US5239593A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Optical pattern recognition using detector and locator neural networks
Assignee: Verizon Patent and Licensing Inc
•
Filing: 1991-04-03
•
Granted: 1993-08-24
•
Expires: 2011-04-03
Abstract
Abstract A system for performing optical pattern recognition includes a first detector neural network for detecting the presence of a particular optical pattern in an input image and a second locator neural network for locating and/or removing the particular optical pattern from the input image. The detector network and the locator network both comprise nodes which can take on the -1, +1, or undefined states. The nodes are arranged in layers and each node in a layer has a location corresponding to a pixel in the input image. A particular application of this neural network is in finding the amount field on a check and removing the line which borders the amount field.
Inventors
- 👤 Ben S. Wittner
- 👤 Keith Loris
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-01 21:39
CPC Classifications
G; G06; G06V; G06V30/00; G06V30/10; G06V30/14; G06V30/146; G06V30/148; G06V30/155
Claims Analysis
Total Claims: 23
Claim 1
Independent Claim
160 words
A system for processing an image comprising a generator for generating electronic signals representative of a digitized image, and a recognition engine for receiving the digitized image generated by said generator and for electronically converting information contained in said digitized image into symbolic form, and a memory for storing said information in symbolic form, said recognition engine including: a first detector neural network comprising a plurality of detector nodes selectively interconnected only by forward feeding communication channels for receiving said digitized image produced by said generator as an input and for outputting an indication as to whether or not a particular optical pattern is present in said digitized image, and a second locator neural network comprising a plurality of locator nodes including one locator node corresponding to each detector node in said detector network, for receiving as an input the output of said detector network and for outputting an indication of the location of said optical pattern in said image.
Claim 2
Independent Claim
160 words
A system for processing an image comprising a generator for generating electronic signals representative of a digitized image, and a recognition engine for receiving the digitized image generated by said generator and for electronically converting information contained in said digitized image into symbolic form, and a memory for storing said information in symbolic form, said recognition engine including: a first detector neural network comprising a plurality of detector nodes selectively interconnected only by forward feeding communication channels for receiving said digitized image produced by said generator as an input and for outputting an indication as to whether or not a particular optical pattern is present in said digitized image, and a second locator neural network comprising a plurality of locator nodes including one locator node corresponding to each detector node in said detector network, for receiving as an input the output of said detector network and for outputting an indication of the location of said optical pattern in said image.
Claim 3
Dependent Claim
19 words
The system of claim 1 wherein said system includes means for removing pixels comprising said pattern from said image.
Claim 4
Dependent Claim
24 words
The system of claim 1 wherein said generator for generating a digitized image comprises a scanner for receiving a document and scanning the document.
Claim 5
Dependent Claim
30 words
The system of claim 1 wherein said digitized image is a bitonal image comprising pixels each of which is represented by a first predetermined state or a second predetermined state.
Claim 6
Dependent Claim
26 words
The system of claim 4 wherein in each layer of said detector neural network there is one detector node for each pixel in said digitized image.
Claim 7
Dependent Claim
38 words
The system of claim 5 wherein each detector node in said detector neural network has one or more inputs and an output which can take on a first predetermined state, a second predetermined state, or an undefined state.
Claim 8
Dependent Claim
75 words
The system of claim 6 wherein said detector neural network comprises an input layer, wherein the detector nodes in the input layer are set to the states of the corresponding pixels in the digitized image, a plurality of intermediate layers, and an output layer arranged in sequence, the detector nodes in the input and intermediate layers being connected by said forwarding feeding communication channels to particular detector nodes of the next layer in the sequence.
Claim 9
Dependent Claim
17 words
The system of claim 7 wherein two or more of the intermediate layers are arranged in parallel.
Claim 10
Dependent Claim
38 words
The system of claim 7 wherein the detector nodes in the output layer of the detector network indicate the presence of said pattern in said digitized image but not the location of said pattern in said digitized image.
Claim 11
Dependent Claim
54 words
The system of claim 9 wherein each node in said intermediate and output layers of said detector network comprises summing means value for forming a weighted sum of input state values, subtracting means for subtracting a threshold value from the weighted sum, and means for determining the algebraic sign of the subtracting means output.
Claim 12
Dependent Claim
85 words
The system of claim 9 wherein said locator neural network comprises an input layer corresponding to the output layer of the detector network, a plurality of intermediate layers corresponding to the intermediate layers of said detector network, and an output layer corresponding to the input layer of the detector network, each layer in said locator network comprising one locator node which corresponds to each detector node in the corresponding layer of the detector network and which also corresponds to one pixel in said digitized image.
Claim 13
Dependent Claim
20 words
The system of claim 11 wherein said locator nodes in said locator network can take on said first predetermined state.
Claim 14
Dependent Claim
35 words
The system of claim 12 wherein the locator nodes in the input layer of the locator network take on the same states as the corresponding detector nodes in the output layer of the detector network.
Claim 15
Dependent Claim
37 words
The system of claim 13 wherein the states of the locator nodes in the intermediate layer and output layer of the locator network are set in the reverse order of the corresponding layers in the detector network.
Claim 16
Dependent Claim
82 words
The system of claim 14 wherein the state of a specific locator node in an intermediate or output layer of the locator network is set to said first predetermined state a) if the output of the specific detector node corresponding to the specific locator node has the first predetermined state, and b) if the state of a locator node which corresponds to a detector node which receives a positively weighted input from the specific detector node is in the first predetermined state.
Claim 17
Dependent Claim
31 words
The system of claim 15 wherein the locator nodes in the output layer of the locator network which take said first predetermined state correspond to the pixels of the optical pattern.
Claim 18
Dependent Claim
35 words
The system of claim 1 wherein said generator for generating a digitized image comprises a scanner for scanning and digitizing a document, and field finding means for finding a particular field in the digitized document.
Claim 19
Dependent Claim
40 words
The system of claim 17 wherein said business document is a check, wherein said field finding means finds an amount field on said check, and wherein said optical pattern is a line which forms a border of said amount field.
Claim 20
Dependent Claim
19 words
The system of claim 1 wherein each layer of detector nodes in said detector network has the convolutional property.
Claim 21
Independent Claim
167 words
A method for processing an image comprising the steps of: generating electronic signals representative of a digitized image, and electronically converting information contained in said digitized image into symbolic form using an optical recognition engine, said step of electronically converting comprising detecting and locating an optical pattern in said digitized image by the steps of inputting said digitized image into a first detector neural network comprising a plurality of detector nodes selectively interconnected only by forward feeding communication channels and outputting from said detector neural network an indication as to whether or not said optical pattern is present in said image, and inputting the output of said detector network into a second locator neural network comprising a plurality of locator nodes including one locator node corresponding to each detector node and outputting from said locator network an indication of the location of said pattern in said digitized image, said method further comprising the step of storing said symbolic information outputted by said recognition engine into a memory.
Claim 22
Independent Claim
167 words
A method for processing an image comprising the steps of: generating electronic signals representative of a digitized image, and electronically converting information contained in said digitized image into symbolic form using an optical recognition engine, said step of electronically converting comprising detecting and locating an optical pattern in said digitized image by the steps of inputting said digitized image into a first detector neural network comprising a plurality of detector nodes selectively interconnected only by forward feeding communication channels and outputting from said detector neural network an indication as to whether or not said optical pattern is present in said image, and inputting the output of said detector network into a second locator neural network comprising a plurality of locator nodes including one locator node corresponding to each detector node and outputting from said locator network an indication of the location of said pattern in said digitized image, said method further comprising the step of storing said symbolic information outputted by said recognition engine into a memory.
Claim 23
Dependent Claim
24 words
The method of claim 20 wherein said method comprises the step of removing the pixels in said digitized image which form said optical pattern.
References Cited (Backward Citations)
US4876731A
Cited
US5014219A
Cited
US5048100A
Cited
US4941122A
Cited
US4945494A
Cited
US5060276A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1996-08-23 | 1997-02-22 | 1997-08-24 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2000-08-23 | 2001-02-22 | 2001-08-24 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2004-08-23 | 2005-02-22 | 2005-08-24 | $7700.00 | Window Closed / Lapsed |