03-21-2017 11:55 PM
Looking for suggestions..... I am trying to fuse optical imagery with GPR data. Have been able to generate a representative image from basic GPR data. I am trying to zoom this resultant GPR image using the resample function to spatially match the optical image so I can fuse the data (already done). But the large image size difference results in a greatly, and unusable, pixelated GPR image. I am trying to correct this. I am currently trying to convert the image into a picture control which then I have an example of good quality zooming. But then I can not use the AND/OR type functions to accomplish the fusion. Any other thoughts on quality image zooming?
03-22-2017 08:07 AM
Once you have a digital picture, the number of data points (pixels) you have is fixed. Suppose you have 1000 x 1000 image, and you want to zoom it 10 times to view the center 100 x 100. This means that you are discarding 99% of your data and need to "smear" the remaining 1% across the entire 1000 x 1000 space. There are 2D smoothing functions that can help "de-pixelate" the resulting image (much easier to do with grey-scale than with color), but you are still forced to deal with "making up new data". Note that there are such functions in the Vision palette.
Bob Schor
03-22-2017 12:25 PM
Thanks for the thoughts,
I am trying to zoom the GPR image a long way but the resample function seems to simply duplicate pixels, even though it appears to hide this with a few built in algorithms. I know there are other methods such as using the r-squared concepts but you just confirmed that there are no other ways built into Labview…. Nuts.
Everything I always thought I knew and read about GPR, and the massive amount of data it produces, was just wrong. As is, I don't think what I am doing useful. Disappointed but I can't change the GPR and perhaps this is why data fusion between Optical devices and GPR is not being done.