powerful feature representation capabilities in image processing and are often used as feature extraction modules in object detection, their introduction has changed conventional techniques of object identification and detection. Models trained using deep learning may be used as classifiers or regression tools, and they don't need any further human input. This suggests that object identification is a promising area for deep learning research. In order to recognize items in a picture, object detection first needs to pinpoint their exact position (object localization). [ABSTRACT FROM AUTHOR]

The development of aberrant cell proliferation in the lungs is a problematic condition that has the potential to result in death. On the list of diseases that most frequently result in mortality, lung cancer takes first place. The early stages of lung cancer are notoriously difficult to diagnose due to the fact that cancer cells with dimensions less than very small are notoriously difficult to spot by imaging. If the cell abnormalities are discovered in the early stages, it will be possible to begin therapy sooner, which will result in an improved chance of the patient surviving the illness. Several different image processing strategies can be utilized in the diagnostic phase of patient care to help spot signs of disease. In this paper, classification of Lung Cancer from chest X-ray images has been done using optimized Convolutional Neural Network (OCNN) and Ant Lion Optimization (ALO) algorithm. In pre-processing step, the contrast of all images are enhanced using Histogram Equalization (HE) method and the noises are removed from all images using Median Filtering. After the pre-processing step, feature extraction is performed using Gray Level Spatial Dependence (GLSD) to extract the statistical features. The feature vector is then trained and classified using OCNN-ALO algorithm. The ALO algorithm is used to optimize the hyper parameters of CNN layers. It classifies the lung images into normal and lung tumor affected. Performance results have indicated that OCNN-ALO attains the superior performance with 95.15% accuracy, 85.43% sensitivity, 93.4% specificity and 76.43% F1-score. [ABSTRACT FROM AUTHOR]

Diabetic patients persist at risk of acquiring Proliferative Diabetic Retinopathy, a retinal condition (PDR). development of neovascularization, a disease in which aberrant blood vessels form on retina, one of key hallmarks of PDR. If not diagnosed & treated early enough, aforementioned illness preserve lead via blindness. Various image processing algorithms considering detecting neovascularization in fundus images have been proposed in a number of studies. Neovascularization, on other hand, difficult via identify due via its erratic growth pattern & tiny size. As a result of their ability via execute automatic feature extraction on objects among complex properties, deep learning approaches persist becoming more common in neovascularization recognition. A fore mentioned research proposes a transfer learning-based approach considering detecting neovascularization. We show certain suggested technique outperforms each individual in detecting neovascularization. [ABSTRACT FROM AUTHOR]

The initial process of developing a Javanese script transliteration system to other scripts using a character recognition approach requires training data in the form of script images of all possible forms. The source of the dataset are script images from a book written in Javanese and then processed using image processing approach. The captured images were then grouped into their respective classes. The study starts with pre-processing the document images that includes the sub-processes of binarization, inverse, filtering, and followed by script segmentation using the projection profile method. Each script image is then processed in the feature extraction steps using the Intensity of Character or IoC algorithm. The feature data of each script image is then grouped using the K-Means clustering algorithm. The data was taken from the scan results of Hamong Tani's book on pages 2 and 59. After pre-processed and segmented images, 597 images of Javanese script were obtained. Using the IoC 3x3 feature, and the number of groups determined by 65 classes, the silhouette index value of the grouping results was found to be 0.5060. After calculating the ground truth value, it was found that the accuracy of the results was 86%. It can be concluded that the steps taken in this research can be used as a model in the process of providing a Javanese script images dataset. [ABSTRACT FROM AUTHOR]

All traditional surgical procedures cause a lot of harm to the patient, while the biggest advantage of minimally invasive surgery is that the wound is small, the pain is small, and the recovery is fast. Recently, minimally invasive surgery has been widely adopted in the medical field and has been widely used as an important technique. Because of the high degree of coordination between the hand and the eye, there is a high demand for the operating ability of the physician. The images obtained by endoscopy do not have sufficient depth and are difficult to localize. The key to improving the visualization of such surgeries lies in optimizing image processing methods and feature point extraction algorithms that augment SLAM (Simultaneous Localization and Map Construction) techniques. This paper reviews several methods, such as the Contrast Constrained Adaptive Histogram Equalization (CLAHE) method, which solves the illumination problem by optimizing the grayscale transformation. Other methods such as wavelet transform, homomorphic filtering, and multiscale bilaterally weighted retina are also discussed for their image enhancement capabilities. For feature point extraction, AKAZE-ORB and SuperPoint algorithms combined with k-means algorithms are emphasized for their advancement in resolving uneven illumination and ensuring accurate extraction. [ABSTRACT FROM AUTHOR]

Gastritis is a disease that occurs in the stomach and can be dangerous if not treated quickly. Gastritis can affect anyone, young or old. Therefore, a system is needed to identify early gastritis identification so that it can help the community to find out early on their gastritis. The identification method used in this study is the Radial Basis Function Neural Network with several image processing techniques. Tongue image is used as input image for image processing. Prior to the initial identification, the image preprocessing process is carried out, namely cropping, resizing, brightness, image feature extraction using color extraction using Hue Saturation Value (HSV). The dataset used in this study are limited to 150 tongue images and resulted in the ability to identify early gastritis with an accuracy of 93.3% for 50 testing images. [ABSTRACT FROM AUTHOR]

The diverse motifs of Lampung Batik require an automated identification process using digital image processing technology. This research aims to identify Lampung Batik motifs by utilizing computer technology, particularly in digital image processing, to assist in recognizing these motifs. In this era of digitization, there are various object recognition methods, one of which is image recognition that utilizes GLCM feature extraction to obtain the characteristics of Lampung Batik motifs. After the feature extraction process on Lampung Batik images is completed, this data is used for classification. This study employs the K-Nearest Neighbor and Naïve Bayes classification methods, with a total dataset consisting of 300 images, comprising 150 images of "siger" motifs and 150 images of "sembagi" motifs. In testing with the K-Nearest Neighbor method, the highest accuracy is achieved at K = 3, approximately 77% for the training data and 81% for the testing data. Meanwhile, in testing with the Naïve Bayes method, an accuracy of around 96% is attained for both the training and testing data. [ABSTRACT FROM AUTHOR]

Skin diseases are among the most common and widespread diseases in the world, with people developing them as a result of inheritance, ageing, hormones, allergic reactions, sun or toxic chemical exposure, and environmental factors. Despite its prevalence, diagnosis is extremely difficult due to subtle differences in skin tone, hue, and hair presence. Many people ignore the effects of skin disease when it is in its early stages. Skin diseases are currently diagnosed using a biopsy procedure, which is then examined, and medications are administered manually by physicians. To avoid manual inspection and provide accurate results in a timely manner, this paper proposes a skin disease detection method based on image processing, Python, and the Yolov3 tool. The patient can provide an image of the infected skin area as an input to the system. On this image, image processing techniques are used, and feature values are extracted and detected, as well as the disease being analyzed. The proposed system is especially useful in areas where dermatologists are scarce. The paper describes a method for identifying four types of skin disease: acne, melanoma, blisters, and cold sores. The primary goal of this method is to achieve the highest possible level of skin disease prediction accuracy. [ABSTRACT FROM AUTHOR]

Computed Tomography (CT) is far and wide utilized to make a diagnosis and access thoracic diseases. The enhanced resolution of CT examination has resulted in a considerable investigation of statistics for analysis. Computerizing the scrutiny of such facts is consequently necessitate and fashioned a hastily emergent research region in medical imaging. The finding of thoracic diseases by means of image processing directs to a pre-processing step identified as "Lung segmentation" which portrays a wide range of techniques starts with simple Thresholding and numerous image processing elements are incorporated to progress segmentation, precision and heftiness. In image processing, techniques like image pre-processing, segmentation and feature extraction have been thrashed out in detail. This paper suggestions investigation of literature on computer examination of the lungs in CT scans and statements the Preprocessing ideas, segmentation of a choice of pulmonary arrangements, and Feature Extraction intended at recognition and categorization of chest abnormalities. As well as, research developments and disputes are recognized and instructions for further examinations are discussed. [ABSTRACT FROM AUTHOR]

The advancement made in image processing techniques has gained much attraction among the researchers in recent years. Object detection and classification are the recent topics that annoys the researchers for developing new techniques in this field. Conventional object detection models degrade the performance due to complexity in feature selection process. Thus, the rapid development in deep learning is combined with object detectors to resolve the addressed problems. In present research, we introduce detecting of an object and classification model using improved segmentation and classifier. The aim of the study is to detect the weapon in human object and then classified onto its relevant classes for upcoming testing models. The proposed model comprises of five phases where each phase serves as input to its predecessor output. Initially, the thermal images are collected from a public repository and then preprocessed using median filtering which eliminates the irrelevant noises by replacing median values to its neighboring pixels. Further, Adaptive Histogram Equalization (AHE) is used for improving the finest quality of an image. The preprocessed image is further fed into segmentation model. Here, OTSUs, one of the successful models is employed for deriving edges preserved segmented image. By computing SURF, GLCM and matching features on segmented images, the relevant features are extracted. These extracted features are then given to DCNNs which develops a layering stack and creates training and testing model. Based on satisfied requirements of training features, testing images are classified to its relevant classes. The performance assessment of classification accuracy dictates 84% of efficiency than the prior works which is implemented using Fuzzy Segmentation. [ABSTRACT FROM AUTHOR]

Ground Penetrating Radar (GPR) is a non-destructive test used as imaging tool for exploration of shallow subsurface such locating the buried infrastructures. Due to the existence of various subsurface material and environmental noise, such as bricks and tree branch, it is a challenging task to interpret the GPR data into a meaningful information. Thus, this project proposes the hyperbolic recognition of buried pipes in GPR images in the presence of scattering objects. In this framework, the GPR images were firstly subjected to image pre-processing. Then, the GPR images were decomposed using Discrete Wavelet Transform (DWT) to analyze the texture analysis of hyperbola signature. Then, the approximation subband of DWT were extracted and used as features to recognize the hyperbolic of buried pipes and scattering objects presence in GPR images. A series of experiment has been conducted on GPR data collection at Agency Nuclear Malaysia. Based on the results obtained, the average recognition rate of extracted approximation subband of DWT features using k-NN classifier is 99.75%, thus shows a promising results in recognizing the buried pipes in the presence of scattering objects. [ABSTRACT FROM AUTHOR]

Breast cancer is a vital problem for women across the globe. Mostly breast cancer is been identified at the later stages. In existing approach, the breast cancers are inspected by doctors bur accuracy and time are major concerns. To overcome the existing issue, we proposed breast cancer detection technique using computer vision invoking image processing and deep learning technique. The trained model is been built using convolutional neural network (CNN). Feature extraction is performed using principal component analysis (PCA). For experimental results, we use google colab with python script to analyze and recommend outcomes such as benign, malignant. [ABSTRACT FROM AUTHOR]