Finally, we come to what I believe is the most interesting facet of the JPEG 2000 standard. The people at JPEG have figured out a way for the user to define a Region of Interest (ROI) within the image. This is a region where the user specifies less loss of image quality. This region can be, of course, a square or rectangle. However, it can also be a circle, oval, triangle, or any blob-like shape the user wants to specify. We can then protect this ROI with one of two methods.
In the first method, we can create an ROI mask. To do this, we must make the bits that describe the region in the ROI more "important". That is, we must trick the utility into thinking that every bit inside the ROI is important. We can do this scaling down the coefficients that are in the background section of the image. When we do this, the coefficient associated with the ROI are coded first in the bit modeling step. This also means that, compared to the now scaled down coefficients of the background, the coefficients of the ROI seem quite important. Thus, those coefficients do not change much during the course of the compression.
The second method is called the Maxshift Method. We must remember that the tiles of the image can be processed and compressed independently. This means that each tile can be compressed using a different method. In fact, some tiles can be compressed using a lossy method while other can be compressed using the lossless method. This feature can be exploited to protect the ROI. We simply specify the tiles in the ROI to be compressed with a lossless format.