Nowadays digital scanning techniques are widely used in many fields such as medical imaging and office affairs. In order to achieve fast and accurate text image acquisition and recognition, a new digital scanning pen as well as some image processing techniques is developed in this paper. An adaptive image division algorithm based on potential function clustering is achieved by the combination of total threshold and local threshold, which can eliminate the influence of illumination and reflection. Redundant scanning frames are eliminated by exclusive OR algorithm. This is essential to reduce data stream and perform following processing steps. The scanning image mosaic technique is achieved by projection correlation algorithm, and an automatic line division algorithm is developed for the sake of convenient character recognition. When being connected with a computer, this new digital scanning penpsilas scanning and recognition speed can achieve 16 cm/s. Its recognition rate of English characters and Chinese characters are 99 percents and 95 percents respectively. In future it may be used in foreign language learning, business affairs, news and publication.