The main features of DjVu compared to previous implementations of compression algorithms are as follows.
First, it is made as the final image would be of two layers - one with the areas that have sharply contrasting border (foreground, typically used for text or graphics in the form of thin lines), another layer - with photographic quality of the picture (background, bearing the image itself).
Second, it is accordingly applied the two-pass algorithm for compression / decompression of images that occurs in real time (sequential processing of both layers).
Third, DjVu involved well-established method for the compression algorithm Wavelet - IW44 (by the way, it has long been used to store Corel fotoIMAGES supplied in the library ready IMAGES).
Fourth, to determine the compression method is selected for a certain area of ??the image is an intelligent encoder binary adaptive entropy coder.
Fifth, this Format does not build in memory the entire image, as is usual, but only the currently visible part of the screen. And the output to the screen is the same as that of the progressive GIF (with a gradual manifestation of the image).