Ncr Corporation

United Parcel Service of America, Inc.

Jie Zhu

Michael C Moed

Daniel B Seevers

Izrail S Gorian

Techniques for identifying labels appearing in images are described. A digital image is analyzed to identify individual components. Each of the individual components is analyzed to determine whether or not it is a text component by comparing it against criteria such as size, aspect ratio, and proximity to other components. Each component identified as a text component is compared against criteria such as size in order to identify it as a label or not. Location coordinates of each label are stored in association with the label and optical character recognition is performed on the labels. Once the labels are identified, each image can be used as an online catalog page. For example, an image may be used to construct a web page containing pictures of available products with each label serving as a hypertext link to retrieve further information about the product or to enter an order for the product. Automatically identifying labels simplifies the conversion of preexisting paper catalog pages to online catalog pages or similar digitized images.

Methods and apparatus for locating and identifying text labels in digital images

A produce recognition method which determines an optimal number of candidate identifications in a candidate identification list. The method includes the steps of obtaining produce data associated with a produce item, determining distances between the produce data and reference produce data, determining confidence values from the distances, determining first confidence values which are greater than a threshold confidence value, displaying candidate identifications associated with the first confidence values, and recording an operator choice of one of the candidate identifications.

Produce recognition method

A triggering method for a produce recognition system which uses historical ambient light level readings. The method includes the steps of obtaining first ambient light levels from an ambient light sensor of the produce data collector with a data collection aperture covered, obtaining second ambient light levels from the ambient light sensor with the data collection aperture uncovered, determining a threshold ambient light level from the first ambient light levels and a difference between the first and second ambient light levels, obtaining a third ambient light level from the ambient light sensor with a produce item adjacent the data collection aperture, comparing the third ambient light level to the threshold ambient light level, and capturing data associated with the produce item if the third ambient light level is less than the threshold ambient light level.

Triggering method for a produce recognition system

Methods and apparatus for gray image based text identification. A gray image of a document is preferably subsampled to reduce the amount of information to be processed, while retaining sufficient information for successful processing. The subsampled image is subjected to preprocessing to remove horizontal and vertical lines. The image is then subjected to a morphological open operation. The image is then segmented to separate foreground and background information to produce a foreground image. Region filtering and merging are performed on the foreground image. Region features are then extracted and region identification performed. Homogenous regions are grouped and noise elimination performed, resulting in a number of small regions of known types. Optical character recognition can then be performed on each of the regions. The use of the information provided by variations in pixel lightness and darkness enables text identification to proceed quickly and efficiently.

Methods and apparatus for gray image based text identification

A method and system including a high speed computer algorithm for rotating images stored in a computer memory, such as those captured by a CCD array. The method and system may be used with an over-the-belt optical character recognition (OCR) reader that captures a pixelized image of the surface of a parcel as it moves along a conveyor. A run-length encoded output image is obtained from a run-length encoded input image, wherein the output image is a rotated version of the input image. A predefined pixel mapping sequence for a rotation angle with a rational tangent is used to map image pixel data. Mapping image pixel data using a predefined pixel mapping sequence allows rotation without floating point computations, and allows the use of a reverse pixel mapping technique to map output runs from an input image in order to avoid the creation of holes or artifacts in the output image. Mapped output runs are scaled to preserve the aspect ratio of the rotated image. The order of processing the elements of the input run-length input table is selected so as to directly create a run-length encoded output table.

Growing community of inventors

Fairport, NY, United States of America

Jie Zhu