Scholarlink

<p>In medical imaging, high-quality images are essential for accurate diagnosis and effective treatment planning. However, real-world imaging systems often suffer from degradation caused by noise, blur, and other distortions during acquisition, transmission, or storage. Such degradation not only reduces visual clarity but also affects the performance of computer vision models. To overcome these issues, this paper develops the Siamese Convolution Deep Belief Network (SCDBNet) for image degradation source identification with image enhancement. Deep Kronecker Network (DKN) performs the noisy, and blurred pixel identification from the input image. The identification of the image degradation source is performed using the proposed SCDBNet. The image denoising is done by the Convolution Neural Network (CNN) and kernel estimation is utilized for image denoising. The image enhancement is performed by the fusing of denoised image and deblurred image. Moreover, the SCDBNet-based image degradation source identification with image enhancement attains the superior Degree of Distortion (DD), Pear Signal to Noise Ratio (PSNR), Mean Square Error (MSE), structural similarity index measure (SSIM), Visual Information Fidelity (VIF), and Figure of Merit (FOM) of 0.260, 0.542, 48.47&#xa0;dB, 0.961, 0.865, and 0.978, confirming its effectiveness for real-time medical image enhancement and analysis.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Medical Image Enhancement with Image Degradation Source Identification Using Hybrid Siamese Convolutional Deep Belief Network

  • 期刊论文

  • 发表日期：

    2025年12月15日

查看全文
  • Mankala Narender,
  • Seetharam Khetavath

摘要

In medical imaging, high-quality images are essential for accurate diagnosis and effective treatment planning. However, real-world imaging systems often suffer from degradation caused by noise, blur, and other distortions during acquisition, transmission, or storage. Such degradation not only reduces visual clarity but also affects the performance of computer vision models. To overcome these issues, this paper develops the Siamese Convolution Deep Belief Network (SCDBNet) for image degradation source identification with image enhancement. Deep Kronecker Network (DKN) performs the noisy, and blurred pixel identification from the input image. The identification of the image degradation source is performed using the proposed SCDBNet. The image denoising is done by the Convolution Neural Network (CNN) and kernel estimation is utilized for image denoising. The image enhancement is performed by the fusing of denoised image and deblurred image. Moreover, the SCDBNet-based image degradation source identification with image enhancement attains the superior Degree of Distortion (DD), Pear Signal to Noise Ratio (PSNR), Mean Square Error (MSE), structural similarity index measure (SSIM), Visual Information Fidelity (VIF), and Figure of Merit (FOM) of 0.260, 0.542, 48.47 dB, 0.961, 0.865, and 0.978, confirming its effectiveness for real-time medical image enhancement and analysis.