To provide an abnormal behavior detection apparatus and an abnormal behavior detection method for detecting any abnormal behavior on the basis of animations.The abnormal behavior detection apparatus 11 creates frame-to-frame difference data from animation data inputted from a video camera 10, extracts characteristic data through cubic higher-degree local autocorrelation from three-dimensional dataconsisting of a plurality of frame-to-frame difference data, and calculates the distance between a partial space derived from a principal component vector obtained from previous characteristic data by use of the technique of principal component analysis and the latest characteristic data. The distance being greater than a predetermined value is determined to be an abnormality. Normal behavior islearned as the partial space and any abnormal behavior is detected as an aberration therefrom, so that, for example, even if there are a plurality of persons within a screen, it is possible to detectthe abnormal behavior of one person. Also, real-time processing is made possible with a small amount of computation.