A goal-oriented verification-based approach for target text line extraction from a document image captured by a pen scanner

We present a goal-oriented verification-based approach for target text line extraction from a document image captured by a pen scanner. Given a binary image, a series of processing steps are invoked adaptively, guided by the text line verification result in the preceding step. Each step adopts a strategy that is most effective for dealing with the problem concerned. Consequently, the target text line can be extracted in a more efficient and reliable way depending on the nature of the captured image. The effectiveness of the above approach is confirmed by a benchmark test.