This paper presents a novel approach for detecting and removing cross-color noise. Specifically, a unique characteristic of the noise is first recognized in several color spaces, and a single detection measure is defined in the form of a correlation coefficient. Then, a simple reduction scheme is provided. Main difference between the prior methods and the proposed one is that the previous approaches have been working on the reduction scheme during decoding the transmitted signal with available useful information, while the proposed scheme is focused on the images without any information from the decoder as a post-processing method. The proposed algorithm can be applicable to any video sources contaminated by cross-color noise including television.