What are Text, Generic and Halftone regions?
Each page of a JBIG2 source image can be partitioned into one or more rectangular regions. A region can be as large as the entire page or as tiny as a few pixels on a side. How a page is partitioned into one or more regions is up to the user. Source images generally contain three types of image content – text such as a letter, line art or halftone image. Each type of content compresses best using a unique encoding technique. Text regions are best coded using a symbol dictionary coding approach, while line art is best coded using either a generic coding. Generic coding does not analyze the region for repetitive symbols while halftone coding applies an error diffusion algorithm to the input to improve coding efficiency. Generic region coding is always lossless while text and halftone coding can be lossy or lossless. Text region lossless coding gives the best compression rate at the expense of longer encoding times and lost pixels. The JBIG2 standard does not state how a page is partition
