Scholarlink

In precision-based tasks such as medical imaging and computer vision, using high quality and noise free images is of utmost importance. Existing techniques fail to satisfactorily denoise the images in large datasets, causing the noise to distort the image quality. Recognizing the role of artificial intelligence in advancing image processing techniques, this paper proposes GAN-CNN ImaDe, a Generative Adversarial Network and Convolutional Neural Network-Based Image Denoising Model. In the proposed model, the generator and discriminator component of Generative Adversarial Network has been utilized for denoising and quality assessment of the images respectively. On the other hand, to perform enhanced feature extraction, the Convolutional Neural Network has been utilized. Extensive experimentation has been performed and the performance of the proposed model has been evaluated using the CBSD68-dataset. Analysis of the loss function values of the generator as well as the discriminator over several training sessions gave deep insight into the performance of the model. The effectiveness of the proposed method is highlighted by the loss value of 0.5 for the generator and 0.7 for the discriminator.

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

Image Denoising with Generative AI: A GAN-CNN-Based Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Kanika Dhiman,
  • Pankaj Deep Kaur

摘要

In precision-based tasks such as medical imaging and computer vision, using high quality and noise free images is of utmost importance. Existing techniques fail to satisfactorily denoise the images in large datasets, causing the noise to distort the image quality. Recognizing the role of artificial intelligence in advancing image processing techniques, this paper proposes GAN-CNN ImaDe, a Generative Adversarial Network and Convolutional Neural Network-Based Image Denoising Model. In the proposed model, the generator and discriminator component of Generative Adversarial Network has been utilized for denoising and quality assessment of the images respectively. On the other hand, to perform enhanced feature extraction, the Convolutional Neural Network has been utilized. Extensive experimentation has been performed and the performance of the proposed model has been evaluated using the CBSD68-dataset. Analysis of the loss function values of the generator as well as the discriminator over several training sessions gave deep insight into the performance of the model. The effectiveness of the proposed method is highlighted by the loss value of 0.5 for the generator and 0.7 for the discriminator.