Image noise recognition algorithm based on BP neural network
Yi Huo, Xiaoxuan Ma · 2020
As an important research subject in the field of digital image processing, noise recognition of noisy images is still faced with many problems of low accuracy in recognition algorithm, such as noise value of misjudgment and miscalculation recognized by the recognition algorithm, truth-value misjudgment and miscalculation, etc. By analyzing BP neural network, a recognition algorithm based on BP neural network is designed to solve the above problems. Firstly, the cause of image noise and two classical recognition algorithms are analyzed. Secondly, the input value and network structure of BP neural network identification algorithm are designed. Then, the training method and decision criteria of BP neural network recognition algorithm are described. Finally, Matlab software is used to simulate and identify the noise points of the image containing salt and pepper noise. Simulation results show that the BP neural network identification algorithm has a very low noise leakage number and False Alarming Ratio(FAR), and a good identification effect for salt and pepper noise.