Variations of diffusion functions on Perona and Malik model for noise removal

In 1990, Perona and Malik introduced diffusion model which is known as Perona and Malik (P-M) model to preserve the edge or boundaries. The edge indicator used as the variable exponent controls the diffusion mode as edge stopping between P-M diffusion and Gaussian smoothing depending on the image. The application of nonlinear second-order PDE model of P-M model is designed such that smooth areas diffused faster than the other areas and consequently may cause the presence of blocky effect. Hence, numerous modified diffusion functions have been developed to enhance image denoising results in order to overcome the loss of detail and blur in images. The objectives of this paper are to propose modified diffusion functions and compare with existing functions in order to provide alternative diffusion functions with edge and details preservation. Besides that, the modified functions were implemented on an image with different level of noise to made visual comparison with existing functions The proposed functions must be a positive function, monotonically decreasing and also convergent to zero. Results shows that the modified functions were converged rapidly as compared to the existing diffusion functions and able to remove different level of noise. In addition, the quality of image and their details are also preserved after the denoising process.