A Scalar Median Filter Algorithm Based on the Rotation of Color Space

A scalar median filter algorithm for color image based on the rotation of color space is presented in this paper. It converts the vector median filter to simpler scalar median filters by rotating color space. The rotation of color space can make the majority noise pixels satisfying scalar median filter conditions at each scalar and being filtered. This method is iterative, and the rotation matrix used in the color space rotation generated by the Principle Component Analysis can reduce iteration times. Tested on several color images and compared with other traditional algorithm, this presented filter method is acceptable. It not only makes the filter computationally simpler, but also preserves the image details well while effectively suppressing impulsive noises.