We report an improved methodology for training classi- fiers for document image content extraction, that is, the lo- cation and segmentation of regions containing handwriting, machine-printed text, photographs, blank space, etc. Our previous methods classified each individual pixel separately (rather than regions): this avoids the arbitrariness and re- strictiveness that result from constraining region shapes (to, e.g., rectangles). However, this policy also allows content classes to vary frequently within small regions, often yield- ing areas where several content classes are mixed together. This does not reflect the way that real content is organized: typically almost all small local regions are of uniform class. This observation suggested a post-classification method- ology which enforces local uniformity without imposing a restricted class of region shapes. We choose features ex- tracted from small local regions (e.g. 4-5 pixels radius) with which we train classifiers that operate on the output of previous classifiers, guided by ground truth. This pro- vides a sequence of post-classifiers, each trained separately on the results of the previous classifier. Experiments on a highly diverse test set of 83 document images show that this method reduces per-pixel classification errors by 23%, and it dramatically increases the occurrence of large contigu- ous regions of uniform class, thus providing highly usable near-solid `masks' with which to segment the images into distinct classes. It continues to allow a wide range of com- plex, non-rectilinear region shapes.