Vector neural networks

• US 5,432,889 A
• Filed: 11/30/1992
• Issued: 07/11/1995
• Est. Priority Date: 07/19/1990
• Status: Expired due to Term

• Current Assignee: Northrop Grumman Systems Corporation (Northrop Grumman Corporation)
• Original Assignee: Litton Systems Incorporated (Northrop Grumman Corporation)
• Inventors: Smith, William F., Johnson, David D., Ekchian, Leon K.
• Primary Examiner(s): Downs, Robert W.
• Application Number:
  US07/983,374
• Time in Patent Office: 953 Days
• Field of Search: 395/21, 395/22
• US Class Current: 706/27
• CPC Class Codes: G01S 13/726 Multiple target tracking
• CPC Class Codes: G01S 7/417 involving the use of neural networks
• CPC Class Codes: G06N 3/04 Architecture, e.g. interconnection topology
• CPC Class Codes: G06T 7/20 Analysis of motion motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
• CPC Class Codes: Y10S 348/909 Noise responsive signal processing

Abstract

A vector neural network (VNN) of interconnected neurons is provided in transition mappings of potential targets wherein the threshold (energy) of a single frame does not provide adequate information (energy) to declare a target position. The VNN enhances the signal-to-noise ratio (SNR) by integrating target energy over multiple frames including the steps of postulating massive numbers of target tracks (the hypotheses), propagating these target tracks over multiple frames, and accommodating different velocity targets by pixel quantization. The VNN then defers thresholding to subsequent target stages when higher SNR'"'"'s are prevalent so that the loss of target information is minimized, and the VNN can declare both target location and velocity. The VNN can further include target maneuver detection by a process of energy balancing hypotheses.

Claims

1. A vector neural network (VNN) comprising:

a) a neural network of interconnected neurons; and

b) means for directing energy flow within said neural network by the directional encoding of said energy flow as a positive energy flow hypothesis downstream of a selected neuron and a negative energy flow hypothesis upstream of said selected neuron, and for propagating postulated multiple hypothesis over said neural network.

2. The VNN of claim 1 in which said means for directing further comprises propagating said multiple hypothesis recursively over multiple frames of said neural network, and integrating the energy values of said positive energy flow hypothesis and said negative energy flow hypothesis upstream of said selected neuron using an exhaustive matched filter test on said multiple frames so that an ordering of said energy values at respective ones of said neurons is completed at each integration.

3. The VNN of claim 2 in which said means permits directional flow through said network.

4. The VNN of claim 2 in which said integrating further comprises determining and declaring said directional encoding as a maneuver with a resultant energy exchange amongst positive flow hypothesis so that the ordering of the upstream negative flow hypothesis is matched.