Fast Capture of Sheet Music for an Agile Digital Music Library

A personal digital music library needs to be “agile”, that is, it needs to make it easy to capture and index material on the fly. A digital camera is a particularly effective way of achieving this, but there are several issues with the quality of the captured image, including distortions in the shape of the image due to the camera not being aligned properly with the page, non-planarity of the page, lens distortion from close-up shots, and inconsistent lighting across the page. In this paper we explore ways to improve the quality of music images captured by a digital camera or an inexpensive scanner, where the user is not expected to pay a lot of attention to the process. Such pre-processing will significantly aid Music Information Retrieval indexing through Optical Music Recognition, for example. The research presented here is primarily based around using a Fast Fourier Transform (FFT) to determine the orientation of the page. We find that a windowed FFT is effective at correct...

• Type: Conference
• Year: 2005
• Where: ISMIR
• Authors: Richard Lobb, Tim Bell, David Bainbridge
• Keywords: Digital Camera, Distortion, Information Retrieval, ISMIR 2005, Personal Digital Music