The hand-written information input system "Denshi-Pen" consists of special output paper with code images printed thereon and the pen used to scan this paper. The information encoded in the code images includes page IDs for identifying the paper being used and the paper's horizontal and vertical coordinates. Taking a notebook for example, this information indicates the page and the point where something is written. When the pen is used to write text or figures, the code images filmed by the built-in miniature camera are taken into the pen and converted into a bit sequence by image processing technology. Decoding this bit sequence identifies the page ID and location of the pen point.
Code images must be inconspicuous enough so as not to interfere with the user's writing, thereby enabling both the special paper with embedded code and the pen to be used just like normal paper and a ballpoint pen. The preciseness of the coordinates is also essential for the accurate identification of text and figures. This can be achieved, for example, by dividing the paper surface into small sections and then encoding the locations of the sections in the paper and a page ID in each section. However, in this method, the size of a section determines the preciseness of the coordinates. It is difficult to make the size of each section small enough for accurately identifying written characters. In addition to detecting where each section is divided, the rotation of the paper must also be detected upon decoding, assuming that the paper may be used upside down or sideways, but code images will be too noticeable should a special synchronization pattern for detecting rotation also be inserted in each section.
In our two-dimensional coordinate encoding technology, information is represented by combining a certain number of dots arranged in a square area called a "code block" as shown in Fig. 1(a). The coordinates and page ID are encoded by the arrangement of code blocks as shown in Fig. 1(b). For the code blocks representing coordinates, a maximum length sequence—a type of numerical sequence characterized by unique subsequences not matching any other subsequence—is used. Code blocks are arrayed in the horizontal and vertical directions of paper, so that the sequence of blocks will be the maximum length sequence representing the coordinate information on paper. Therefore, a given code block's location can be identified from any part of the array of blocks. In code blocks representing the page ID information, the same page information is cyclically encoded in the horizontal and vertical directions of the paper. The code blocks used for synchronization simultaneously detect where code blocks are divided and whether the paper is rotated. As one set of blocks contains one set of basic information, a coordinate point and page ID can be decoded wherever a set of blocks is located, because the same set of information is included when the set is the same size. This method disperses the dots in code images and makes code images inconspicuous, as no special synchronization pattern is required and there is the same number of dots in each code block. This technology has enabled the identification of one point written on a page amongst about a quadrillion sheets of A0 paper with an accuracy of 0.17 mm.
Fig. 1: Structure of Two-dimensional Coordinate Encoding
The area within the red square includes one set of basic information.
As long as the size is the same, as shown in the square area indicated by the red dashed lines,
the area includes the required blocks regardless of location. Therefore,
a coordinate point and a page ID can be identified from anywhere.