Google Inc.

Other

Raymond W Smith

Luc Vincent

Andrea Lynn Frome

Bo Wu

Alessandro Bissacco

Shlomo Reuben Urbach

Kong Man Cheung

Sergey Ioffe

David Wasserman

Zongyi Liu

Ranjith Unnikrishnan

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for implementing long-short term memory cells with saturating gating functions. One of the systems includes a first Long Short-Term Memory (LSTM) cell, wherein the first LSTM cell is configured to, for each of the plurality of time steps, generate a new cell state and a new cell output by applying a plurality of gates to a current cell input, a current cell state, and a current cell output, each of the plurality of gates being configured to, for each of the plurality of time steps: receive a gate input vector, generate a respective intermediate gate output vector from the gate input, and apply a respective gating function to each component of the respective intermediate gate output vector, wherein the respective gating function for at least one of the plurality of gates is a saturating gating function.

Long short-term memory cells with saturating gating functions

A system and method is provided for automatically recognizing building numbers in street level images. In one aspect, a processor selects a street level image that is likely to be near an address of interest. The processor identifies those portions of the image that are visually similar to street numbers, and then extracts the numeric values of the characters displayed in such portions. If an extracted value corresponds with the building number of the address of interest such as being substantially equal to the address of interest, the extracted value and the image portion are displayed to a human operator. The human operator confirms, by looking at the image portion, whether the image portion appears to be a building number that matches the extracted value. If so, the processor stores a value that associates that building number with the street level image.

System and method of determining building numbers

A system and method of assigning diacritics in an electronic image using optical character recognition (OCR) is disclosed. In one example, the method comprises analyzing, by a computer system, the electronic image to generate a plurality of bounding blocks associated with text lines within the electronic image. The method further comprises establishing a plurality of bounding boxes for diacritics and base text with the electronic image. The method also comprises determining a distance from a diacritic to a nearest base text character and a nearest text line. The method also comprises evaluating a base box distance and the nearest text line distance to assign the diacritic to a correct text line in the electronic image.

Detection of diacritics in OCR systems with assignment to the correct text line

A system and method of extracting formulas in an electronic image of a document using optical character recognition (OCR) is disclosed. In one example, the method comprises analyzing the electronic image, including a plurality of text lines, to generate a plurality of bounding blocks, each bounding block associated with a text line detected within the electronic image, searching the plurality of text lines to detect at least one character matching one of a plurality of character groups, calculating a symbol density of each of the plurality of character groups for each of the plurality of text lines, and classifying each of the plurality of text lines as at least one of an equation block type, an inline equation block type, and a descriptive block type, based on the symbol density, wherein each of the plurality of text lines classified as the equation block type is extracted.

System and method for detecting equations

A page layout module receives a page image displaying text in an unknown script and unknown orientation, determines a text section in the received image and transmits the text section to an orientation and script module. The orientation and script module comprises a training module, a classifier and a recognition module. The training module trains the classifier to identify connected components that include a connected portion of one or more characters of text. The recognition module uses the trained classifier to identify in the received text section a set of connected components. The recognition determines the likely orientation and script for the connected components and then uses the determined information to determine the orientation and script for the text section. The determined orientation and script for the text section is transmitted to the OCR module. The OCR module uses the determined orientation and script to recognize text in the text section.

Text script and orientation recognition

Systems, methods and computer program products on storage devices for shape clustering and applications in processing various documents, including an output of an optical character recognition (OCR) process. The output of an OCR process is classified into a plurality of clusters of clip images and a representative image for each cluster is generated to identify clusters whose clip images were incorrectly assigned character codes by the OCR process.

Techniques for shape clustering and assignment of character codes in post optical character recognition processing

Physical page layout analysis for optical character recognition is performed. A physical page layout analysis method finds constituent parts of an image and gives an initial data-type label, such as text or non-text. Within the text data, connected components are identified and analyzed. Tab-stops are detected from groups of edge-aligned connected components. The detected tab-stops are used to deduce the column layout of the page by finding column partitions. The column layout is then applied to find the polygonal boundaries of and a reading order of regions containing flowing text, headings, and pull-outs.

Physical page layout analysis via tab-stop detection for optical character recognition

Shape clustering in post optical character recognition processing

Systems, methods, and computer program products on storage devices for shape clustering and applications in processing various documents, including an output of an optical character recognition (OCR) process. The output of an OCR process is classified into a plurality of clusters of clip images and a representative image for each cluster is generated to identify clusters whose clip images were incorrectly assigned character codes by the OCR process.

Techniques for shape clustering and applications in processing various documents, including an output of an optical character recognition (OCR) process. The output of an OCR process is classified into a plurality of clusters of clip images and a representative image for each cluster is generated to identify clusters whose clip images were incorrectly assigned character codes by the OCR process.

Systems, methods and computer program products for shape clustering and applications in processing various documents, including an output of an optical character recognition (OCR) process. Clip images defined in a received OCR output are classified into a plurality of clusters of clip images. Clip images in each of the plurality of clusters are processed to generate a cluster image for each cluster. Shape differences between the cluster images of a first cluster and a second cluster and between the cluster images of the first cluster and a third cluster are used to determine a level of confidence in one or more first OCR character codes assigned to the first cluster.

Techniques for shape clustering and applications in processing various documents, including an output of an optical character recognition (OCR) process.

Growing community of inventors

Mountain View, CA, United States of America

Raymond W Smith