Movie segmentation into scenes and chapters using locally weighted bag of visual words