In this paper, we propose a method to select the most informative iris image from video stream of capture equipment. This method is a combination of three steps. The first is eye existence verifying which based on Haar cascade classifier. This step ensures the accepted image is an eye-image without users'' co-operation when data acquisition performs. The next steps are locating components of eye and pupil tracking. Finally, we use computed parameters from the second step to select the most informative iris image. Our method can be comparable to the previous approaches which are complex or require hardware support. It makes real-time iris based systems work effectively in data acquisition with acceptable speed and high accuracy.