Digital Image Processing Approach Research Articles

Comparison of a conventional image processing approach with an artificial neural network approach to three‐dimensionally trace multiple particles in dynamic x‐ray microtomography experiments under laboratory conditions

AbstractThis paper compares a classic digital image processing approach to trace particles in laboratory x‐ray micro‐computed tomography (µCT), which is based on a combination of random sample consensus (RANSAC) algorithm and least squares ellipse fitting (LSF), with an approach based on artificial neural networks (ANNs). In order to be able to perform the comparison, dynamic experiments were carried out in a laboratory microtomography facility. During active scans with a duration of 30–75 s several sedimentation experiments have been carried out with an exposure time of 0.13 s/projection. Through the movement of the particles during the scan curved motion artefacts in the reconstructed data occur, where the vertex marks the coordinate of the particle. It could be shown that both approaches enable the tracing of particles in laboratory x‐ray µCT with deviations from a manually evaluated result of exemplarily 1.25% for the conventional digital image processing (CDIP) and 0.48% for the ANN. It was found that ANNs are able to identify particle positions in non‐symmetrical motion artefacts, appearing around the first and last position of the particles in the scan, allowing an extension of the motion range of the particles that can be evaluated. Both methods have their advantages and disadvantages. Due to the high complexity and size as well as partly black box structures of neural networks, they are not 100% comprehensible whereas conventional image processing is 100% transparent and understandable. But because of the complexity of the tracing of particles, the CDIP code offers many parameters that can be set, which is why the application is therefore slightly more complex.

Comparison of a conventional image processing approach with an artificial neural network approach to three‐dimensionally trace multiple particles in dynamic x‐ray microtomography experiments under laboratory conditions

AbstractThis paper compares a classic digital image processing approach to trace particles in laboratory x‐ray micro‐computed tomography (µCT), which is based on a combination of random sample consensus (RANSAC) algorithm and least squares ellipse fitting (LSF), with an approach based on artificial neural networks (ANNs). In order to be able to perform the comparison, dynamic experiments were carried out in a laboratory microtomography facility. During active scans with a duration of 30–75 s several sedimentation experiments have been carried out with an exposure time of 0.13 s/projection. Through the movement of the particles during the scan curved motion artefacts in the reconstructed data occur, where the vertex marks the coordinate of the particle. It could be shown that both approaches enable the tracing of particles in laboratory x‐ray µCT with deviations from a manually evaluated result of exemplarily 1.25% for the conventional digital image processing (CDIP) and 0.48% for the ANN. It was found that ANNs are able to identify particle positions in non‐symmetrical motion artefacts, appearing around the first and last position of the particles in the scan, allowing an extension of the motion range of the particles that can be evaluated. Both methods have their advantages and disadvantages. Due to the high complexity and size as well as partly black box structures of neural networks, they are not 100% comprehensible whereas conventional image processing is 100% transparent and understandable. But because of the complexity of the tracing of particles, the CDIP code offers many parameters that can be set, which is why the application is therefore slightly more complex.

Morphometric Analysis of Chili Leaves with Yellow Curly Symptom Using Digital Image Processing Approach and Data Mining Algorithm

Gejala kuning keriting pada daun cabai umumnya disebabkan oleh infeksi Begomovirus. Daun tanaman terinfeksi tidak hanya mengalami perubahan warna sebagai indikator rusaknya klorofil tetapi juga mengalami perubahan morfologi bentuk. Penelitian ini bertujuan menguantifikasi gejala infeksi Begomovirus berdasarkan perubahan morfologi bentuk daun menggunakan pengolahan citra digital dan algoritma data mining yang akan memudahkan dalam pemantauan dan analisis perkembangan penyakit tanaman. Total 33 citra daun cabai rawit bergejala kuning keriting maupun tidak bergejala menjadi dataset penelitian ini. Citra daun cabai tersebut diolah dan diekstrak karakteristik bentuknya berupa circularity, aspect ratio, roundness, dan solidity menggunakan aplikasi Fiji-ImageJ. Selanjutnya dilakukan uji beda (uji-t), pengelompokan citra menggunakan algoritma Simple K-Means, dan evaluasi ketepatan hasil pengelompokan berdasarkan indeks ARI dan NMI. Hasil penelitian menunjukkan bahwa secara umum ada perbedaan bentuk yang nyata antara daun bergejala dengan daun tidak bergejala. Daun cabai rawit bergejala kuning keriting memiliki rata-rata nilai aspect ratio dan solidity yang lebih kecil dibandingkan daun cabai tidak bergejala, sebaliknya memiliki rata-rata nilai circularity dan roundness yang lebih besar dibandingkan daun cabai tidak bergejala. Evaluasi ketepatan pengelompokan sampel daun cabai rawit bergejala maupun tidak bergejala berdasarkan indeks ARI dan NMI menghasilkan nilai terbaik untuk pengelompokkan ke dalam dua kelompok.

INTEGRATING REMOTE SENSING DATA WITH HYDROTHERMAL ALTERATION MAPPING AND GEOCHEMICAL CHARACTERISTICS OF PRECAMBRIAN ROCKS ACROSS MAMBILLA PLATEAU, NORTHEASTERN, NIGERIA

Mineral exploration must include the mapping of hydrothermally altered regions, which are typically connected to mineralization. This study presents a strategy for integrating remote sensing data with the geochemical characteristics of Precambrian rocks from the Mambilla Plateau in northern Nigeria. Tectonically, the lineaments were fractures with the following orientations: NE-SW, NW-SE, ENE-WSW, and N-S. The distinct spectral reflectance and absorption properties of remotely sensed Landsat 8 data in the visible, near-infrared, shortwave-infrared, and thermal infrared portions of the electromagnetic spectrum were exploited in various digital image processing approaches. Mapping hydrothermal alteration minerals was done effectively and efficiently using band ratios (red/blue, SWIR 2/NIR, SWIR 1/NIR), spectral band combinations (Kaufmann and Sabins ratios), and principal component analysis. Ferric, phyllic, propylitic, and argillic iron all underwent changes as a result of hydrothermal alteration. Regional metamorphism also resulted in important alteration processes such as epidotization, sericitization, muscovititization, kaolinitization, and chloritization. Geochemically, on average, the Al2O3, SiO2, Fe2O3, CaO, MgO, K2O, Na2O, Li2O content are 12.04 %; 54.83%; 4.76%, 7.16%, 3.02%, 10.22%, 4.53%, 1.35% while MnO, P2O5 and TiO2 are less than 1%. The average base metal composition revealed that 60% of them had positive anomalies, which indicate mineralization, and 40% had negative anomalies. About 50% of the samples had positive anomalies, which indicated strong potential for harbouring mineralization based on the makeup of trace elements. Geochemical studies revealed the presence of lithium, titanium, and silica oxides in substantial concentrations. The local mineralization is governed by structural factors.

Reinforced concrete surface cracks length detection and length estimation by using digital image processing approach

Surface cracks are a common failure that occurs in reinforced concrete structures (RC). With the help of new technologies, access to crack properties should be easier and help the inspector to provide better results. However, most inspectors still prefer manual visual inspection approach, which leads to inconsistent results when investigating this flaw. Moreover, cracks with inconsistent shapes and irregularities are a difficult task for crack extraction, and inspectors overlook the details of cracks. Therefore, in this study, crack detection and thin crack appearance enhancement using various digital image processing algorithms (DSP) were proposed to improve the accuracy of crack length estimation. By using certain DSP on the captured crack images on RC, several algorithms were created and coded in MATLAB via the morphological approach to produce good quality of the original crack images. At the end of this study, the appearance of the thin crack was improved and helped to improve the estimate of the total length of the crack in pixels. The maximum percentage error between the estimated crack length was calculated and compared to the actual length and was 7.10 %. The surface crack detection algorithm and the has the potential as a helpful structural health monitoring (SHM) tool for crack inspection.

PERBANDINGAN KETAHANAN DAUN CABAI RAWIT DAN TOMAT TERHADAP SERANGAN Thrips sp. BERBASIS PENGOLAHAN CITRA DIGITAL

The presence of thrips pests that attack the leaves of cayenne pepper and tomato can cause high yield losses in both crops. Plant resistance to thrips is thought to be related to the pest's preference for plant leaves. The difference in the resistance level of the two plants to thrips attack needs to be studied based on a digital image processing approach as an accurate method of assessing plant damage. This study compares the resistance of cayenne pepper and tomato leaves to thrips attack based on digital image processing. The study was designed for feed testing with the option of using cayenne pepper and tomato leaves as treatments that were placed in pairs in a testing container. There were five pairs of tests for the first and second repetitions, respectively, and each pair of tests used five individual thrips. The results showed tomato leaves were relatively more resistant to thrips attack because they tended to be less preferred than cayenne pepper. The average area of thrips attack on cayenne pepper leaves for two repetitions of testing was higher than tomato. The assessment of the extent of thrips attack tomato and cayenne pepper leaves based on image processing has high validity, characterized by the accuracy and precision of the assessment results, which reach 80-100%. However, the reliability of the assessment results for two repetitions of testing is still relatively low, especially in assessing damage to cayenne pepper leaves.

Classification for Papaya Fruit Maturity Level With Convolutional Neural Network

Papaya California (Carica papaya L) is one of the agricultural commodities in the tropics and has a very big opportunity to develop in Indonesia as an agribusiness venture with quite promising prospects. So the quality of papaya fruit is determined by the level of maturity of the fruit, the hardness of the fruit, and its appearance. Papaya fruit undergoes a marked change in color during the ripening process, which indicates chemical changes in the fruit. The change in papaya color from green to yellow is due to the loss of chlorophyll. The papaya fruit is initially green during storage, then turns slightly yellow. The longer the storage color, the changes to mature the yellow. The process of classifying papaya fruit's ripeness level is usually done manually by business actors, that is, by simply looking at the color of the papaya with the normal eye. Based on the problems that exist in classifying the ripeness level of papaya fruit, in this research, we create a system that can be used to classify papaya fruit skin color using a digital image processing approach. The method used to classify the maturity level of papaya fruit is the Convolutional Neural Network (CNN) Architecture to classify the texture and color of the fruit. This study uses eight transfer learning architectures with 216 simulations with parameter constraints such as optimizer, learning rate, batch size, number of layers, epoch, and dense and can classify the ripeness level of the papaya fruit with a fairly high accuracy of 97%. Farmers use the results of the research in classifying papaya fruit to be harvested by differentiating the maturity level of the fruit more accurately and maintaining the quality of the papaya fruit.

Image Acquisition, Preprocessing and Classification of Citrus Fruit Diseases: A Systematic Literature Review

Different kinds of techniques are evaluated and analyzed for various classification models for the detection of diseases of citrus fruits. This paper aims to systematically review the papers that focus on the prediction, detection, and classification of citrus fruit diseases that have employed machine learning, deep learning, and statistical techniques. Additionally, this paper explores the present state of the art of the concept of image acquisition, digital image processing, feature extraction, and classification approaches, and each one is discussed separately. A total of 78 papers are selected after applying primary selection criteria, inclusion/exclusion criteria, and quality assessment criteria. We observe that the following are widely used in the selected studies: hyperspectral imaging systems for the image acquisition process, thresholding for image processing, support vector machine (SVM) models as machine learning (ML) models, convolutional neural network (CNN) architectures as deep learning models, principal component analysis (PCA) as a statistical model, and classification accuracy as evaluation parameters. Moreover, the color feature is the most popularly used feature for the RGB color space. From the review studies that performed comparative analyses, we find that the best techniques that outperformed other techniques in their respective categories are as follows: SVM among the ML methods, ANN among the neural network networks, CNN among the deep learning methods, and linear discriminant analysis (LDA) among the statistical techniques.This study concludes with meta-analysis, limitations, and future research directions.

Classification for Papaya Fruit Maturity Level with Convolutional Neural Network

Papaya California (Carica papaya L) is one of the agricultural commodities in the tropics and has a very big opportunity to develop in Indonesia as an agribusiness venture with quite promising prospects. So the quality of papaya fruit is determined by the level of maturity of the fruit, the hardness of the fruit, and its appearance. Papaya fruit undergoes a marked change in color during the ripening process, which indicates chemical changes in the fruit. The change in papaya color from green to yellow is due to the loss of chlorophyll. During storage, the papaya fruit is initially green, then turns slightly yellow. The longer the storage color, the changes to mature the yellow. The process of classifying papaya fruit's ripeness level is usually done manually by business actors, that is, by simply looking at the color of the papaya with the normal eye. Based on the problems that exist in classifying the ripeness level of papaya fruit, in this research, we create a system that can be used to classify papaya fruit skin color using a digital image processing approach. The method used to classify the maturity level of papaya fruit is the Convolutional Neural Network (CNN) Architecture to classify the texture and color of the fruit. This study uses eight transfer learning architectures with 216 simulations with parameter constraints such as optimizer, learning rate, batch size, number of layers, epoch, and dense and can classify the ripeness level of the papaya fruit with a fairly high accuracy of 97%. Farmers use the results of the research in classifying papaya fruit to be harvested by differentiating the maturity level of the fruit more accurately and maintaining the quality of the papaya fruit.

Role of artificial intelligence based image-processing techniques for port development activities in physical tidal models

ABSTRACT The Cochin Port physical tidal model available at Central Water and Power Research Station (CWPRS), Ministry of Jal Shakti, Government of India, is used to simulate the tidal conditions of real-time Cochin Port. Analysis of physical tidal model takes time and requires specialised knowledge in order to analyse and identify the velocity profile through manual instrumentation. The use of artificial intelligence (AI) based image processing techniques may reduce the analysis time and improve the accuracy of physical tidal model. An attempt was made to determine the velocity profile of Cochin Port model (scaled down model) at CWPRS through AI-based video image-processing analysis using MATLAB. Based on the analysis, it was observed that magnitude of horizontal velocity is varying for 1579 times in about 30.36 sec duration from 0.01156 m/s to 0.32986 m/s with a mean velocity magnitude of about 0.20357 m/s. The results indicated that digital videography-based image processing approach with the assistance of trained AI algorithms gives the précised velocity profile having 578 unique magnitude variations out of 1579 readings in 30.36 sec, which is far better than the manual instrumentation, and therefore, manual instrumentation may be substituted with the developed artificial intelligence (AI) based image processing technique.

Brain Tumor Segmentation in Multimodal MRI Using U-Net Layered Structure

The brain tumour is the mass where some tissues become old or damaged, but they do not die or not leave their space. Mainly brain tumour masses occur due to malignant masses. These tissues must die so that new tissues are allowed to be born and take their place. Tumour segmentation is a complex and time-taking problem due to the tumour’s size, shape, and appearance variation. Manually finding such masses in the brain by analyzing Magnetic Resonance Images (MRI) is a crucial task for experts and radiologists. Radiologists could not work for large volume images simultaneously, and many errors occurred due to overwhelming image analysis. The main objective of this research study is the segmentation of tumors in brain MRI images with the help of digital image processing and deep learning approaches. This research study proposed an automatic model for tumor segmentation in MRI images. The proposed model has a few significant steps, which first apply the pre-processing method for the whole dataset to convert Neuroimaging Informatics Technology Initiative (NIFTI) volumes into the 3D NumPy array. In the second step, the proposed model adopts U-Net deep learning segmentation algorithm with an improved layered structure and sets the updated parameters. In the third step, the proposed model uses state-of-the-art Medical Image Computing and Computer-Assisted Intervention (MICCAI) BRATS 2018 dataset with MRI modalities such as T1, T1Gd, T2, and Fluid-attenuated inversion recovery (FLAIR). Tumour types in MRI images are classified according to the tumour masses. Labelling of these masses carried by state-of-the-art approaches such that the first is enhancing tumour (label 4), edema (label 2), necrotic and non-enhancing tumour core (label 1), and the remaining region is label 0 such that edema (whole tumour), necrosis and active. The proposed model is evaluated and gets the Dice Coefficient (DSC) value for High-grade glioma (HGG) volumes for their test set-a, test set-b, and test set-c 0.9795, 0.9855 and 0.9793, respectively. DSC value for the Low-grade glioma (LGG) volumes for the test set is 0.9950, which shows the proposed model has achieved significant results in segmenting the tumour in MRI using deep learning approaches. The proposed model is fully automatic that can implement in clinics where human experts consume maximum time to identify the tumorous region of the brain MRI. The proposed model can help in a way it can proceed rapidly by treating the tumor segmentation in MRI.

Automated Detection and Characterization of Colon Cancer with Deep Convolutional Neural Networks.

Colon cancer is a momentous reason for illness and death in people. The conclusive diagnosis of colon cancer is made through histological examination. Convolutional neural networks are being used to analyze colon cancer via digital image processing with the introduction of whole-slide imaging. Accurate categorization of colon cancers is necessary for capable analysis. Our objective is to promote a system for detecting and classifying colon adenocarcinomas by applying a deep convolutional neural network (DCNN) model with some preprocessing techniques on digital histopathology images. It is a leading cause of cancer-related death, despite the fact that both traditional and modern methods are capable of comparing images that may encompass cancer regions of various sorts after looking at a significant number of colon cancer images. The fundamental problem for colon histopathologists is differentiating benign from malignant illnesses to having some complicated factors. A cancer diagnosis can be automated through artificial intelligence (AI), enabling us to appraise more patients in less time and at a decreased cost. Modern deep learning (MDL) and digital image processing (DIP) approaches are used to accomplish this. The results indicate that the proposed structure can accurately analyze cancer tissues to a maximum of 99.80%. By implementing this approach, medical practitioners will establish an automated and reliable system for detecting various forms of colon cancer. Moreover, CAD systems will be built in the near future to extract numerous aspects from colonoscopic images for use as a preprocessing module for colon cancer diagnosis.

Opening Doors Using Internet Of Things (IoT) Based Face Recognition

Face recognition is a digital image processing approach that uses face photographs as input to identify a person. Face recognition is important since the face is a person's primary means of identification because the shape of a person's face differs significantly, which is easy to do intuitively using the visual senses. Image processing, face detection, feature extraction, and classification are all aspects of the face recognition system, which seeks to determine whether the image obtained is a person's face stored in the database. Principles of operation If a human face appears in front of the camera, the system quickly executes a facial recognition procedure and compares the face to facial data kept on the website. If a face detected by the camera matches the face stored on the website, the solenoid will automatically be in the on position or the door will be open, and vice versa, if the face detected by the camera does not match, the solenoid will remain in the off position or the door will remain locked. This tool can be used to improve the security system on the door of a private room or a room that can only be accessed by certain people.

The Assessment of Coral Reefs Mapping Methodology: An Integrated Method Approach

Indonesian waters hold the world’s mega biodiversity of coral reefs. However, a range of anthropogenic pressures are threatening the coral reefs persistence. Since the early 20th century, remote sensing data has been assessed to map and monitor coral reefs. The reef habitats are monitored at various hierarchical spatial scales using integrated remote sensing and field data, but the level of detail and accuracy at a single point still questionable. Therefore, this study aims to assess the coral reefs methodology based on an integrated digital image processing approach. The method will employ a multi-pair and a single pair or an initial pair of Depth Invariant Index (DII) transformation bands, pixel-based Isodata and K-Means algorithm, and supervised classification method based on maximum likelihood and nearest neighbor algorithms. Object-based classification images, training areas, and data references were supported by previous research. The findings indicate that the maximum likelihood algorithm is better to apply for supervised classification for a single transformation band, while the K-Means algorithm is better for pixel-based classification since better accuracy can be obtained. However, various remote sensing data, band combinations, and clusters may affect the difference in results.

Study on the micro-mechanism and structure of unsaturated polyester resin modified concrete for bridge deck pavement

To solve the problems of durability reduction and service-life shortening of bridge deck pavement concrete, this study investigated the effect of unsaturated polyester resin (UPR) as a modifier on the cement hydration and the microstructure morphology of pore, crack and interface transition zone (ITZ) of cement concrete. The enhancement mechanism of UPR incorporation into bridge deck pavement concrete was determined with the aid of meso- and micro-structure analysis technology and digital image processing approach. The results indicated that the incorporation of UPR-emulsion might have a significant retarding effect on the cement hydration process while greatly enhanced the compactness and anti-permeability of the concrete. In addition, the addition of UPR emulsion could effectively delay crack initiation and restrain the propagation of crack during consolidation process, and reduce the zone area and materialize intense densification and homogenization in terms of ITZ. In comprehensive consideration of the characteristics of pore structure and micro-crack, the introduction of 3% UPR-emulsion could be recommended to achieve the performance reinforcement of deck pavement concrete to a certain degree.

Deteksi Api pada Video dengan Gaussian Mixture Model Untuk Deteksi Gerakan Dan Segmentasi Warna Api Dalam Ruang Warna YCbCr

Fire is a disaster that can endanger lives and cause property loss. The solution to detect fire that is commonly used today is to use a sensor. Fire sensors can be used together with surveillance cameras (CCTV) which are now being installed in many office buildings. This study tries to build a model for detecting fire in video with a digital image processing approach using the Gaussian Mixture Model for motion detection and fire color segmentation in the YCbCr color space. The model is then tested with metrics for accuracy, precision, recall, and processing speed. The dataset used is in the form of videos with small, medium, large fire sizes, and videos that only have fire-like objects. The test results show that the algorithm is able to detect fire when the size of the fire is not too small or the position of the fire is close enough to the camera. For videos with a resolution of 800x600 and a framerate of 30 fps, it can achieve 66.89% accuracy, 73.77% precision, and 66.66% recall. The performance during the day is relatively better than at night. Algorithm processing speed is too slow to be implemented in real-time

Turbidity Level Prediction Based on Suspended Particle Counting Through Image Processing Approach

Monitoring of pollutant concentrations in surface water becomes a concern, considering the utilization of surface water as the raw water for drinking water treatment plants (WTP). The fluctuation of pollutant concentrations in surface water can affect the performance of WTP. This research was conducted to assess the potential for turbidity level prediction based on the calculation of the number and surface area of suspended particles through a digital image processing approach. Measurements of the amount and surface area were carried out in the form of laboratory-scale experiments using the open source software ImageJ 1.46r. The algorithm in ImageJ can convert pixels into a number “value” and surface area through a series of digital image processing steps, henceforth compared with the existing measurement method. The results showed that there was a strong correlation between the number of particles and the concentration of formazine suspension (r = 0.9821), but does not apply to the surface area. Referring to the results of laboratory experiments, it can be concluded that the approach to measure the number of suspended particles can be the basis for predicting the turbidity level in the turbidity range 100-800 NTU, but does not apply to the turbidity range 0.02-20 NTU.

Local Features-Based Watermarking for Image Security in Social Media

The last decade shows an explosion of using social media, which raises several challenges related to the security of personal files including images. These challenges include modifying, illegal copying, identity fraud, copyright protection and ownership of images. Traditional digital watermarking techniques embed digital information inside another digital information without affecting the visual quality for security purposes. In this paper, we propose a hybrid digital watermarking and image processing approach to improve the image security level. Specifically, variants of the widely used Least-Significant Bit (LSB) watermarking technique are merged with a blob detection algorithm to embed information into the boundary pixels of the largest blob of a digital image. The proposed algorithms are tested using several experiments and techniques, which are followed by uploading the watermarked images into a social media site to evaluate the probability of extracting the embedding watermarks. The results show that the proposed approaches outperform the traditional LSB algorithm in terms of time, evaluation criteria and the percentage of pixels that have changed.

EVALUATING THE PERFORMANCE OF A SEMI-AUTOMATIC APPLE FRUIT DETECTION IN A HIGH-DENSITY ORCHARD SYSTEM USING LOW-COST DIGITAL RGB IMAGING SENSOR

Abstract: This study investigates the potential use of close-range and low-cost terrestrial RGB imaging sensor for fruit detection in a high-density apple orchard of Fuji Suprema apple fruits (Malus domestica Borkh). The study area is a typical orchard located in a small holder farm in Santa Catarina’s Southern plateau (Brazil). Small holder farms in that state are responsible for more than 50% of Brazil’s apple fruit production. Traditional digital image processing approaches such as RGB color space conversion (e.g., rgb, HSV, CIE L*a*b*, OHTA[I 1 , I 2 , I 3 ]) were applied over several terrestrial RGB images to highlight information presented in the original dataset. Band combinations (e.g., rgb-r, HSV-h, Lab-a, I” 2 , I” 3 ) were also generated as additional parameters (C1, C2 and C3) for the fruit detection. After, optimal image binarization and segmentation, parameters were chosen to detect the fruits efficiently and the results were compared to both visual and in-situ fruit counting. Results show that some bands and combinations allowed hits above 75%, of which the following variables stood out as good predictors: rgb-r, Lab-a, I” 2 , I” 3 , and the combinations C2 and C3. The best band combination resulted from the use of Lab-a band and have identical results of commission, omission, and accuracy, being 5%, 25% and 75%, respectively. Fruit detection rate for Lab-a showed a 0.73 coefficient of determination (R2), and fruit recognition accuracy rate showed 0.96 R2. The proposed approach provides results with great applicability for small holder farms and may support local harvest prediction.

Keratoconus Detection Using the Fusion Features of Anterior and Lateral Segment Photographed Images

Keratoconus (KC) is a type of eye disease that involves the thinning of the corneal layer and a change in the semispherical shape of the cornea to a bulging cone shape when viewed laterally. KC is difficult to detect in the early stages of the disease, as the patient does not feel any pain. Hence, the development of a KC detection (KD) method using a digital image processing approach is needed for the early detection of KC so that physicians can provide patients with the subsequent treatment sooner. The objective of this study was to develop a method of KD using a camera from a smart device to capture anterior and lateral segment photographed eye images (A&LSPIs). A total of 280 images comprising 140 KC and 140 normal A&LSPIs were used in this study, and all images were validated by a qualified expert. First, the corneal area of both image views was segmented so the geometric features could be extracted using the automated modified active contour model and the semiautomated spline function for the anterior and lateral images, respectively. Then, the features were selected using infinite latent feature selection (ILFS) by identifying the feature rankings based on the graph weighting that was automatically learned by the probabilistic latent semantic analysis (PLSA). The results showed that the all-combined features, where the proposed and improved features were successfully top ranked, had 96.05% accuracy, 98.41% sensitivity and 93.65% specificity with the RFn=50 classifier, outperforming the 7-NNMaha and SVMRBF classifiers. In conclusion, this study successfully developed a keratoconus detection system based on fusion features using a digital image processing approach for A&LSPIs captured with a smartphone camera.