Abstract: A method for compressing a digitized image of a document using optical character recognition (OCR). The method includes performing optical character recognition (OCR) on the digitized image, identifying, based, at least in part, on a result of the performing step, a plurality of classes of characters comprised in the image, each the class of characters having an associated character value and comprising at least one character, pruning each class of characters, thereby producing information describing the plurality of classes of characters and a residual image, and utilizing the information describing the plurality of classes of characters and the residual image as a compressed digitized image in further processing. Related methods and apparatus are also disclosed.

Abstract: A method of binarization used in an OCR system involves in determining text pixels by checking, for each pixel, that the difference between its value and the values of a plurality of pixels located at a predetermined distance therefrom is greater than a relative threshold corresponding to the difference in intensities between the text and the background of the image, subsampling the image at a rate corresponding to at least two pixels in order to detect kernels of text, and then binarizing the image pixels only in tiles of several stroke width sides containing text kernels by using in each tile, an absolute threshold estimated in that tile.

Abstract: A method for automatic sorting includes receiving an item in a sequence of items to be sorted, each such item marked with a respective machine-readable identifying code and with respective characters in a location relative to the code that varies from one item to another in the sequence. A position of the code on the item is determined and, responsive to the position of the code, the location of the characters on the item is found. The characters are processed to determine a destination of the item.

Abstract: A method for locating a structured field in a gray-scale image of an object, including choosing a plurality of anchor points in the image, each anchor point having a gray-scale value associated therewith. For each anchor point there is determined a horizontal variation dependent on a difference between the gray-scale value of the anchor point and the gray-scale value of a horizontally neighboring anchor point, and there is also determined a vertical variation dependent on a difference between the gray-scale value of the anchor point and the gray-scale value of a vertically neighboring anchor point. Those anchor points whose vertical and horizontal variations obey a first or a second predefined condition are defined as vertically or horizontally dominant respectively.

Abstract: The invention locates address indicia on digitized images of mail pieces by dividing the mail piece images into blocks; identifying rows of blocks that are adjacent along a first direction and that have similar defined image features by comparing such image features within adjacent blocks; assembling at least some of the identified rows of blocks that are adjacent along a second direction perpendicular to the first direction into address block candidate regions; and selecting the address block by scoring the various address block candidate regions on the basis of defined criteria. The criteria initially score the candidate regions based on closeness to the center of the rectangle and then reduce this initial score if an edge of the region is close to an edge of the image, the region contains a small or very high number of text rows, or the aspect ratio between the width and height of the region is high.