match pattern 4

[LOC91]

Hi everybody!

I have a question about IMAQ match pattern 4 VI in labview 2013. this VI has 2 output matches (in pixel) and matches (in real world). I want to know what they different?

Anyone can help me?

Thanhk in advance.

[William1225 · Accepted solution]

Hi, LOC91

It is like involvinig with the calibration process.

You can apply the image calibration to get the real-world dimension related to the image.

[LOC91]

Hi, William1225

I run example of labview( Pattern Matching). I connect the two output of Matching Pattern 4, matches(pixel) and matches( real- world) to indicators. I think two indicators will indicate different results. I get the same result in two indicators( matches and Matches 2).

can you explain for me? I attach file below.

Thank in advance.

[Poolean]

As William mentioned, the Real-World output corresponds to an image calibration that must be completed before performing the pattern match. I recommend viewing the Simple Calibration.vi shipping example. You can find this VI in LabVIEW by clicking Help>>Find Examples.. then navigating to Toolkits and Modules>>Vision>>Calibration.

-Mike

[LOC91]

Thank you.

[LOC91]

Hi Poolean

I run example, which you recommend for me. I read the comments in block diagram. have a comment: "Convert the image coordinate pixel position to the real world position for the mounting hole. Display both the pixel and real world position. Notice the real world position stays consistent except when the mounting hole's position is incorrectly drilled relative to the board corner". But I run and observe real world position. I realize that the result of real world position has difference, it is very small. This is tolerance of algorithm, isn't this?

Thanks a lot.

[Poolean]

You're correct. This is due to slight variations in the performance of the pattern matching algorithm. It does its best to locate the image in every frame, but the slightly different results of the "real" coordinates (with respect to the axis) come from the object being rotated in some of the images.

-Mike

[LOC91]

Hi Poolean

I understand my problem.I must attach calibration information to processed image before I use matching pattern. I have a question. I have a calibration grid, use camera to capture a calibration gird image. distance from calibration grid to camera is equal to distance from cinema screen to camera, isn't it?

I fixed position of camera. template object move on background. I mean when I take a photo calibration grid, I can set calibration grid abitrary as long as it is in region of image.

Thank you.

[Poolean]

I'm not sure what you mean from your last post. What's the cinema screen you're referring to?

Are these questions in reference to a specific VI?

What aribitrary calibration grid setting are you referring to?

-Mike

[LOC91]

Hi Poolean

I misunderstand about calibration. my problem has file VI, which simulate a moving rectangle box, use A laptop. I have a camera connect to another laptop, camera is fixed, distance from camera to screen laptop( A laptop) don't change. I acquire image from camera, find rectangle box position.

My idea grabs a calibration grid image, learns calibration information, saves image and calibration information to disk. Next I grab object image, read image( contain information calibration) from disk, copy information into object image, use IMAQ match pattern 4 to find the real world coordinate of object.

I use IMAQ get image info VI to display calibration information before and after save. I realise that axis info is same, but error statistics is different, The value of error satatistics before  save isn't zero. The value of error satatistics after save is equal to zero.

I have a question. Did I make a mistake in write and read image( contain calibration information) process?

I use IMAQ write image and vision info and IMAQ read image and vision info VIs, save image .png format.

I am so sorry for borthering you.

Thanks in advance.