Method and system for single pass optical character recognition

Abstract

A computer implemented method of performing single pass optical character recognition (OCR) including at least one fully convolutional neural network (FCN) engine including at least one processor and at least one memory, the at least one memory including instructions that, when executed by the at least processor, cause the FCN engine to perform a plurality of steps. The steps include preprocessing an input image, extracting image features from the input image, determining at least one optical character recognition feature, building word boxes using the at least one optical character recognition feature, determining each character within each word box based on character predictions and transmitting for display each word box including its predicted corresponding characters.

1 . A computer implemented method of performing single pass optical character recognition (OCR) including at least one fully convolutional neural network (FCN) engine including at least one processor and at least one memory, the at least one memory including instructions that, when executed by the at least processor, cause the processor to perform the steps of:
 preprocessing an input image, wherein the input image includes machine printed characters;   extracting image features from the input image using a plurality of convolutional layers included in the FCN engine, wherein image features of the input image are extracted at each convolutional layer in the FCN engine;   aggregating the image features extracted at each convolutional layer of the FCN engine;   determining at least one optical character recognition (OCR) feature from the aggregated extracted image features;   building word boxes using the determined at least one optical character recognition feature;   determining each character within each word box based on character predictions; and   transmitting for display each word box including its corresponding determined characters.   
     
     
         2 . The computer implemented method of  claim 1 , wherein the step of preprocessing the input image includes padding the image with zero values to a specified dynamic denominator. 
     
     
         3 . The computer implemented method of  claim 1 , wherein the step of extracting image features from the input image includes filtering the input image at each convolutional layer in the FCN engine thereby extracting features from the image at each convolutional layer. 
     
     
         4 . The computer implemented method of  claim 1 , wherein the step of determining at least one OCR feature from the aggregated extracted image features includes calculating a wordiness score. 
     
     
         5 . The computer implemented method of  claim 1 , wherein the step of determining at least one OCR feature from the aggregated extracted image features includes calculating a character gap score. 
     
     
         6 . The computer implemented method of  claim 1 , wherein the step of determining at least one OCR feature from the aggregated extracted image features includes calculating character predictions. 
     
     
         7 . The computer implemented method of  claim 1 , wherein the step of building word boxes includes determining boundaries between words. 
     
     
         8 . The computer implemented method of  claim 1 , wherein the step of building word boxes includes determining centers of words. 
     
     
         9 . The computer implemented method of  claim 1 , wherein the step of determining each character within each word box based on character predictions includes determining character boxes for each character in the word box by delineating each character in the word box by character gaps. 
     
     
         10 . The computer implemented method of  claim 9 , wherein a number of character predictions are made for each character box and the number of character predictions are aggregated. 
     
     
         11 . The computer implemented method of  claim 10 , wherein a character determination is made by selecting a highest prediction from a result of multiplying the aggregated character predictions by a character weighting index. 
     
     
         12 . A system for performing single pass optical character recognition (OCR), the system comprising:
 a network interface configured to receive an input image, the input image includes machine printed characters;   a memory configured to store electronic program guide data and computer executable instructions;   an FCN engine including at least one processor configured to execute the computer executable instructions to:
 preprocess the input image by padding the image with zero values to a specified dynamic denominator; 
 extract image features from the input image using a plurality of convolutional layers included in the FCN engine, wherein image features of the input image are extracted at each convolutional layer of the FCN engine; 
 aggregate the image features extracted at each convolutional layer of the FCN engine; 
 determine at least one optical character recognition (OCR) feature from the aggregated extracted image features; 
 build word boxes using the determined at least one optical character recognition feature; 
 determine each character within each word box based on character predictions; and 
 transmit for display each word box including its predicted corresponding characters. 
   
     
     
         13 . The system of  claim 12 , wherein the determined at least one OCR feature is a calculated wordiness score. 
     
     
         14 . The system of  claim 12 , wherein the determined at least one OCR feature is a calculated character gap score. 
     
     
         15 . The system of  claim 12 , wherein the determined at least one OCR feature is calculated character predictions. 
     
     
         16 . The system of  claim 12 , wherein building word boxes includes determining boundaries between words. 
     
     
         17 . The system of  claim 12 , wherein building word boxes includes determining centers of words. 
     
     
         18 . The system of  claim 12 , wherein determining each character within each word box based on character predictions includes determining character boxes for each character in the word box by delineating each character in the word box by character gaps. 
     
     
         19 . The system of  claim 12 , wherein a number of character predictions are made for each character box and the number of character predictions are aggregated. 
     
     
         20 . The system of  claim 12 , wherein a character determination is made by selecting a highest prediction from a result of multiplying the aggregated character predictions by a character weighting index.