Inconsistent Caliper (NI Vision) results Assistance mode to VI mode

[Imagineer]

Hi Chris,

Thanks for the update! My apology for not getting back on this, as I have shifted to report generation, but shall work on this weekend and get back by Monday with my analysis.

Regards,

Tirthankar De

[Imagineer]

Dear All, 

As promised I have done my homework and am presenting it here. 

I have tested for individual block/tool and put the result in an excel sheet to do a comparative study. The file is attached with this post. Following are the observation (Feel free to add your perspective): 

1] The Straight Edge Rake tool in Vision Assistant mode outputs NaN when it’s unable to detect an edge, but when LabVIEW calling VA Express VI the same tool outputs 0 when unable to detect an edge. Is this a bug?

 

2] The Straight Edge Rake tool sometimes has wide differences between results obtained when in Vision Assistant mode and when LabVIEW calling VA Express VI.

  

3] The Pattern Matching tool has same performance in both cases.

4] The Caliper tool has similar results provided the points derived as inputs for the Caliper tool, are same.

From [1] it is seen that the caliper considers 0.00, 0.00 as a point in origin and calculates the angle whereas when its NaN,NaN the caliper just leaves it blank. This inconsistent output under same scenario seems more like a bug in the Straight Edge Rake tool! 

Could it be safe to deduce that the Straight Edge Rake tool needs further tuning/tweaking as a tool? OR is it that am using it improperly? 

Awaiting a response! 

Thanks and Regards,

Tirthankar De

Message Edited by Imagineer on 12-15-2008 02:15 PM

[Imagineer]

Addenum:

I have missed out to point another possibility - the environment also could make a difference. There could be something wrong (or different) eithere in Rake Tool<->Vision Assistant Environment OR Rake Tool in SubVI<->LabVIEW Environment , OR both?

Currently am using a workaround for this - ensuring the Rake Tool detects an edge somehow (even wrong edge), and later in caliper am validating the points by use of angle formed with other points.

Thanks and Regards,

Tirthankar De

[chris.b]

Hi Tirthankar De,

This was reported to R&D (CAR # 137072) for further investigation.

The best workaround right now is to do exactly what you are doing: ensure an edge is found.  All of my investigation brought me to the Rake Tool, so this output will have to be monitored.

Thanks so much for the feedback!

[Imagineer]

Thanks Chris, for your continued assistance. Shall I expect R&D team to come up with a solution by Jan 2009 ?

[chris.b]

I cannot promise you a patch fix, but we can definitely see what the best workaround/solution is.

[seVEMes]

The issue has been investigated and a fix may be included in the next module version. There will not be a patch fix for this problem.

Regards,

[Imagineer]

Thanks for the update! Am curious to know if the fix has taken care of inconsistency between the two environments.......i mean even for small differences? For illustration if in Assistance mode it gives P1(x=123.45, y=54.321) then as Sub-VI in LabVIEW it gives the same P1(x=123.45, y=54.321) since the image and algorithm, both remains the same and if input to a software remains same then output also should remain same?

May I also know when can I expect the new revision to be released ? 

DOUBT: Is it valid that if I set the edge strength as 108 and in a scenario, an image has maximum edge strength of 48, the Straight Edge Rake Tool will still detect edges and give two points P1(X1,Y1) and P2(X2,Y2) as output?

Awaiting your response.

Thanks and Regards,

Tirthankar De

[seVEMes]

Hi Tirthankar De,

There is no fix yet because it is in progress under CAR ID 137072. Since it is in progress the version release is yet to be determined. In order to know the version in which this fix will be implemented, please refer to the CAR number in the release notes of the following versions.