Yazar "Tunga, Burcu" seçeneğine göre listele

Listeleniyor 1 - 6 / 6
  • Küçük Resim
    Öğe
    A novel image denoising technique with Caputo type space-time fractional operators
    (Springer, 2024) Tanrıöver, Evren; Kiriş, Ahmet; Tunga, Burcu; Tunga, Mehmet Alper
    A novel image denoising model, namely Full Fractional Total Variation (TVFF), based on the Rudin-Osher-Fatemi (ROF) and the fractional total variation models is presented. The leading advantage of TVFF model is that it uses fractional derivatives with length scale parameters instead of ordinary derivatives with respect to both time and spatial variables in the diffusion equation. The Riesz-Caputo fractional derivative operator is used to disperse nonlocal influence throughout all directions, whereas the Caputo fractional derivative concept is employed for time fractional derivatives. Therefore, the influence of neighboring pixels is given greater weight compared to those situated farther away and this reflects the consideration behind denoising process better. Moreover, the numerical approach is constructed, and its stability and convergence properties are thoroughly examined. To show the superiority of our model, the denoised images are subjected to visual and numerical comparisons using metrics such as the Signal-to-Noise Ratio (SNR), the Structural Similarity Index Measure (SSIM) and the Edge-Retention Ratio (ERR). The performance of the TVFF method is evaluated under various types of noise, including Poisson, Speckle, and Salt & Pepper, and the results are compared with those obtained using Gauss and Median Filters. Furthermore, the proposed method is applied to both blind and synthetic images, thereby showcasing its versatility and applicability across diverse datasets. The outcomes showcase the substantial potential of our enhanced model as a versatile and efficient tool for image denoising.
  • Küçük Resim Yok
    Öğe
    An efficient feature extraction approach for hyperspectral images using Wavelet High Dimensional Model Representation
    (Taylor and Francis Ltd., 2022) Tuna, Süha; Korkmaz Özay, Evrim; Tunga, Burcu; Gürvit, Ercan; Tunga, Mehmet Alper
    Hyperspectral (HS) Imagery helps to capture information using specialized sensors to extract detailed data at numerous narrow wavelengths. Hyperspectral imaging provides both spatial and spectral characteristics of regions or objects for subsequent analysis. Unfortunately, various noise sources decrease the interpretability of these images as well as the correlation between neighbouring pixels, hence both reduce the classification performance. This study focuses on developing an ensemble algorithm that enables to denoise the spectral signals while decorrelating the spatio-spectral features concurrently. The developed method is called Wavelet High Dimensional Model (W-HDMR) and combines High Dimensional Model Representation (HDMR) with the Discrete Wavelet Transform (DWT). Through W-HDMR, denoised and decorrelated features are extracted from the HS cubes. HDMR decorrelates each dimension in HS data while DWT denoises the spectral signals. The classification performance of W-HDMR as a new feature extraction technique for HS images is assessed by exploiting a Support Vector Machines algorithm. The results indicate that the proposed W-HDMR method is an efficient feature extraction technique and is considered an adequate tool in the HS classification problem.
  • Küçük Resim Yok
    Öğe
    DeepEMPR: coffee leaf disease detection with deep learning and enhanced multivariance product representation
    (PeerJ Inc., 2024) Topal, Ahmet; Tunga, Burcu; Tirkolaee, Erfan Babaee
    Plant diseases threaten agricultural sustainability by reducing crop yields. Rapid and accurate disease identification is crucial for effective management. Recent advancements in artificial intelligence (AI) have facilitated the development of automated systems for disease detection. This study focuses on enhancing the classification of diseases and estimating their severity in coffee leaf images. To do so, we propose a novel approach as the preprocessing step for the classification in which enhanced multivariance product representation (EMPR) is used to decompose the considered image into components, a new image is constructed using some of those components, and the contrast of the new image is enhanced by applying high-dimensional model representation(HDMR)to highlight the diseased parts of the leaves. Popular convolutional neural network (CNN) architectures, including AlexNet, VGG16, and ResNet50, are evaluated. Results show that VGG16 achieves the highest classification accuracy of approximately 96%, while all models perform well in predicting disease severity levels, with accuracies exceeding 85%. Notably, the ResNet50 model achieves accuracy levels surpassing 90%. This research contributes to the advancement of automated crop health management systems. © 2024 Topal et al.
  • Küçük Resim
    Öğe
    An efficient feature extraction approach for hyperspectral images using wavelet high dimensional model representation
    (TAYLOR & FRANCIS LTD, 2022) Tuna, Süha; Özay, Evrim Korkmaz; Tunga, Burcu; Gürvit, Ercan; Tunga, Mehmet Alper
    Hyperspectral (HS) Imagery helps to capture information using specialized sensors to extract detailed data at numerous narrow wavelengths. Hyperspectral imaging provides both spatial and spectral characteristics of regions or objects for subsequent analysis. Unfortunately, various noise sources decrease the interpretability of these images as well as the correlation between neighbouring pixels, hence both reduce the classification performance. This study focuses on developing an ensemble algorithm that enables to denoise the spectral signals while decorrelating the spatio-spectral features concurrently. The developed method is called Wavelet High Dimensional Model (W-HDMR) and combines High Dimensional Model Representation (HDMR) with the Discrete Wavelet Transform (DWT). Through W-HDMR, denoised and decorrelated features are extracted from the HS cubes. HDMR decorrelates each dimension in HS data while DWT denoises the spectral signals. The classification performance of W-HDMR as a new feature extraction technique for HS images is assessed by exploiting a Support Vector Machines algorithm. The results indicate that the proposed W-HDMR method is an efficient feature extraction technique and is considered an adequate tool in the HS classification problem.
  • Küçük Resim
    Öğe
    High dimensional model representation median filter for removing salt and pepper noise
    (Springer science and business media deutschland GmbH, 2024) Kaçar, Sena; Tunga, Burcu
    In this paper, we introduce a novel method for noise reduction called the High-Dimensional Model Representation (HDMR) Median Filter. HDMR, functioning as a divide and conquer algorithm, dynamically adapts its filtering parameters based on local image characteristics. This adaptation aims to preserve image quality while efficiently eliminating noise. The HDMR Median Filter excels in retaining fine details, preserving edges, and safeguarding essential image features. To assess the performance of the proposed algorithm, we conducted comparisons with results from recent studies. These comparisons were based on two commonly used metrics: Structural Similarity (SSIM) and Peak Signal-to-Noise Ratio (PSNR) The obtained results demonstrate that the noise reduction algorithm obtained with HDMR performs better than the most recent studies in the literature.
  • Küçük Resim Yok
    Öğe
    A novel multistage CAD system for breast cancer diagnosis
    (Springer London Ltd, 2023) Karacan, Kubra; Uyar, Tevfik; Tunga, Burcu; Tunga, M. Alper
    Computer-aided diagnosis (CAD) systems are widely used to diagnose breast cancer using mammography screening. In this research, we proposed a new multistage CAD system based on image decomposition with High-Dimensional Model Representation (HDMR) which is a divide-and-conquer algorithm. We used digital mammograms from Digital Database for Screening Mammography as dataset. We neglected BIRADS classification and used a brand-new clustering based on HDMR constant and breast size. To find the best performance of HDMR-based CAD system, we compared different pre-processing settings such as contrast enhancement with CLAHE and HDMR, feature extraction with HDMR, feature scaling, dimension reduction with Linear Discriminant Analysis. We used several Machine Learning algorithms and measured the performance of proposed system for normal-benign-malign classification, cancer detection, mass detection and found that the proposed system achieves 66%, 71% and 87% accuracy, respectively. We were able to achieve 92% accuracy, 100% sensitivity and 91% specificity in specific clusters. These results are comparable with deep learning-based methods although we simplified the pipeline and used brand-new HDMR-based processes.