	Machine Vision Systems

Normalized Correlation Search

Tips for Effective Use

The Pattern Position tool, which utilizes a type of processing called Normalized Correlation, can be used to locate a pattern on a target image that best matches the specified reference pattern from the registered "master" image. This tool enables the detection of position, rotation and match percentage (known as correlation value) of an object.

Effective Use of Pattern and Search Windows

1.	The pattern window should be set with an area of more distinct contrast, where no other similarly-shaped objects exist.

Avoid areas with low contrast or features for which rotation can't be detected

2.	Since a large search window will prolong the processing time, the window should be set to the minimum size needed.

Judgment by Correlation Value

Advantages of Using Correlation Value for Presence/Absence Detection

Pattern Position is a mode used to detect the positional coordinates of an object by using normalized correlation. It can also be used for presence check or shape detection by using correlation values. For example, to inspect print, the pattern window can be configured by setting the normal date as the template, as shown in the figure below. If any print is missing or smudged, the correlation value decreases and the item in question is
judged to be defective. Detection by binarization (CV: area mode by Keyence)
is also available, although detection using correlation values is preferable for the following reasons.

Using correlation values for pattern matching

Advantages

1. Normalized correlation search is unaffected by changes in ambient light intensity.
2. Even if the position or angle changes, the object can be steadily tracked.

	Page Top

Copyright(c) 2010 KEYENCE CORPORATION. All rights reserved.	\| Privacy \| Terms of Use \| Contact \| Produced by