ExactImage

The PDF codec was completely rewritten to not only
allow you to dump images into PDFs, but write
arbitrary vector objects and multiple pages. A new
command line tool was added that allows hOCR
annotated HTML to be converted into perfectly laid
out, searchable PDF files. The barcode recognition
was improved regarding CODE25 and CODE28
recognition and checksum computation. De-skew
accuracy was slightly improved and is now exported
through SWIG.

A keybinding was added to the lightweight image
viewer edisplay to invert images. The GIF codec
was fixed to correctly read larger data chunks as
requested by libungif. The RGB to RGBA conversion
buffer use was fixed. The GRAY8 thresholding was
optimized.

Support for circle and ellipse elements was added to the SVG parser. Vector font rendering was cleaned up and optimized and the font on path rendering fixed to use the actual foreground color. A image canvas resize function was exported via the external SWIG API. A possible buffer overrun when parsing numeric values was fixed in the SVG parser.

Support was added to control the barcode scan
directions through the externally available SWIG
interface for Perl, Python, PHP, and Lua. The
contouring and segmentation examples were updated
to compile with the current core library,
resorted, and extended. The Python configure test
was improved to also cover the slightly older
Python version 2.4.

High-quality and high-speed barcode recognition
was added. A new, generic, and better optimizing
image iterator was implemented, resulting in
higher performance of ported algorithms. Support
was added for writing non-RGB BMP files, as well
as writing EPS, PS, PDF, and reading (rendering)
SVG files. Language bindings are now also
available for PHP, Python, and Lua. Compilation
with GCC 4.3 was fixed.