A system and method for extracting feature data of dynamic objects selects sequential N frames of a video file up front, where N is a positive integer, and divides each of the N frames into N*N squares. The system and method further selects any n frames from the N frames, selects any n rows and n columns of the n frames to obtain n*n*n squares, where n is a positive integer. The system and method further extracts feature data from the video file by computing averages and differences for pixel values of the n*n*n squares.