The rapid increase in digital video content demands effective summarization techniques, specially with the creation of RGBD videos. Keyframe extraction significantly reduces the amount of raw data in a video sequence. In this paper, we present a two-stage (histogram and filtering) keyframe extraction algorithm applicable on RGB and RGBD videos. In the first stage, RGB and depth histogram similarities of consecutive frames are computed and candidate keyframes are extracted. In the second stage, we filter neighboring candidate keyframes based on the MAD of their Euclidean distance and their MSE. Subjective and objective experimental results show our algorithm effectively extracts keyframes from both RGB and RGBD videos.