James Russell Rinker

Text recognition of low-resolution document images

Cheap and versatile cameras make it possible to easily and quickly capture a wide variety of documents. However, low resolution cameras present a challenge to OCR because it is virtually impossible to do character segmentation independently from recognition. In this paper we solve these problems simultaneously by applying methods borrowed from cursive handwriting recognition. To achieve maximum robustness, we use a machine learning approach based on a convolutional neural network. When our system is combined with a language model using dynamic programming, the overall performance is in the vicinity of 80-95% word accuracy on pages captured with a 1024/spl times/768 webcam and 10-point text.

A foreground-background separation algorithm for image compression

Many bitmap documents are composed by the superposition of layers with pictures and text. These documents do not compress well using image compression algorithms such as JPEG-2000, because text introduces sharp edges on top of the smooth surfaces typically found in natural images. Similarly, compression algorithms for text facsimiles, such as JBIG2, are not suited for color or gray level images. In this paper the SLIm system for separating text and line drawing from background images, in order to compress both more effectively is proposed. This approach differ from previous ones such as DjVu, Tiff-FX, and MRC, by being extremely simple and fast, while yielding close to state-of-the-art compression performance. Results show that the SLIm compression performance is attractive for many applications.

Automatic Fax Routing

We present a system for automatic FAX routing which processes incoming FAX images and forwards them to the correct email alias. The system first performs optical character recognition to find words and in some cases parts of words (we have observed error rates as high as 10 to 20 percent). For all these “noisy” words, a set of features is computed which include internal text features, location features, and relationship features. These features are combined to estimate the relevance of the word in the context of the page and the recipient database. The parameters of the word relevance function are learned from training data using the AdaBoost learning algorithm. Words are then compared to the database of recipients to find likely matches. The recipients are finally ranked by combining the quality of the matches and the relevance of the words. Experiments are presented which demonstrate the effectiveness of this system on a large set of real data.