Ocular torsion determination using search coils was a preferred method in a past few years. But advantages of the digital image processing techniques allows us to refuse search coils and use video-based techniques instead. The manufactures of the commercial video-based systems declare high accuracy of their product, but they don’t publish result analysis in public to improve it. We chose to use E. Greon’s (1997) suggested method with additional pattern matching using cross correlation. The landmarks of iris texture is used to track from frame to frame, so its important to enhance them. The researches helped to find out what is the most suitable iris processing and what are the optimal correlation parameters. The results showed, that a polar cross correlation can be applicable for ocular torsion measurement.