Median Filtering Technique Research Articles

Roughness measurement results evaluation of 6082 aluminium alloy specimens after fatigue bending tests

In this paper, the topography of 6082 aluminium alloy specimens after fatigue bending tests was studied with a comprehensive evaluation of measurement noise caused by vibration. Roughness results were acquired by contactless Focus Variation Microscopy (FVM). Studied data were pre-processed, removing the non-measured points and outliers with regular methods, respectively, and high-frequency noise was considered. The variations in ISO 25178 roughness parameters were studied. Based on the previous studies, it was found that surfaces after fatigue bending tests can be difficult to consider when analyzing the measurement noise in a selected bandwidth. Some advantages of profile data extraction in selected directions, like horizontal, vertical or crack, were found deficient, even in studies by various functions, like autocorrelation, power spectral density, or texture direction ratio. When noise suppression methods depend on the details studied, boundary areas were extracted to compare and highlight the presence of high-frequency data characteristics. The proposed method was validated when contrasting standardised Gaussian or median filtering techniques with the spline filtering approach. A proper filter for the reduction of vibrational noise from the results of FVM topography measurements was suggested based on the proposed procedure. Finally, it was proposed how use the new method for reducing errors caused by high-frequency measurement noise in the surface topography of specimens after fatigue bending tests.

Transformer-based decoder of melanoma classification using hand-crafted texture feature fusion and Gray Wolf Optimization algorithm

Melanoma is a type of skin cancer that poses significant health risks and requires early detection for effective treatment. This study proposing a novel approach that integrates a transformer-based model with hand-crafted texture features and Gray Wolf Optimization, aiming to enhance efficiency of melanoma classification. Preprocessing involves standardizing image dimensions and enhancing image quality through median filtering techniques. Texture features, including GLCM and LBP, are extracted to capture spatial patterns indicative of melanoma. The GWO algorithm is applied to select the most discriminative features. A transformer-based decoder is then employed for classification, leveraging attention mechanisms to capture contextual dependencies. The experimental validation on the HAM10000 dataset and ISIC2019 dataset showcases the effectiveness of the proposed methodology. The transformer-based model, integrated with hand-crafted texture features and guided by Gray Wolf Optimization, achieves outstanding results. The results showed that the proposed method performed well in melanoma detection tasks, achieving an accuracy and F1-score of 99.54% and 99.11% on the HAM10000 dataset, and an accuracy of 99.47%, and F1-score of 99.25% on the ISIC2019 dataset.• We use the concepts of LBP and GLCM to extract features from the skin lesion images.• The Gray Wolf Optimization (GWO) algorithm is employed for feature selection.• A decoder based on Transformers is utilized for melanoma classification.

An efficient hybrid bert model for brain tumor classification

The central nervous system can develop complex and deadly neoplastic growths called brain tumors. Despite being relatively uncommon in comparison to other cancers, brain tumors pose particular challenges because of their delicate anatomical placement and interactions with critical brain regions. The data are taken from TCIA (The Cancer Image Archive) and Kaggle Datasets. Images are first pre-processed using amplified median filter techniques. The pre-processed images are then segmented using the Grabcut method. Feature extraction is extracted using the Shape, ABCD rule, and GLCM are the features were retrieved. The MRI images are then classified into several classes using the Bi-directional Encoder Representations from Transformers-Bidirectional Long Short Term Memory (BERT-Bi-LSTM) model. Kaggle and TICA datasets are used to simulate the proposed approach, and the results are evaluated in terms of F1-score, recall, precision and accuracy. The proposed model shows improved brain tumour identification and classification. To evaluate the expected technique’s efficacy, a thorough comparison of the current techniques with preceding methods is made. The trial results showed that an efficient hybrid bert model for brain tumor classification suggested strategy provided precision of 98.65%, F1-score of 98.25%, recall of 99.25%, and accuracy of 99.75%.

Mammogram Images Using Noise Removal of Filtering Techniques

Historically, image denoising methods were proposed for reducing salt and pepper noise in mammography images through enhanced filtering techniques. The Mammogram Image Analysis Society (MIAS) benchmark database for DICOM visuals is used to detect benign and malignant breast cancer. Thus, mean and median filtering techniques, as well as spatial filtering methods, are used to implement medical image analysis with the modalities of low and high radiation for medical images. Similar to these experimental and suggested techniques for removing salt and pepper noise, these statistical techniques assess MSE, SNR, and PSNR data for enhanced quality of signal in the image.

Analysis and Comparison of Image Enhancement Technique for Improving PSNR of Lung Images by Median Filtering over Histogram Equalization Technique

Aim: The main goal of this project is image enhancement to improve interpretability or perception of information in images for human viewers and also to provide better input for other automated image processing techniques. Materials and Methods: In this research different sources of lung images collected from Kaggle website were used. Samples were considered as (N=30) for median filtering and (N=30) for novel histogram equalization technique with total sample size calculated using clinical.com. As a result the total number of sample was calculated to be 60.Using SPSS Software and a standard data set,the PSNR was obtained. Both median filter and novel histogram technique image enhancement were implemented on Lung images through Matlab coding and also extracting PSNR values of each image. Then through SPSS software comparison and analysis has been made Results: In an image enhancement of the image processing pathway, novel histogram equalization technique shows the best performance by removing noise to improve PSNR of lung images than median filtering. Comparison of PSNR values are done by independent sample test using IBM-SPSS software. There is a statistical difference between histogram technique and median filtering. The novel histogram equalization technique showed higher results of PSNR (69.6557dB) with (p=0.04) in comparison with median filtering (37.6427dB). Conclusion: Histogram equalization image enhancement technique provides high PSNR values for different sources of lung images than median filtering Technique.

Analysis and Comparison of Image Enhancement Technique for Improving PSNR of Lung Images by Wiener Filtering over Histogram Equalization Technique

Aim: The aim of this study is to compare and analyze wiener filter algorithm for lung image enhancement over novel histogram equalization technique. Materials and Methods:In this research, different sources of lung images collected from the kaggle website were used. Samples were considered as (N=30) for median filtering and (N=30) for novel histogram equalization technique with total sample size calculated using clinical. com. As a result the total number of samples was calculated to be 60. Using SPSS Software and a standard data set, the PSNR was obtained. Both median filter and novel histogram equalization technique image enhancement were implemented on lung images through Matlab coding and also extracting PSNR values of each image. Then through SPSS software comparison and analysis has been made. Result:. The novel histogram equalization technique showed higher results of PSNR (67.2076dB) in comparison with wiener filtering (37.1558dB). It is observed that histogram algorithm performed better than the Wiener filter (p=0.04) by performing an independent sample t-test. Conclusion: Within this research study the histogram equalization image enhancement technique has greater PSNR value of lung images than in wiener filtering technique.

Diabetic Retinopathy Detection: A Blockchain and African Vulture Optimization Algorithm-Based Deep Learning Framework

Blockchain technology has gained immense momentum in the present era of information and digitalization and is likely to gain extreme popularity among the next generation, with diversified applications that spread far beyond cryptocurrencies and bitcoin. The application of blockchain technology is prominently observed in various spheres of social life, such as government administration, industries, healthcare, finance, and various other domains. In healthcare, the role of blockchain technology can be visualized in data-sharing, allowing users to choose specific data and control data access based on user type, which are extremely important for the maintenance of Electronic Health Records (EHRs). Machine learning and blockchain are two distinct technical fields: machine learning deals with data analysis and prediction, whereas blockchain emphasizes maintaining data security. The amalgamation of these two concepts can achieve prediction results from authentic datasets without compromising integrity. Such predictions have the additional advantage of enhanced trust in comparison to the application of machine learning algorithms alone. In this paper, we focused on data pertinent to diabetic retinopathy disease and its prediction. Diabetic retinopathy is a chronic disease caused by diabetes and leads to complete blindness. The disease requires early diagnosis to reduce the chances of vision loss. The dataset used is a publicly available dataset collected from the IEEE data port. The data were pre-processed using the median filtering technique and lesion segmentation was performed on the image data. These data were further subjected to the Taylor African Vulture Optimization (AVO) algorithm for hyper-parameter tuning, and then the most significant features were fed into the SqueezeNet classifier, which predicted the occurrence of diabetic retinopathy (DR) disease. The final output was saved in the blockchain architecture, which was accessed by the EHR manager, ensuring authorized access to the prediction results and related patient information. The results of the classifier were compared with those of earlier research, which demonstrated that the proposed model is superior to other models when measured by the following metrics: accuracy (94.2%), sensitivity (94.8%), and specificity (93.4%).

A Convolution Neural Network based MRI breast mass diagnosis using Zernike moments

The most important aspect of women's death worldwide is currently breast cancer. Advance engineering techniques of classifying the images and machine learning has generally been utilized for classifying the breast images. The association of computerized picture characterization permits the physicians and doctors second opinion hence saving their time. Median filter technique is used for image de-noise. To overcome the issues of the conventional imaging techniques (Mammography, Ultrasound, MRI) used in machine learning supervised algorithms such as small data size, feature introspection, nowadays deep learning techniques are used to make performance getting better as you feed them more information. In this study we develop a technique based on the CNN (LeNet-5) for the diagnosis of tumors in Zernike moments of breast cancer. The performance of the new CNN architecture is 83.33% 88.2 percent sensitivity and 76.92% accurateness, 83.3% sensitivity and 62.5% malignant growth accuracy.

An Optimized Architecture for COVID‑19 Prediction Using Chest X‑Ray Images

In modern times, a disease known as COVID-19 that is highly contagious is continuing to have a profoundly negative influence on the people of the entire world. The fundamental purpose of the model that has been proposed is to improve its predictive capabilities while also providing an effective model for predicting COVID-19 that has a robust diagnostic. Image scaling and noise reduction are two examples of the types of pre-processing techniques that are used at the very first step. The adoption of picture scaling and median filtering techniques, both of which work to enhance the quality of the input data in preparation for further processing steps, allows this goal to be accomplished. Several distinct data augmentation strategies, like flipping and rotation, are utilized to improve the model's performance on a limited dataset and assist it in better comprehending the differences present in the training data. In this article, we will provide a unique Optimized Architecture for COVID-19 Prediction (OACP) model to classify COVID-19 situations as either positive or negative effectively. Using CXR pictures, this novel method, based on a tunable deep learning technique called DenseNet, may predict the presence of COVID-19-positive patients. Based on the findings, it was determined that the proposed model utilized achieved better outcomes, with an accuracy of 98%.

Classification of Multi-view Digital Mammogram Images Using SMO-WkNN

Breast cancer (BCa) is a leading cause of death in the female population across the globe. Approximately 2.3 million new BCa cases are recorded globally in females, overtaking lung cancer as the most prevalent form of cancer to be diagnosed. However, the mortality rates for cervical and BCa are significantly higher in developing nations than in developed countries. Early diagnosis is the only option to minimize the risks of BCa. Deep learning (DL)-based models have performed well in image processing in recent years, particularly convolutional neural network (CNN). Hence, this research proposes a DL-based CNN model to diagnose BCa from digitized mammogram images. The main objective of this research is to develop an accurate and efficient early diagnosis model for BCa detection. This proposed model is a multi-view-based computer-aided diagnosis (CAD) model, which performs the diagnosis of BCa on multi-views of mammogram images like medio-lateral-oblique (MLO) and cranio-caudal (CC). The digital mammogram images are collected from the digital database for screening mammography (DDSM) dataset. In preprocessing, median filter and contrast limited adaptive histogram equalization (CLAHE) techniques are utilized for image enhancement. After preprocessing, the segmentation is performed using the region growing (RG) algorithm. The feature extraction is carried out from the segmented images using a pyramidal histogram of oriented gradients (PHOG) and the AlextNet model. Finally, the classification is performed using the weighted k-nearest neighbor (WkNN) optimized with sequential minimal optimization (SMO). The classified images are evaluated based on accuracy, recall, precision, specificity, f1-score, and mathews correlation coefficient (MCC). Additionally, the false positive and error rates are evaluated. The proposed model obtained 98.57% accuracy, 98.61% recall, 99.25% specificity, 98.63% precision, 97.93% f1-score, 96.26% MCC, 0.0143 error rate, and 0.0075 false positive rate (FPR). Compared to the existing models, the research model has obtained better performances and outperformed the other models.

Computer-aided Diagnosis applied to MRI images of Brain Tumor using Spatial Fuzzy Level Set and ANN Classifier

The most vital organs in the human body are the brain, heart, and lungs. Because the brain controls and coordinates the operations of all other organs, normal brain function is vital. Brain tumour is a mass of tissues which interrupts the normal functioning of the brain, if left untreated will lead to the death of the subject. The classification of multiclass brain tumours using spatial fuzzy based level sets and artificial neural network (ANN) techniques is proposed in this paper. In the proposed method, images are preprocessed using Median Filtering technique, the boundaries of the Brain Tumor are obtained using Spatial Fuzzy based Level Set method, features are extracted using Gabor Wavelet and Gray-Level Run Length Matrix (GLRLM) methods. Finally ANN technique is used for the classification of the image into Normal or Benign Tumor or Malignant Tumor. The proposed method was implemented in the MATLAB working platform and achieved classification accuracy of 94%, which is significant compared to state-of-the-art classification techniques. Thus, the proposed method assist in differentiating between benign and malignant brain tumours, enabling doctors to provide adequate treatment.

Using Drones for Thermal Imaging Photography and Building 3D Images to Analyze the Defects of Solar Modules

In this research, drones were used to capture thermal images and detect different types of failure of solar modules, and MATLAB® image analysis was also conducted to evaluate the health of the solar modules. The processes included image acquisition and transmission by drone, grayscale conversion, filtering, 3D image construction, and analysis. The analyzed targets were the solar modules installed on buildings. The results showed that the employment of drones to monitor solar module farms could significantly improve inspection efficiency. Moreover, by combining the mean and median filtering techniques, an innovative box filtering method was successfully created. Additionally, this study compared the differences between the mean, median, and box filtering techniques, and proved that the 3D image improved by box filtering is a more convenient and accurate way to check the health of solar modules than the mean and median filtering methods. In addition, this new method can simplify the maintenance process, as it helps maintenance personnel to determine whether to replace the solar modules on site, achieving the goal of power generation efficiency enhancement. It is worth noting that 3D image recognition technology can enhance the clarity of thermal images, thereby providing maintenance personnel with better defect diagnosis capability. It is also able to provide the temperature value of the defect zone, and to indicate the scale of defects through the cumulative temperature chart, so the 3D image is qualified as a quantitative and qualitative indicator. The analysis of the transmitted image is innovative that it not only can locate the defect area of the module, but also can display the temperature of the module, providing more information for maintenance personnel.

Distance Metric-based Segmentation and Score-Level Classification for Optimized Tumor Identification in MR Images

Tumors in the brain and pancreas are formed by abnormal cells accumulating in a tissue mass in the human body. The standard technique for MRI tumor detection and categorization would be to investigate them utilizing MR medical images. Due to a discrepancy among many brain and image characteristics, an MR picture is segmented at the early stage of diagnosis. The tumor identification framework is processed in the following order: The initial pre-processing (or) filtering is carried out using a Modified Decision-Based Coupled Windows Median Filter (MDBCWMF) technique. This stage's output is transmitted to a unique Distance Metric-based variant of the Fuzzy C-Means algorithm for segmenting MDM-FCM. This method results in a lower segmentation error. The image is then feature extracted using a Gray-Level Co-Occurrence Matrix (GLCM) algorithm and classified using a parameterized combination of Back Propagation Neural Network (BPNN) and Sparse Representation (SR) approaches known as Dynamic Score Level Integration (DSLI). Compared to other state-of-the-art methodologies, MDM-FCM segmented yields 0.9674 and 0.9481 in classification yield 97.84% and 95.62% accuracy for brain and pancreas images, experimental valuation being done with Datasets from TCIA.

QR Code Detection

Due to the incorrect image collecting approach, QR code identification frequently confronts obstacles such as uneven backdrop fluctuations, inadequate illuminations, and distortions. As a result, identifying QR codes is difficult, and artificial intelligence-based methods were developed to address this challenge. This article uses an improved adaptive median filter technique and a QR code distortion correction method based on backpropagation (BP) neural networks to increase the identification rate of QR image codes. The distorted QR image can be fitted into the geometric deformation pattern using this combination of artificial intelligence methods, and QR code identification is possible. This paper addresses two-dimensional code distortion, which has been a critical research topic in present software systems. The research findings, which focused on the picture preprocessing step, demonstrated a significant improvement of 14 percent in the reading rate of QR image codes following processing using the system algorithm described in this article. The artificial intelligence technique used improves the recognition rate of the two-dimensional code image to some extent.

Detection of rice plant disease using AdaBoostSVM classifier

AbstractThe research in the detection of plant diseases using plant images based on machine learning is widely increased in the field of agriculture. This could be done with the images of infected rice (Oryza sativa L.) plants. The changes in atmospheric condition cause changes in soil condition and in temperature. Both air temperature and soil temperature have distinct roles in crops, which can also lead to diseases in rice plants. In this paper, a prototype is developed for the detection of rice plant diseases like bacterial leaf blight, brown spot, and leaf smut. The proposed prototype is developed by undergoing experiments on image processing using machine learning algorithms. Several images of rice leaf which are infected by diseases had been captured and preprocessed using a median filtering technique. The important features are extracted by using Discrete Wavelet Transform (DWT) for the diseased part of the leaf images. Then the green parts of the leaf have been removed, so as to extract the diseased part. The features are extracted based on the attributes like color, shape, and texture. For multiclass classification process, Adaptive Boosting support vector machine (AdaBoostSVM) Classifier is used. The proposed prototype results in the accuracy of about 98.8% in detecting and classifying the rice leaf disease.

Design and Development of Contrast-Limited Adaptive Histogram Equalization Technique for Enhancing MRI Images by Improving PSNR, UIQI Parameters in Comparison with Median Filtering

Image enhancement is used to improve the quality of images and it enhances, sharpens image features, such as edges, boundaries, and contrast, to make a graphic display useful for display and analysis. In order to enhance the quality of MRI images, histogram-based image enhancement technique is developed in this work. Materials and Methods: In this research, a Contrast Limited Adaptive Histogram Equalization (CLAHE) based image enhancement technique is proposed and developed for MRI images and the proposed work is compared with another image enhancement technique called Median Filtering (MF) method. Input medical images (N=30) of both group were downloaded from standard medical database. The enrollment ratio is obtained as 1 with 95% confidence interval and a threshold value 0.05. Results: The performance of image enhancement is measured using two parameters namely, Peak Signal Noise Ratio (PSNR) and Universal Image Quality Index (UIQI). These parameters are calculated and evaluated to assess the proposed methods efficacy. High values of PSNR and UIQI indicate better enhancement. CLAHE provides mean PSNR values of 18.6968(dB), mean UIQI of 80.9220%, and median filtering method provides mean PSNR values of 14.2261(dB) and mean UIQI of 76.3463%. Conclusion: Based on the experiment's results, the CLAHE image enhancement technique significantly performed better than the MF technique.

Two-dimensional and pseudo three-dimensional visualisation of foreign objects in milk carton using ultrasonic tomography

This paper presents the development of two-dimensional and pseudo three-dimensional visualisation of foreign object in milk carton using ultrasonic tomography system. Mathematical modelling based on ultrasonic amplitude wave and correlation between ultrasonic projection distance and voltage sensor are elaborated in this paper. Physical hazard contaminants such as metal, plastic, wood and glass materials were placed in the milk. The forward problem and the inverse problem of image reconstruction technique have been discussed briefly. 2D images of the contaminants are reconstructed using Linear Back-Projection (LBP) and Convolution Back-Projection (CBP) with Hanning, Hamming and Ram-Lak filter. A median filter technique has been combined to the algorithms for filtering noise of reconstructed images. The results show that image reconstruction using CBP-Ram-Lak has significantly yield the best image reconstruction for most tested phantoms. The application of median filter with 20 × 20 matrix have obtained a better image quality as they achieved the best values of PSNR and SSIM compared to non-median filter with 13.2765 and 0.7318, respectively for pairs of metal A - glass B (phantom Y-2). Pseudo three-dimensional (3D) image reconstruction for the foreign object has been carried out to provide more sufficient monitoring process in milk cartons. The 3D images were constructed based on stacking the 2D slice images in z-axis. Thus, the exact location of the object in the milk carton can be exposed clearly from various viewing angles. The results indicate that the ultrasonic tomography imaging system promises to be a non-invasive tool in quality monitoring especially for the detection of foreign objects in milk carton.

Human activity recognition in WBAN using ensemble model

PurposeFor achieving the profitable human activity recognition (HAR) method, this paper solves the HAR problem under wireless body area network (WBAN) using a developed ensemble learning approach. The purpose of this study is,to solve the HAR problem under WBAN using a developed ensemble learning approach for achieving the profitable HAR method. There are three data sets used for this HAR in WBAN, namely, human activity recognition using smartphones, wireless sensor data mining and Kaggle. The proposed model undergoes four phases, namely, “pre-processing, feature extraction, feature selection and classification.” Here, the data can be preprocessed by artifacts removal and median filtering techniques. Then, the features are extracted by techniques such as “t-Distributed Stochastic Neighbor Embedding”, “Short-time Fourier transform” and statistical approaches. The weighted optimal feature selection is considered as the next step for selecting the important features based on computing the data variance of each class. This new feature selection is achieved by the hybrid coyote Jaya optimization (HCJO). Finally, the meta-heuristic-based ensemble learning approach is used as a new recognition approach with three classifiers, namely, “support vector machine (SVM), deep neural network (DNN) and fuzzy classifiers.” Experimental analysis is performed.Design/methodology/approachThe proposed HCJO algorithm was developed for optimizing the membership function of fuzzy, iteration limit of SVM and hidden neuron count of DNN for getting superior classified outcomes and to enhance the performance of ensemble classification.FindingsThe accuracy for enhanced HAR model was pretty high in comparison to conventional models, i.e. higher than 6.66% to fuzzy, 4.34% to DNN, 4.34% to SVM, 7.86% to ensemble and 6.66% to Improved Sealion optimization algorithm-Attention Pyramid-Convolutional Neural Network-AP-CNN, respectively.Originality/valueThe suggested HAR model with WBAN using HCJO algorithm is accurate and improves the effectiveness of the recognition.

Statistical Methods to Improve the Quality of Real-Time Drilling Data

Abstract The age of easy oil is ending, and the industry started drilling in remote unconventional conditions. To help produce safer, faster, and most effective operations, the utilization of artificial intelligence and machine learning (AI/ML) has become essential. Unfortunately, due to the harsh environments of drilling and the data-transmission setup, a significant amount of the real-time data could defect. The quality and effectiveness of AI/ML models are directly related to the quality of the input data; only if the input data are good, the AI/ML-generated analytical and prediction models will be good. Improving the real-time data is therefore critical to the drilling industry. The objective of this paper is to propose an automated approach using eight statistical data-quality improvement algorithms on real-time drilling data. These techniques are Kalman filtering, moving average, kernel regression, median filter, exponential smoothing, lowess, wavelet filtering, and polynomial. A dataset of +150,000 rows is fed into the algorithms, and their customizable parameters are calibrated to achieve the best improvement result. An evaluation methodology is developed based on real-time drilling data characteristics to analyze the strengths and weaknesses of each algorithm which were highlighted. Based on the evaluation criteria, the best results were achieved using the exponential smoothing, median filter, and moving average. Exponential smoothing and median filter techniques improved the quality of data by removing most of the invalid data-points; the moving average removed more invalid data-points but trimmed the data range.

Hybrid Convolution Neural Network in Classification of Cancer in Histopathology Images.

Cancer statistics in 2020 reveals that breast cancer is the most common form of cancer among women in India. One in 28 women is likely to develop breast cancer during their lifetime. The mortality rate is 1.6 to 1.7 times higher than maternal mortality rates. According to the US statistics, about 42,170 women in the US are expected to die in 2020 from breast cancer. The chance of survival can be increased through early and accurate diagnosis of cancer. The pathologists manually analyze the histopathology images using high-resolution microscopes to detect the mitotic cells. This is a time-consuming process because there is a minute difference between the normal and mitotic cells. To overcome these challenges, an automatic analysis and detection of breast cancer by using histopathology images play a vital role in prognosis. Earlier researchers used conventional image processing techniques for the detection of mitotic cells. These methods were found to be producing results with low accuracy and time-consuming. Therefore, several deep learning techniques were adopted by researchers to increase the accuracy and minimize the time. The hybrid deep learning model is proposed for selecting abstract features from the histopathology images. In the proposed approach, we have concatenated two different CNN architectures into a single model for effective classification of mitotic cells. Convolution neural network (CNN) automatically detects efficient features without human intervention and classifies cancerous and non-cancerous images using a hybrid fully connected network. It is a computationally efficient, very powerful, and efficient model for performing automatic feature extraction. It detects different phenotypic signatures of nuclei. In order to enhance the accuracy and computational efficiency, the histopathology images are preprocessed, segmented, and feature extracted through CNN and fed into a hybrid CNN for classification. The hybrid CNN is obtained by concatenating two CNN models; together, this is called model leveraging. Model averaging can be improved by weighting the contributions of each sub-model to the combined prediction by the expected performance of the sub-model. The proposed hybrid CNN architecture with data preprocessing with median filter and Otsu-based segmentation technique is trained using 50,000 images and tested using 50,000 images. It provides an overall accuracy of 98.9%.