A Pixel-Weighting Method for Discriminating Objects of Different Sizes in an Image Captured from a Single Camera.