An impulse noise fading technique based on local histogram processing

摘要

Impulse noise is one of the common noise models. Therefore, finding methods to remove such noise has high importance in image enhancement. In this paper, we propose a novel noise detection technique called HET (Histogram Evaluation Technique) which assesses image histogram locally to identify noisy pixels. The experimental results have shown that HET outperforms other noise detection techniques and produces images with higher quality. The proposed method works well in a wide range of impulse noise when the noise varies from 1% up to 98%.