Methods and devices receive a document comprising raster images, using an optical scanner. These methods and devices automatically identify topical items within the raster images based on raster content in the raster images, using a processor. Further, these methods and devices automatically associate the topical items with topics in the document based on previously established rules for identifying topical sections, and automatically crop the topical items from the raster images to produce cropped portions of the raster images, using the processor. These methods and devices then automatically create an index for the document by combining the cropped portions of the raster images organized by the topics, using the processor, and output the index from the processor.