Qing Liu, Lu-ping Xu, Yi-de Ma, Yong Wang, and Qiang Xie
Image filtering, Gaussian noise, PCNN, time matrix, bilateral filtering
To effectively remove Gaussian noise in images, a novel bilateral filtering algorithm based on improved pulse coupled neural networks (PCNN) is introduced. Firstly, smooth restraining factor and adaptive linking strength are adopted. They are combined with the synchronous fire of similar neuron in improved PCNN from the aspect of the characteristics of image Gaussian noise. Secondly, the noisy image is processed by using PCNN pre-de-noise iteration. The extreme value noise is removed and the time matrix that can reflect image spatial–temporal information is also formed. Finally, the time matrix is operated in bilateral filtering which is given adaptive improvement and de-noise application. The results of experiments show that the novel algorithm can effectively remove noise in smooth region in the case of good preserving image edges and details; the de-noised images have good subjective vision effects and objective quality. Meanwhile, this algorithm presents high signal-to-noise ratio, strong capability to reduce any and has good adaptability.
Important Links:
Go Back
