Quantitative evaluation of edge preserving noise-smoothing filter

This paper presents a method to evaluate the performance advantages of an edge preserving noise-smoothing filter algorithm in remote sensing based on the objectivity of qualitative and quantitative calculations. The evaluation studies on simulated images were made under the effects of noise. Evaluation methods were used to compare the effectiveness of four different kinds of edge preserving noise-smoothing filters and observations were made. The median filter performed relatively well in both homogenous and edge dominated areas, and tends to distort the edge in the form of blurring. Averaging produces powerful noise reduction in homogenous areas but tends to distort the edge in the form of blurring. Nagao and Matsuyama´s (1981) edge preserving noise-smoothing filter can reduce the noise but in this study that filter also reduces the interesting features. The developed filter produces better smoothing of noise and better quality than the others because this filter has control for every time of processing