Research Of Image Denoising Algorithm Based On Artificial Neural Network

Digital image is closely related to people’s daily life.With the advancement of information technology,the Internet and mobile devices,people are exposed to more and more digital image information in life.Nowadays,various image-related devices have higher requirements on the quality of digital image.Image denoising algorithms are highly related to the quality of digital images in the field of image processing,but it is difficult for traditional image denoising methods to effectively denoise a wide variety of digital images.Thanks to the rapid development of machine learning,especially deep learning,artificial neural network models in the field of deep learning have made outstanding achievements in the field of image processing in recent years.More and more artificial neural networks are applied in the field of image noise denoising.Compared with traditional image denoising methods,image denoising algorithms based on artificial neural network in datasets with correct labels often achieve better results.This paper is based on the artificial neural network image denoising algorithm.Based on the convolution neural network(CNN)algorithm,it mainly studies the effect of the image denoising algorithm model based on U-Net in the high resolution image denoising task.Firstly,the noise data in this dataset is analyzed statistically,and it is known that the noise of the digital image in this paper belongs to a variety of additive Gaussian white noise.Then,in order to solve the problem that the highresolution image is difficult to use due to the large amount of data and computation in participating in model training,the original image data is cut into picture blocks and then used by the model.The Peak Signal-to-Noise Ratio(PSNR)and Structural Similarity(SSIM)were used as evaluation criteria to evaluate the noise denoising effect of the model.The study focused on investigating the effects of image enhancement,loss functions,activation functions,and global connections on optimizing the U-Net network for image denoising.The comparison between the U-Net-based denoising algorithm and several other image denoising algorithms was conducted,which proved that the U-Net network can effectively accomplish the denoising task and achieve the best results in the high-resolution image data used in this study.Finally,combined with the U-Net network performance on the test set,the model is optimized and analyzed using batch normalization,residual connectivity and attention mechanism techniques.