An Improved Non-local Means Denoising Algorithm and Its Application in Denoising Analysis of Metal Fracture
Han Yan, Xuxiu Zhang, Weidong Li · 2020
The analysis of metal fracture image plays an important role in metal intelligent diagnosis. Precise analysis of material properties of metal fractures requires denoising preprocessing of images. This paper presents a non-local means denoising algorithm based on improved neural network to denoise metal fracture images. Specifically, the global connection in the non-local means denoising algorithm is used to extract the non-local data in the noise image as the input to train neural network, and the noise image is iteratively filtered by the trained model. The experimental results show that the denoising performance of the proposed algorithm has higher signal-to-noise ratio than the traditional algorithms, and it is more complete in the preservation of metal fracture boundary information.