Integrated text location and reading
Quixate’s main product is an integrated text-locating and text-reading system.
It finds the lines of text — or even isolated
words — in a photograph
and then reads each line that it finds. The result is a searchable, indexable
database of those lines of text.
The system is capable of finding and reading text:
- in a wide range of fonts;
- at any orientation;
- of any size from a few pixels high to hundreds of pixels high;
- set along a straight line or along an arbitrary curve;
- that is unevenly lit;
- distorted by the camera lens, by noise, by compression, or by the effect of perspective.
See our gallery for some examples of what
the system can do. If you want to know how well it can do on your own
images, send us some samples and we’ll show you.
The input to the system is a photograph, either as a JPEG or in almost any other
common format. The output is a plain text database of the lines of text
giving their contents along with their exact positions and sizes, plus
a score indicating the system’s confidence in the results.
The output is
easy to search and index, and can readily be converted into
other formats as needed. You can read the details of how it all works
The system is written in C++ and will run without difficulty on
almost any desktop computer. It is designed to be fast, flexible and frugal
with machine resources for easy portability to other platforms,
especially embedded systems.
Flexible licence options are available to suit the needs of your particular application.