Re: Apple II Books

[mjmahon@aol.com (Michael J. Mahon)]

Eric Smith replied:

>Paul R. Santa-Maria wrote:
>> The biggest scanning mistake I see folks making is using
>> greyscale instead of monochrome.  This increases the file
>> size and makes the scan useless for OCR.
>
>Michael J. Mahon wrote:
>> Scanning in grayscale, even if only to 2- or 3-bit depth, 
>> should preserve some sub-pixel resolution which can
>> be recovered by thresholding at higher resolution if
>> desired.
>
>That's true, except that people typically store greyscale images using
>JPEG lossy compression, which was designed for continuous tone images
>rather than text and line art.  JPEG discards high frequency
>information.  Guess what the frequency spectrum of text and line art
>looks like?  JPEG makes the edges blurry and indistinct, and introduces
>other artifacts within the 8x8 pixel blocks.  This makes it MUCH worse
>for OCR or even just for viewing.

True, JPEG is not the best choice for text an line art.

I was actually thinking of GIF, since it allows the pallette of grays to
be chosen.

Of course, in monochrome, it's hard to beat run-length coding.  ;-)

>Lossless greyscale coding (such as is available with JPEG 2000 wavelet
>compression) would be worthwhile, but typically doesn't get very good
>compression ratios, and software supporting it is not widely available.

I haven't had much experience with wavelet compression, but I have
heard that the artifacts its lossy forms introduce are much less
noticeable (at least in pictures) than the DCT artifacts of JPEG.

-michael

Check out amazing quality sound for 8-bit Apples on my
Home page:  http://members.aol.com/MJMahon/