ABBYY RTR SDK includes this feature since the last release (R5) of the extended version of the product. In order to get the extended RTR SDK version, please contact our sales at http://rtrsdk.com/licensing/.
In general, RTRTextCaptureService is intended to be used to recognize video frames in real-time video capture. If you want to recognize high resolution single shot images, the way to do this is to use recognizeText from RTRCoreAPI. This method will generally give better results in this scenario.
I also should note that RTRTextCaptureService is configured to work with CMSampleBufferRef containing camera preview images only, not with any CMSampleBufferRef objects which might contain a variety of media types.