Objective Caml image processing library
CamlImages is an image processing library, which provides
In addition, the library can handle huge images that cannot be (or can hardly be) stored into the main memory (the library then automatically creates swap files and escapes them to reduce the memory usage).
Additional example programs are supplied in the examples directory of the library:
Get the source tar ball, decompress it, and read the file INSTALL. You can also access the read-only repository which contains the latest developing (but not yet released) version.
CamlImages supports the following color models:
For each color model, a corresponding module is provided. For example, use the module Rgb24 if you want to access 24bit depth full color images.
CamlImages supports loading and saving of the following file formats:
For each image format, we provide a separate module. For instance, use the Tiff module to load and save images stored in the tiff file format. If you do not want to specify the file format, you can use Image.load: this function automatically analyzes the header of the image file at hand and loads the image into the memory, if the library supports this format.
CamlImages also provides an interface to the internal image format of O’Caml’s Graphics library (this way you can draw your image files into the Graphics window).
You can also draw strings on images using the Freetype library, which is an external library to load and render TrueType fonts.
The modules with names beginning with the letter ‘o’ contains the object class interface for CamlImages.
When you create/load a huge image, the computer memory may not be sufficient to contain all the data. (For example, this may happen if you are working with a scanned image of A4, 720dpi, 24bit fullcolor, even if you have up to 128Mb of memory!) To work with such huge images, CamlImages provides image swaps, which can escape part of the images into files stored on the hard disk. A huge image is thus partitioned into several blocks and if there is not enough free memory, the blocks which have not been accessed recently are swapped to temporary files. If a program requests to access to such a swapped block, the library silently loads it back into memory. By default, image swapping is disabled, because it slows down the programs. To activate this function, you have to modify Bitmap.maximum_live and Bitmap.maximum_block_size:
For example, if you do not want to use more than 10M words (that is 40Mb for a 32bit architecture or 80Mb for a 64bit architecture), set Bitmap.maximum_live to 10000000. You may (and you should) enable heap compaction, look at the GC interface file, gc.mli, in the standard library for more details (you should change the compaction configuration). Bitmap.maximum_block_size affects the speed and frequency of image block swapping. If it is larger, each swapping becomes slower. If it is smaller, more swappings will occur. Too large and too small maximum_block_size, both may make the program slower. I suggest to have maximum_block_size set to !Bitmap.maximum_live / 10. If you activated image swapping, cache files for unused swapped blocks will be removed automatically by Caml GC finalization, but you may free them explicitly by hand also. The functions and methods named “destroy” will free those blocks. The swap files are usually created in the /tmp directory. If you set the environment variable “CAMLIMAGESTMPDIR”, then its value replaces the default “/tmp” directory. The temporary files are erased when the program exits successfully. In other situations, for instance in case of spurious exception, you may need to erase temporary files manually.
You can find some examples in the examples directory including image auto-cropping, resizing and an image viewer with the lablgtk library.