An Adaptive Image Filter Based on Decimal Object Scale for Noise Reduction and Edge Detection

A novel adaptive image filter is proposed for noise reduction and edge detection where the decimal object scale is adopted to detect the size of the local structure. The image filter contains two parts: an adaptive median filter and an adaptive Gaussian filter. Its main idea is: the strong noise points are removed by using an adaptive median filter which smoothes the image selectively according to the decimal object scale, then, the remaining weak noise points are removed by using an adaptive Gaussian filter of which the variance and mask size are controlled by the decimal object scale and integral object scale respectively. The main contribution of this paper is that the integral object scale is upgraded to the decimal object scale for reflecting the local structure size more accurately. Both subjective and objective comparisons indicate that our method can preserve the fine details while suppress noise, and is superior to the several traditional methods.