For your scenario we usually recommend to use the processImage method with the textExtraction profile. This profile is suitable for extracting all text from the input image, including small text areas of low quality. Also it could be helpful to export text and its coordinates to XML and then extract the necessary information on your side using text coordinates.
The field-level recognition cannot be used for your scenario because we can recognize fields on a page only with the specified coordinates of each these fields. Thus field-level recognition can be used in case you process some static forms, not when the images are quite different each time.
answered 02 Jul '15, 15:56
Oksana Serdyuk ♦♦