Video frames of a document are captured. The edges of the document are identified from the frames as they appear within a field of view of a camera and a document rectangle for the document is tracked within the field of view. An area of the document determined from the frames is compared against the available area of the camera preview mode and a larger superimposed guiding rectangle is displayed as a guide within the camera preview mode. When a frame for the document is detected as having document edges that substantially fit within the guiding rectangle, the guiding rectangle is removed and an optimal image of the document is captured for subsequent downstream document image processing.