Noise level is a major factor when assessing the video quality. Although digital technologies have advanced, imaging is an analogue technology based on the photo-electric conversion imaging characteristics. Recently the sensitivity of imaging devices improved dramatically. However, noise is visible when the content is created under poor lighting condition. A signal processing called noise reducer (NR) was used to cope with this problem. When we use the NR, the noise is one of the important parameters. When the NR works on lower noise level than the actual noise level in the video, noise cannot be reduced sufficiently. In contrast, when the NR works on the higher noise level, video blur occurs and details might lose. It is necessary to detect the exact noise level in order to take the full advantage of the NR. It, however, is very difficult to pinpoint the actual noise level in general videos. In this paper, we propose a novel noise level estimation algorithm.