Computer vision system to detect 3-D rectangular objects

• US 5,666,441 A
• Filed: 08/30/1995
• Issued: 09/09/1997
• Est. Priority Date: 03/17/1994
• Status: Expired due to Term

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Texas Instruments, Inc.
• Inventors: Rao, Kashi, Sarwal, Piyush
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): Tran, Phuoc
• Application Number:
  US08/521,512
• Time in Patent Office: 741 Days
• Field of Search: 382/203, 382/103, 382/154, 382/181, 382/199, 382/204, 382/282, 382/285, 382/286, 348/143, 348/152, 348/154
• US Class Current: 382/203
• CPC Class Codes: G06V 10/457 by analysing connectivity, e.g. edge linking, connected component analysis or slices
• CPC Class Codes: G06V 10/757 Matching configurations of points or features

Abstract

In general, the invention includes a method and system of rectangular solid detection comprising: receiving an image of data; detecting edges from the image of data; finding lines from the image of data; finding junctions from the lines; filtering junctions and lines that meet a first set of certain criteria; grouping the junctions and lines that meet a second set of certain criteria; and detecting whether a rectangular solid exists in the image according to the grouping. The criteria includes using triplets of lines and fork junctions. The system and method may also analyze a single frame, or a single frame paired with a reference frame.

Claims

1. A method of rectangular solid detection comprising:

receiving a reference image of a scene;

receiving a new image of said scene;

differencing said new image of said scene from said reference image of said scene to produce a differenced image of said scene with a modified differencing technique; and

identifying a rectangular solid from said differenced image of said scene wherein said identifying said rectangular solid includes;

detecting edges from said differenced image of said scene;

finding lines from said edges and said differenced image of said scene;

finding fork junctions from lines and said differenced image of said scene;

filtering fork junctions and triplets of lines that meet a first set of certain criteria;

grouping said junctions and said triplets of lines that meet a second set of certain criteria; and

determining whether a rectangular solid exists in said image according to said grouping.

2. The method of claim 1, wherein said finding fork junctions utilizes a quality factor, wherein said quality factor relates to a position of an intersecting point of two lines of said fork junction.

3. The method of claim 1, wherein said first set of certain criteria includes:

accepting said triplets of lines with outer lines parallel and overlapping;

rejecting said fork junctions with angles greater 270°

;

rejecting said fork junctions with two obtuse angles and one angle acute; and

accepting all remaining fork junctions.

4. The method of claim 1, wherein said second set of certain criteria includes:

grouping said fork junctions and said triplets of lines that share two lines;

grouping said triplets of lines that share two lines; and

grouping said fork junctions that share one line.

5. A method of rectangular solid detection comprising:

receiving an image of a scene;

detecting edges from said image of said scene;

finding lines from said edges and said image of said scene;

finding triplets of lines from said lines and said image of said scene;

finding fork junctions from said lines;

filtering fork junctions and triplets of lines that meet a first set of certain criteria;

grouping said fork junctions and triplets of lines that meet a second set of certain criteria; and

determining whether a rectangular solid exists in said image according to said grouping.

6. The method of claim 5, wherein said first set of certain criteria includes:

accepting said triplets of lines with outer lines parallel and overlapping;

rejecting said fork junctions with angles greater 270°

;

rejecting said fork junctions with two obtuse angles and one angle acute; and

accepting all remaining fork junctions.

7. The method of claim 5, wherein said finding fork junctions utilizes a quality factor, wherein said quality factor relates to a position of an intersecting point of two lines of said fork junction.

8. The method of claim 5, wherein said second set of certain criteria includes

grouping said fork junctions and said triplets of lines that share two lines;

grouping triplets of lines that share two lines; and

grouping said fork junctions that share one line.

9. The method of claim 8, wherein said grouping of said fork junctions that share one line includes grouping said fork junctions only where said one line is a center line of at least one of said fork junctions.

10. A method of rectangular solid detection comprising:

receiving an image of a scene;

detecting edges from said image of said scene;

finding lines from said edges and said image of said scene;

finding triplets of lines from said lines and said image of said scene;

finding fork junctions from said lines;

filtering fork junctions and triplets of lines, wherein said filtering includes;

accepting said triplets of lines with outer lines parallel and overlapping to produce an accepted set of triplet of lines;

rejecting said fork junctions with angles greater 270°

;

rejecting said fork junctions with two obtuse angles and one angle acute; and

accepting all remaining fork junctions to produce an accepted set of fork junctions;

grouping said fork junctions and triplets of lines, wherein said grouping includes;

grouping said accepted set of fork junctions and said accepted set of triplets of lines that share two lines;

grouping said accepted set of triplets of lines that share two lines; and

grouping identifying said accepted set of fork junctions that share one line; and

determining whether a rectangular solid exists in said image when a grouping is formed.