PHOTOTRACE The Art of Recognition
A PHOTOTRACE image is classified as a "bilevel" or "binary" image. Fortunately all the widely-used image file formats already accomodate PHOTOTRACE images since "bilevel" or "binary" image support has been around for a very long time.
An uncompressed PHOTOTRACE image is stored as only 1-bit-per-pixel (1=white, 0=black). With such a small stored image file size, you may not need further image compression. In this case, your image applications can be much simpler.
What if you do require a smaller image file size? When further compression of a PHOTOTRACE image is required, among commonly used compression methods, CCITT Group4 works well with PHOTOTRACE images, providing around 2:1 further compression. JBIG, the standard for bilevel compression, improves on this only slightly. JBIG and JBIG2 are kind of disappointing for this type of image when you consider how much more complicated they are than CCITT Group4.
When more aggressive compression of a PHOTOTRACE image is required, Principal Component Analysis (Karhunen-Loeve expansion), localized forms of Principal Component Analysis, and Shapiro zerotree embedded Wavelets are among the better advanced techniques.
Today even better Wavelet image compression solutions are available, to meet your specific requirements.
People tend to get carried away bragging about image compression and "compression ratios". Its a good idea to confirm the quality of the image you have after going through a number of compression and decompression cycles! A 500:1 compression ratio doesn't do you much good if the image quality becomes a joke in the process, and this has happened even in widely publicized examples. Quite often a "lossless" image compression approach will be available for your specific requirements using PHOTOTRACE images, meaning you are always able to re-create the original image exactly.
Image Compression is further discussed at the "Waterloo brag zone" Image Compression Home Page at: the University of Waterloo.