| 摘要 |
A method of processing video comprising a succession of images of a scene to detect movement of an object in the scene, comprising storing a matrix of bytes each representing the difference in intensity of a respective pixel between temporally adjacent images, creating from those bytes a histogram of the frequency at each of a range of positive and negative values of the intensity difference, locating the limits of a central peak of the frequency distribution in the histogram so as to include in the central peak changes due to illumination change and repetitive motions in the background of the scene, determining the proportion of bytes in the remainder of the distribution that are in tails of the distribution on each side of the central peak, repeating the aforesaid process for each successive pair of temporally adjacent images, and providing an output indicative of a moving object and not indicative of illumination change or repetitive motions, in the event that that proportion meets a predetermined criterion. |
