Abstract: Raster reduction is performed on a binary image. The binary image includes a number of features, and the raster reduction decomposes the binary image into rectangular areas of features. The binary image is separately decomposed into row-oriented rectangles and column-oriented rectangles. Row- and column-oriented decomposition involves identifying rectangles that are as long as possible along the orientation direction including only contiguous features, and then widening the rectangles to include segments of contiguous features in neighboring contiguous rows or columns, without including segments that belong in other rectangles. The row- and column-oriented rectangles are then merged into a set. The rectangle set is then ordered, and a subset of the rectangles is selected. The selected rectangles cover every feature in the binary image. The binary image is then displayed using the set of selected rectangles, each selected rectangle being displayed in a different color.