In the arena of digital image processing, mixed noise is normally depicted. Nevertheless, the main limitation of the existing noise removal methods is the exploitation of the statistics of the original contaminated image. Concerns such as high computational complexity, halo artifacts and over-saturation issue are presented in the edges of the noisy images i.e. high-frequency components of the mixed noise digital images which lead to false guesstimate of noisy patches that negatively results on the task of recovery processes of the restoration images. This paper proposes a rapid and high accurate mixed noise removal method by combining pulse coupled neural networks (PCNN) and regularization of Perona-Malik equation (P-M equation) in order to remove unwanted contamination. In this regard, the locations of impulse noise are positioned using PCNN, the second-generation wavelet filter is used to suppress the mixed noise into small local neighborhood, and then the full noisy image is denoised by exploiting the regularization of P-M equation. The fine details and sharp edges are well preserved in the proposed method. In addition, subjective and objective analyses are showed that the visual quality of the proposed mixed noise removal technique outperformed state of the art noise removal methods. Extensive results show that the proposed model outperforms current state-of-the-art mixed noise removal techniques, with PSNR improvements of (0.85 dB to 1.54 dB) and SSIM improvements range of (0.0132-0.1521). Moreover, improvements in running time of the proposed technique was the second best technique with improvements range of (0.33-0.56 s) on wide range of benchmark digital images.
