Pattern recognition apparatus and method

• US 4,385,322 A
• Filed: 08/13/1979
• Issued: 05/24/1983
• Est. Priority Date: 09/01/1978
• Status: Expired due to Term

• Current Assignee: View Engineering, Inc. (Quality Vision International, Inc.)
• Original Assignee: View Engineering, Inc. (Quality Vision International, Inc.)
• Inventors: Sacks, Jack, Hubach, Richard A., DeZotell, Gary L.
• Primary Examiner(s): Martin, John C.
• Assistant Examiner(s): Coles, Edward L.
• Application Number:
  US06/066,031
• Time in Patent Office: 1,380 Days
• Field of Search: 358/221, 358/169, 358/166, 358/160, 358/96, 358/101, 358/138
• US Class Current: 348/243
• CPC Class Codes: G06V 10/20 Image preprocessing
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
• CPC Class Codes: G06V 10/32 Normalisation of the pattern dimensions
• CPC Class Codes: G06V 30/2504 Coarse or fine approaches, e.g. resolution of ambiguities or multiscale approaches

Abstract

Video information obtained in real time is clocked and digitized according to light intensity. Changing light intensities are adaptively determined to vary the threshold level for determining black and white signals. The scene to be used as a reference is first recorded in a fine format and then in a coarse format. In the search mode the coarse stored information is compared in real time with coarse information from the field of view and convolved to generate a correlation number indicating the percentage of match. The location coordinates in the X and Y dimension for the best match is determined at the location of the highest correlation number. A fine search is then made around the coarse location previously determined by convolving in two dimensions previously stored information against information in real time from the scene under investigation to again determine the highest correlation as a measure of the location where the best possible match can be determined.

Claims

1. A pattern recognition system comprising:

video camera means for generating a video signal representative of an area being viewed,

opaque means located on a portion of said video camera means for producing on each line scan a black level video output signal corresponding to a black optical scene produced by said opaque means,

means for averaging the output signal of said video camera relative to said black video signal for each line scan to generate a varying output threshold voltage,

means for continuously comparing the output of said video camera means with said threshold voltage to vary the output level on each line scan relative to changes in light intensity,

clock means for digitizing the compared video output signal thereby generating a clock digitized video output signal having a leading edge and adaptive to varying light intensities, and

an analog comparator and in which the output of said analog comparator and said clock means are fed to a flip flop circuit for generating a change in state output in the presence of a change in the video signal as measured by the leading edge of each clock pulse.

2. A pattern recognition system comprising:

video camera means for generating a video signal representative of an area being viewed,

opaque means located on a portion of said video camera means for producing on each line scan a black level video output signal corresponding to a black optical scene produced by said opaque means,

means for averaging the output signal of said video camera relative to said black video signal for each line scan to generate a varying output threshold voltage,

means for continuously comparing the output of said video camera means with said threshold voltage to vary the output level on each line scan relative to changes in light intensity,

clock means for digitizing the compared video output signal thereby generating a clock digitized video output signal having a leading edge and adaptive to varying light intensitives, and

in which the output signal is averaged for each line scan and that value is stored in a sample and hold circuit and applied as the threshold voltage for the next scan thereby providing an adaptive updating of the average light level for each scan.