National Instruments IMAQTM 用户手册

A high score does not reflect the accuracy of the system.

If the learning process returns a learning score below 600, try the following: 

1.

Make sure the grid complies with the guidelines listed in the 

 section.

2.

Check the lighting conditions. If you have too much or too little 
lighting, the software may estimate the center of the dots incorrectly. 
Also, adjust the threshold range to distinguish the dots from the 
background.

3.

Select another learning algorithm. When nonlinear lens distortion is 
present, using perspective projection sometimes results in a low 
learning score.

An error map helps you gauge the quality of the complete system. The error 
map returns an estimated error range to expect when a pixel coordinate 
is transformed into a real-world coordinate. The transformation 
accuracy may be higher than the value the error range indicates. Set 

CWIMAQLearnCalibrationOptions.LearnErrorMap

 to 

True

 to learn 

the error map.

If the speed of image correction is a critical factor for the application, use 
a correction table. The correction table is a lookup table that contains 
the real-world location information of all the pixels in the image. The 
correction table is stored in memory. The extra memory requirements for 
this option are based on the size of the image. Use this option when you 
want to simultaneously correct multiple images in the vision application. 
Set 

CWIMAQLearnCalibrationOptions.LearnCorrectionTable

 to 

True

 to learn the correction table.

Use the scaling mode option to choose the appearance of the 
corrected image. Set 

CWIMAQLearnCalibrationOptions.CorrectionScalingMode

 to 

cwimaqScaleToFit

 or 

cwimaqScaleToPreserveArea

. For more 

information about the scaling mode, refer to Chapter 3, System Setup and 
Calibration, in the IMAQ Vision Concepts Manual.