From oleg@ponder.csci.unt.edu Sun Jul  3 12:25:47 1994
From: oleg@ponder.csci.unt.edu (Kiselyov Oleg)
Newsgroups: comp.lang.c++,comp.graphics,alt.graphics.pixutils,comp.sys.mac.programmer
Subject: grayimage C++ classlib posted to comp.sources.misc
Date: 30 Jun 1994 15:19:45 -0500
Organization: University of North Texas
Reply-To: oleg@ponder.csci.unt.edu
NNTP-Posting-Host: sparky.sterling.com
Summary: grayscale image processing class library and verification code
Keywords: grayscale, image processing, C++, class library
Environment: C++, tested_on{G++ 2.2.2+, Symantec 6.0.1+, Sun, HP, Mac, Sequent}

This package has just been posted to comp.sources.misc.

	grayimage_classlib is a C++ class library to do a generic
processing of grayscale images.  It is one part of my image
compression code I've been messing with for almost 3 years. This part
is general enough to be of some interest to folks at large (I
hope). It lets you do a variety of different operations on images,
rectangular areas, etc: say, add two images, compute their "scalar
product", modify pixel values over the entire image or some
rectangular area of it in a _variety_ of ways, equalize the image
histogram, stuff like that.  One function, shift_clip_add(), does
actually the filtration (convolution). Morphological filtration is
implemented as well. I tried to make the code as optimal as I could
think of (without getting into assembly -:). The package can
read/write XWD, Group G TIFF and PGM file formats; actually, the
package understands which file format it's asked to read and selects
the appropriate method automatically.  I demonstrated pieces at Data
Compression Conferences, and some people suggested me submit it into
the public domain.

	The file images.dr tells exactly which modules are in the
package and what they are good for. There are several v*.cc files in
the package: these are verification files, you can compile them and
run to make sure everything works as it's supposed to.  The validation
files can also be used as an example as to how to use the package
(because they test, that is, use all the constructions/classes
/functions /methods of the package, and not once 8:)

	The package has been compiled with gcc/g++ version 2.2.2 and
later on Sun Sparc/SunOS, HP 9000, Sequent Symmetry/Dynix,
Concurrent/RTU, and with Symantec C++ 6.0.1 (through 7.0.2) on Mac.
The package has been in operation for 2.5 years.

	I also have to mention that the image package I'm submitting
is just the very bottom layer of the image and video processing stuff
I've been working on. The bottom layer is meant to give the basic
working environment rather to do something fancy with the image
(though, BTW, function shift_clip_add() makes the convolution in two
stmts; and this lets you do a whole lot of the fancy things in the
digital darkroom). So, there is much more to the story. BTW, it also
means that I'm constantly adding to the package (as I come across a
need for some function, option, etc), so your comments/suggestions
will be highly appreciated.

	And now, here's a quick glance at the package capabilities:

// Note! This is the *correct* code; Put it within main(){}, and it will
// compile, run and finish without errors

// Pixel operations

IMAGE im1(256,256,8); IMAGE im2(im1);
im1 = 1; im2 = im1; im1 *= 4;		// Assigning a value to every pixel
assert( im1 == 4 ); assert( im1 >0 );	// Checking up the value of every pixel
im1.invert(); im1.clear(); im2 <<= 2; im1 = 5; 
im1 &= 0xfe;  assert( im1==im2);	// Modifying every pixel
im1.clear(); im1 -= im2; im2 = im1; 	// Subtracting two images
					// Applying *any* function to every
im1.abs(); im2.apply(::abs); assert(im1 == im2); // pixel (abs() in the ex.)
assert( im1 * im1 == norm_2_sqr(im1) );	// Computing the "scalar product"
					// and image norm

// Square and rectangular parts of image

im1 = 2*pattern; im2 = pattern;
rowcol rightbottom(im1.q_nrows()-1,im1.q_ncols()-1);
rowcol center(im1.q_nrows()/2,im1.q_ncols()/2);
					// Modifying pixels only within the
					// lower left quadrant
im1.rectangle(center,rightbottom) -= 2*pattern;
assert( !(im1 == 2*pattern) && !(im1 != 2*pattern) );
					// Another way of doing this
im1.shift_clip_add(center,2,im2);  assert(im1 == 2*pattern);

// Miscellanea
					// Note "expanded" file name of the
					// image
IMAGE raw_image("zcat ../../pictures/lenna.xwd.Z |");
//  IMAGE image(IMAGE::Shrink,raw_image);
IMAGE image = raw_image.square_of(256,rowcol(120,100));
image.write("/tmp/ior","Original image");
image.display("Image to display");