Whiteboard sharing: capture, process, and print or email Michael Gormish, Berna Erol, Daniel G. Van Olst, Tim Li, Andrea Mariotti Ricoh Innovations, Inc. 2882 Sand Hill Rd. Ste 115, Menlo Park, CA USA 94025 http://beta.rii.ricoh.com, lastname@rii.ricoh.com, +1.650.496.5700 ABSTRACT Whiteboards support face to face meetings by facilitating the sharing of ideas, focusing attention, and summarizing. However, at the end of the meeting participants desire some record of the information from the whiteboard. While there are whiteboards with built-in printers, they are expensive and relatively uncommon. We consider the capture of the information on a whiteboard with a mobile phone, improving the image quality with a cloud service, and sharing the results. This paper describes the algorithm for improving whiteboard image quality, the user experience for both a web widget and a smartphone application, and the necessary adaptations for providing this as a web service. The web widget, and mobile apps for both iPhone and Android are currently freely available, and have been used by tens of thousands of people. Keywords: Image capture, enhancement, mobile phone, web service 1. INTRODUCTION Whiteboard Usage Whiteboards that print in black and white, spot color, or full color have existed for many years. These whiteboards typically have a writing rotating surface, a line scanner and built-in printer. Such whiteboards are quite common in Japan and produce quite readable text on A4 or letter sized pages. Their value is well recognized with meetings benig preferentially held in rooms with these boards. Obviously, such printers are not available with every whiteboard. Furthermore, even when a whiteboard has a printer it can be out of paper or ink or suffer from paper jams. Whiteboard printers are typically fairly slow, and thus getting everyone in the meeting a copy of the whiteboard can require a trip to the copier. The information from whiteboards is so valuable there are also whiteboards with dedicated cameras mounted above to capture images. The left side of Figure 1 shows a whiteboard with a built-in printer connected just below the pen tray. This whiteboard also has a camera mounted to the wall to capture images of the whiteboard, only the bottom of the mounting bracket of this camera is visible in the picture. While mounted cameras avoid the problem of running out of supplies, again they are relatively expensive, not available everywhere, and don't work with portable whiteboards. Also, it is not immediately obvious how to get images captured by the mounted camera to the meeting participants. For these reasons we initially approached the problem of capturing information from a whiteboard by using a portable digital camera, and then as mobile phone cameras improved in quality we made use of these devices. Pictures taken with portable cameras of whiteboards have lots quality problems: reflection from room lights or camera flash, uneven illumination, and sometimes very dark "white" boards. Low quality images do not serve the purpose of capturing the information from a meeting. Thus we provide an algorithm to improve the quality of whiteboard images and compare it to other techniques. Image Quality Improvements We take the approach of segmenting the image into foreground and background, removing the background (presenting it as white), and enhancing the foreground (presenting as more saturated color). There are a variety of background subtraction methods for different applications. 1 Many background subtraction algorithms make use of a reference image, e.g. taken before the foreground existed, or a sequence of images, and assume that the background is the portion of the image that does not move. While we could have asked users to take an image of the background before writing on