A novel denoising method using connectivity judgment and adaptive filtering

The noise caused by various external factors gives a negative effect on optical measurements, object detection, etc. The filter-based random impulse denoising methods have the disadvantage of poor noise detection accuracy when the noise density is high and poor noise removal for weak intensity noises. In this paper, a novel denoising method using connectivity judgment and adaptive filter is proposed. First, the background noise and edge noise are separated individually by the connectivity judgment based on the brightness different of the pixels. And, the filter size is actively adapted according to the magnitude of the local noise density, and reasonable filter weight is determined. Then, high density background noise is removed by using the recursive method and edge noise is removed by using the edge direction information. Experiments were performed with image data which is added random impulse noise between 0-255 and Gaussian noise with 0 mean and variance of 20 at the ranging from 10 to 50%, and the proposed method was superior in terms of noise detection accuracy, speed and noise reduction accuracy compared to the previous methods.