SIFT Image Stitching Technology Based on Music Score Scanning Recognition System

摘要

Music score scanning recognition system, to identify musical notes through hand-held device by scanning five-line staff. Most of the music score scanning equipment equips with linear sensors, acquires the binary image according to the intensity of reflection of light with the irradiation of the front LED. However, because of the influence of illumination and differential rotation of this method, the acquired picture signal is incomplete with lots of noise. It describes a new collection method that is to use video instead of images and cut out several images, using the image stitching algorithm, to improve matching precision with the RANSAC algorithm. Simultaneously, use the weighted smoothing fusion algorithm based on the matching principle. Ultimately, realizes the stitching method based on SIFT feature matching in the music score scanning recognition system. Because of its high matching capacity, problems such as translation between two different images and illumination could be perfectly solved, the pictures look much more verisimilitude and the traditional deficiency could be improved.