NI Vision Concepts Help

Content Type
Programming Language
Current manual

Geometric Matching Using Calibrated Images

Geometric Matching Using Calibrated Images

During matching, the geometric matching algorithm uses calibration information attached to the inspection image to return the position, angle, and bounding rectangle of a match in both pixel and real-world units. In addition, if the image is calibrated for perspective or nonlinear distortion errors, geometric matching uses the attached calibration information directly to find matches in the inspection image without using time-consuming image correction.

 

Simple Calibration or Previously Corrected Images

If an inspection image contains simple calibration information, or if the inspection image has been corrected prior to being used by geometric matching, the matching stage performs the same way that it does with uncalibrated images. However, each match result is returned in both pixel and real-world units. The pixel-unit results are identical to the results that would have been returned from matching the same, uncalibrated image. Geometric matching converts the pixel units to real-world units using the simple calibration information attached to the inspection image.

Perspective or Nonlinear Distortion Calibration

If an inspection image contains calibration information for perspective or nonlinear distortions, the first step in the matching process is different than it would be with uncalibrated images. In the first step, curves extracted from the inspection image are corrected for distortion errors using calibration information. The remaining four steps in the matching process are performed on the corrected curves. Each match result is returned in pixel and real-world units.

Match results in pixel units are returned to be consistent with the inspection image. As a result, the bounding rectangle of a match in pixel units may not be rectangular, as shown in the following figure. Figure A shows the template image of a metallic part. Figure B shows an image of a calibration grid. The image exhibits nonlinear distortion. Figure C shows an image of metallic parts taken with the same camera setup used in figure B. The gray lines depict the bounding rectangle of each match found by geometric matching.

Geometric Matching Using Calibration Information
Was this information helpful?