06-20-2012 11:46 AM
Hi!
I am using Labview 2011, NI Vision Assistant, and a Kinect. I have been successful in using the color location processing function for the images collected by the Kinect, but for improved location accuracy, I thought I would try to use geometric matching (under the machine vision of the NI Vision Assistant), but using the same basic vi framework, I can't seem to get my vi to work without the Error -1074396080 (invalid image type). To do geometric matching, I have to first convert my RGB image from the Kinect to an 8-bit image. I do this by using the Color Plane Extraction function in the NI Vision Assistant, and then I use the Geometric Matching function with the 8-bit image. My error comes up with the conversion of the RGB 32 Kinect image to an 8-bit image, but I am not entirely sure why.
I feel like this should be a simple vi, but after looking at many forum posts and struggling with this vi for days, I am stuck.
I have attached my vi (kinectgeometricmatching) and the sub vi created with the NI Vision Assistant (geometricmatching) below. Any help would be much appreciated!!
Thanks!
06-21-2012 09:55 AM
Hi,
I just tried to convert a 32-Bit RGB by extracting the Luminance as you have done. This works and I was able to use the geometric pattern matching step. Could you post an image taken with the Kinect? Also, do you get this error in Vision Assistant with the geometric pattern matching or only when you try it in LabVIEW?
Regards,
Greg H.