Switched resistive neural network for sensor fusion

• US 5,218,440 A
• Filed: 06/07/1991
• Issued: 06/08/1993
• Est. Priority Date: 06/07/1991
• Status: Expired due to Fees

• Current Assignee: Conexant Systems Incorporated (Synaptics Incorporated)
• Original Assignee: Rockwell International Corporation
• Inventors: Mathur, Bimal P.
• Primary Examiner(s): MacDonald, Allen R.
• Application Number:
  US07/712,240
• Time in Patent Office: 732 Days
• Field of Search: 395/22, 395/24, 382/14, 382/15, 358/213.27, 358/213.15
• US Class Current: 348/25
• CPC Class Codes: G06N 3/065 Analogue means
• CPC Class Codes: G06T 2207/20084 Artificial neural networks [ANN]
• CPC Class Codes: G06T 2207/20221 Image fusion; Image merging
• CPC Class Codes: G06T 5/50 using two or more images, e.g. averaging or subtraction

Abstract

An electronic image processing system uses data provided by one or more sensors to perform cooperative computations and improve image recognition performance. A smoothing resistive network, which may comprise an integrated circuit chip, has switching elements connected to each node. The system uses a first sensory output comprising primitives, such as discontinuities or object boundaries, detected by at least a first sensor to define a region for smoothing of a second sensory output comprising at least a second, distinct output of the first sensor or a distinct output of at least a second sensor. A bit pattern for controlling the switches is generated from the detected image discontinuities in the first sensory output. The second sensory output is applied to the resistive network for data smoothing. The switches turned off by the data from the first sensory output define regional boundaries for smoothing of the data provided by the second sensory output. Smoothing operations based on this sensor fusion can proceed without spreading object characteristics beyond the object boundaries.

Claims

1. An electronic processing system, comprising:

a processor having a resistive network, said network comprising a plurality of nodes;

each of said nodes connected to a neighboring node by a resistive component and a switch connected in electrical series;

at least one sensor connected to said network;

a first output of said at least one sensor directed to said nodes to control said switches and thereby set boundaries within said network; and

a second output of said at least one sensor directed to said nodes for resistive smoothing within said boundaries of said network set by said switches.

2. The system of claim 1, further comprising output means connected to said processor for providing a bounded, smoothed data output.

3. The system of claim 2, wherein said first output comprises data corresponding to discontinuities detected by said at least one sensor.

4. The system of claim 2, further comprising a plurality of latches, each of said latches connected to a corresponding one of said switches, said latches receiving said discontinuity data for setting said switches.

5. The system of claim 4, wherein said output means further comprises a plurality of output amplifiers, each of said output amplifiers connected to one of said plurality of nodes.

6. The system of claim 2, further comprising:

a first sensor for providing said first output;

a second sensor for providing said second output; and

said first and second sensors co-boresighted on a common field.

7. An electronic image processing system, comprising:

a processor having a resistive network, said network comprising a plurality of nodes;

each of said nodes connected to a neighboring node by a resistive component and a switch connected in electrical series;

one or more sensors connected to said network, said one or more sensors generating at least two distinct sensor images of an object field;

at least a first of said sensor images directed to said nodes to set said switches, said switch settings forming boundaries within said network corresponding to discontinuities detected in said at least a first of said sensor images;

at least a second of said sensor images directed to said nodes for resistive image smoothing within said boundaries of said network set by said switches; and

output means connected to said processor for providing a bounded, smoothed output image.

8. The system of claim 7, further comprising a plurality of latches for receiving said at least a first of said sensor images and setting said switches, each of said latches connected to a corresponding one of said switches.

9. The system of claim 8, wherein said output means further comprises a plurality of output amplifiers, each of said output amplifiers connected to a corresponding one of said plurality of nodes.

10. The system of claim 9, further comprising:

a first sensor for providing said first sensor image;

a second sensor for providing said second sensor image; and

said first and second sensors co-boresighted on said object field.

11. A method of electronic image processing, comprising the steps of:

providing a processor with a resistive network having a plurality of nodes;

connecting each of said nodes to a neighboring node with a resistive component and a switch in electrical series;

connecting at least one sensor to said network;

generating a plurality of sensor images of an object field;

detecting discontinuities in at least a first of said sensor images;

directing data representing said detected discontinuities to said nodes for setting said switches;

forming boundaries within said network, said boundaries defined by said switch settings corresponding to said detected discontinuities;

directing data representing at least a second of said sensor images to said nodes for resistive image smoothing within said boundaries formed in said network; and

providing a bounded, smoothed output image.

12. The method of claim 11, wherein the step of setting said switches further comprises the step of directing data representing said detected discontinuities to a plurality of latches, each of said latches connected to a corresponding one of said switches.

13. The method of claim 12, wherein the step of providing said bounded, smoothed output image further comprises the step of connecting each of a plurality of output amplifiers to a corresponding one of said plurality of nodes.

14. The method of claim 13, further comprising the steps of:

connecting a plurality of sensors to said network for generating said plurality of sensor images; and

co-boresighting said plurality of sensors on said object field.