Method and apparatus for background determination and subtraction for a monocular vision system

• US 5,684,898 A
• Filed: 06/01/1995
• Issued: 11/04/1997
• Est. Priority Date: 12/08/1993
• Status: Expired due to Fees

• Current Assignee: 3M Company
• Original Assignee: 3M Company
• Inventors: Cerny, Darin G., Brady, Mark J.
• Primary Examiner(s): COUSO, YON JUNG
• Application Number:
  US08/458,577
• Time in Patent Office: 887 Days
• Field of Search: 382/107, 382/104, 382/215, 382/216, 382/103, 382/100, 382/276, 382/282, 382/286, 382/291, 358/464, 348/148, 348/149, 348/152, 348/155, 348/143
• US Class Current: 382/282
• CPC Class Codes: G01V 8/10 Detecting, e.g. by using light barriers by reflection from the object G01S17/00
• CPC Class Codes: G08G 1/04 using optical or ultrasonic detectors

Abstract

A method and apparatus for producing a background image from a plurality of images of a scene and for subtracting a background image from an input image are described. A background image is produced by dividing an image into subimages, acquiring reference subimages for each subimage location and comparing subsequent subimages with the reference subimage to determine if any objects have passed between the reference subimage and the video camera that acquired images. When objects have passed between the reference subimage and the video camera, the reference subimage is designated as background and stored in a background image. Background portions of an input image can be removed or their intensity diminished with a background image. Foreground weights can be determined by comparing the difference between a background image and an input image. To the extent that corresponding pixels are the same, the pixel is given a low foreground weight, indicating that the pixel is a background weight. The background subtraction method can further employ a weighting curve to take into account noise considerations. The foreground weights are then applied to an input image to diminish or remove pixels in the background.

Claims

1. A method of producing a background image from a plurality of images of a scene, said plurality of images acquired from a machine vision system with image acquisition means, said method comprising the steps of:

a) selecting within each image of said plurality of images a subimage;

b) acquiring a reference subimage corresponding to said subimage;

c) designating said reference subimage as a background subimage if any object has passed between said reference subimage and said image acquisition means in subsequent images; and

d) storing said background subimage in a corresponding position in said background image;

wherein said steps of acquiring reference subimages and designating said reference subimage as a background subimage comprise the steps of;

e) initializing a state table for said image, said state table having a first state for each subimage location corresponding to each said subimage after said step of initializing;

f) storing a first subimage from a first subimage location in a reference buffer and assigning a second state to said first subimage location;

g) comparing subsequent subimages in said first subimage location with said first subimage;

h) assigning a third state to said first subimage location when a predetermined number of subsequent subimages are similar to said subimage stored in said reference buffer when said first subimage location is assigned said second state;

i) assigning a fourth state to said first subimage location when said subsequent subimage is significantly different than said subimage stored in said reference buffer and said first subimage location is assigned said third state;

j) designating said subimage stored in said reference buffer as said background subimage when said subsequent subimage is substantially similar to said subimage stored in said reference buffer and said first subimage location is assigned said fourth state; and

k) repeating steps f) through j) for each subimage location and said subsequent subimages corresponding to said subimage location.

2. The method of producing a background image from a plurality of images of a scene according to claim 1, wherein a plurality of subimages are selected within each image.

3. The method of producing a background image from a plurality of images of a scene according to claim 1, further comprising the step of assigning said first state to said first subimage location if said first subimage location is assigned said fourth state for a predetermined time.

4. The method of producing a background image from a plurality of images of a scene according to claim 1, wherein each said subimage comprises a single pixel within said image.

5. The method of producing a background image from a plurality of images of a scene according to claim 1, wherein said predetermined number of subsequent subimages is zero.

6. A machine vision system for producing a background image from a plurality of images of a scene, said system comprising:

a) image acquisition means for acquiring images from three-dimensional space;

b) dividing means for dividing said images into subimages;

c) memory means for storing said images, said subimages, reference subimages and said background image; and

d) processor means for determining when any object passes between said reference subimages and said image acquisition means in subimages acquired subsequent to acquiring said reference subimage;

wherein said processor means determines whether an object passes between said reference subimages and said image acquisition means by the steps of;

e) initializing a state table for said image, said state table having a first state for each subimage location corresponding to each said subimage after said step of initializing;

f) storing a first subimage from a first subimage location in a reference buffer and assigning a second state to said first subimage location;

g) comparing subsequent subimages in said first subimage location with said first subimage;

h) assigning a third state to said first subimage location when a predetermined number of subsequent subimages are similar to said subimage stored in said reference buffer when said first subimage location is assigned said second state;

i) assigning a fourth state to said first subimage location when said subsequent subimage is significantly different than said subimage stored in said reference buffer and said first subimage location is assigned said third state;

j) designating said subimage stored in said reference buffer as said background subimage when said subsequent subimage is substantially similar to said subimage stored in said reference buffer when said first subimage location is assigned said fourth state; and

k) repeating steps f)-j) for each subimage location and said subsequent subimages corresponding to said subimage location.

7. The machine vision system for producing a background image from a plurality of images of a scene according to claim 6, further comprising a second processing means for comparing said background image with input images and subtracting background from said input images.

8. The machine vision system for producing a background image from a plurality of images of a scene according to claim 6, wherein said image acquisition means comprises a monocular vision device.

9. The machine vision system for producing a background image from a plurality of images of a scene according to claim 7, wherein said second processing means comprises:

a) means for generating foreground weights based on the difference between said background image and said input images as weighted by a weighting curve; and

b) multiplier means for multiplying said foreground weights with a measure of edge contrast of said input image.

10. The machine vision system for producing a background image from a plurality of images of a scene according to claim 9, wherein said means for generating foreground weights comprises a lookup table.