Method for recognizing a target component image within an image having multiple component images

{
  "publication_number": "US6718048B1",
  "country": "US",
  "kind": "B1",
  "title": "Method for recognizing a target component image within an image having multiple component images",
  "abstract": "A position detection tool in a machine vision system finds the position of target objects in a digital image using the length and width of the target object as parameters. The input length and width dimensions are measured by the developer as the length and width of a simple hypothetical rectangle that surrounds the object. For most shapes, the developer can easily obtain these two measurements. The position detection tool can easily be adapted to discern particular image patterns from multiple component images.",
  "claims": [
    "1. A method for recognizing a target component image from a larger image having a plurality of component images, the method comprising: receiving a first linear dimension measurement and a second linear dimension measurement of a minimum enclosing rectangle of an object corresponding to the target component image; segmenting the larger image into the plurality of component images; calculating an error amount for each of the plurality of component images based on a calculation of two local maximum distances between edge pixels in each of the component images for each of a plurality of orientations of an orthogonal coordinate system defined by the first and second linear dimensions and comparing the calculated two local maximum distances to the first and second linear dimensions; and recognizing the target component image as the component image with the smallest error amount.",
    "2. The method of claim 1, further comprising performing edge detection of the component images to define an outlining area for each of the component images.",
    "3. The method of claim 1, wherein recognizing the target component image as the component image with the smallest error amount additionally includes comparing the error amount with a predetermined error value and recognizing the target component image when the error amount is less than the predetermined value.",
    "4. The method of claim 1, wherein calculating two local maximum distances between edge pixels further comprises determining distances between all pairs of edge pixels that lie along lines parallel to the direction of the linear dimensions for each of the plurality of orientations of the orthogonal coordinate system.",
    "5. The method of claim 1, wherein the first and second linear dimensions define dimensions of a hypothetical minimum enclosing rectangle surrounding the object."
  ],
  "description_excerpt": "1. Field of the Invention\n\nThe present invention relates to machine vision systems, and more particularly to machine vision systems for detecting the position of an object.\n\n2. Description of Background Information\n\nMany manufacturing processes are automated to enhance speed and efficiency. For example, in the assembly of printed circuit boards (PCBs), robotic arms are often used to insert surface mounted devices (SMDs), such as semiconductor chips, resistors, and capacitors, onto the PCBs. Machine vision systems assist in the automated manufacturing process. Generally, in a machine vision system, a digital picture of the manufacturing area of interest is taken and interpreted by a computer. Machine vision systems perform a variety of tasks, including machine guidance (e.g., guiding the robotic arm to insert its SMD at the correct location), part identification, gauging, alignment, and inspection.\n\nOne particular task performed by machine vision systems is the task of position detection or pattern recognition. In position detection problems, a description of the object of interest is given to the machine vision system, which then applies a position detection algorithm to find the location of the object in images taken during the manufacturing process.\n\nConventional pattern matching algorithms include the so-called “caliper” and “blob” matching algorithms. The caliper algorithm is modeled after a mechanical caliper. The developer specifies the desired separation between the caliper “jaws” - or the approximate distance between parallel edge pairs of interest on the object.",
  "cpc": [
    "G06T 7/73",
    "G06T 2207/30141",
    "G06T 7/564",
    "G06V 10/25"
  ],
  "ipc": [
    "G06T 7/00",
    "G06V 10/25"
  ],
  "assignees": [
    "Cognex Technology and Investment LLC"
  ],
  "inventors": [
    "Masayuki Kawata",
    "Kenji Okuma",
    "Hiroyuki Hasagawa"
  ],
  "filing_date": "2002-08-12",
  "publication_date": "2004-04-06",
  "grant_date": "2004-04-06",
  "priority_date": "1999-09-24",
  "application_number": "US-21717002-A",
  "family_id": "23599989",
  "cited_by_count": 13,
  "citations": [
    "US4007440A",
    "US4739401A",
    "US6002793A",
    "US5495537A",
    "US5602937A",
    "US5796868A",
    "US5872870A",
    "US5933523A",
    "US5974169A"
  ]
}