Efficient impulse noise reduction based on the improved direction impulse detector and PDE

In this paper, we present a new improved impulse detector, aiming at locating the polluted pixel more exactly. To achieve the goal, we make efforts in both reducing the miss rate and false-warn rate and get satisfactory results. And at the same time, in order to preserve the details or edges of image more efficiently and get more satisfactory visual effects, we also involve in making new nonlinear anisotropic diffusion filter model based on the primary P-M method. Therefore, we propose an improved model based on open system and adopt a coefficient to avoid the diffusion of the contaminated neighborhood pixels to the current processing pixel. The experiment results can fully illustrate its advantages in removing the impulse noise with high noise level and preserving more detailed feature of the image.