A technique for analyzing urban space structures quantitatively based on fractal theory is presented. The technique employs the fractal sequence of a shade and shadow image of an illuminated three-dimensional block model. Changes in the height, relative position and shape of the buildings in the block are sensitively reflected in the image feature distance calculated from the fractal sequence, providing a much more sensitive index than the fractal dimension. The technique is applied to the analysis of a real urban block with good results. The proposed technique is therefore a useful tool for performing structural analyses and precise, quantitative classification, with potential uses in urban planning and environmental engineering.