Abnormal event detection in crowd scenes using quaternion discrete cosine transformation signature

In this paper, an abnormal event detection system inspired by the saliency attention mechanism of human visual system is presented. Conventionally, training-based methods assume that anomalies are events with rare appearance, which suffer from visual scale, complexity of normal events and insufficiency of training data. Instead, we make the assumption that anomalies are events that attract human attentions. Temporal and spatial anomaly saliency are considered consistently by representing the value of each pixel in each frame as a quaternion composed of intensity, contour, motion-speed and motion-direction feature. For each quaternion frame, Quaternion Discrete Cosine Transformation (QDCT) and signature operation are applied. The spatio-temporal anomaly saliency map is developed by inverse QDCT and Gaussian smoothing. Abnormal events appear at those areas with high saliency values. Experiments on typical datasets show that our method can achieve high accuracy results.