Comments

• Moderator

  MarioSpeedwagon August 13, 2015 at 11:08am

  Let's see it
• 

  Vega Tech Group LLC August 13, 2015 at 11:04am

  It doesn't have to he low quality. This is a machine vision application. You are looking at the viability of the crop and sorting it. It can be done while the crops area growing or after harvest with a high speed system.Since this requirement is for a single attribute. A single monochrome camera and filter can he used. Cost if a total system I can build under $1k.
• Moderator

  MarioSpeedwagon August 13, 2015 at 9:35am

  Jimmy, I just uploaded 4 from the same set...but they're still the blurry ones.

  Ned,

  That index is easier to see for me and the center portion looks accurate, but I'd still like to have the agronomist ground truth a field after we fly it. 

  I know that we can isolate the rye wheat with the current filter, but it would have to be set up again every flight and adjusted to the flight conditions. If it is possible to modify a low end camera to be able to perform this task reliably, I think there is a lot more potential for business than with NDVI. The growers here can't sell their wheat unless it is 98%(I think) pure and rye-free. 

  Vega, you've posted links to publications and manufacturers before...can you give us a recommendation on the absolute lowest quality, but "ready to map" package that YOU would recommend? Basically, what you would consider a starting point. 
• 

  Jimmy Oliver August 13, 2015 at 9:06am

  Can I get 3 or more images to run through PIX4D from someone here? 
• 

  Vega Tech Group LLC August 13, 2015 at 9:00am

  I have done several projects over the past seven years in plant identification from aircraft. From oak trees to illicit drugs in corn fields. Simple to do, since I have done it.
• 

  Greg Nuspel August 13, 2015 at 8:53am

  I do wonder about the lens quality. With that short of a focal length the depth of field should be incredible. A 4.5mm f1.8 lens hyper focal distance is about 3.4 ft. with a 1/3" sensor. Sorry I missed you actual focal length but you can use this calculator http://www.dofmaster.com/dofjs.html 

  

  Online Depth of Field Calculator

  Depth of field and hyperfocal distance calculator
• Ned Horning August 13, 2015 at 8:04am

  Sorry about that Mario - Here is a colorblind friendly look up table I created but if you have suggestions to improve it please pass them on. I also resampled the image by a factor of 4 taking the mean of aggregated pixels. 

  You can test different filters for separating Rye wheat. You could look at the spectral curves for Rye wheat and compare them with spectral curves of plants you're trying to separate Rye wheat from to see if there are any spectral difference that are obvious. You might be able to use that information to select a good filter. If there are other factors that make Rye wheat different form other plants like image texture you can use that to help separate them. Often a good way to detect specific species/varieties is to time your image acquisition so the difference between your plant of interest and other plants is greatest. 

  If there is a better way for me to post these images please let me know.

  

  
• Moderator

  MarioSpeedwagon August 13, 2015 at 7:26am

  Uploading them now, should be on my profile. 
• Moderator

  MarioSpeedwagon August 13, 2015 at 7:23am

  Vega - If I can get the lens set for 118m, I'll at least be satisfied with using it on the fixed wing. I agree though...trying to get the lens focus set has been a lot of trial and error. Since I'm going to google it anyway, and just in case someone else here isn't familiar with the terminology, what to expand on chromatic aberration?

  Ned - I'm pretty damn colorblind, which is a big part of why I like Agribotix's NDVI color scale and post processing. I can't tell whats what in your image. 

  I will say that the image is part of a larger set that I sent through post processing at Agribotix, even though they were still out of focus. I did a small survey near my boss' house because they used the wrong chemicals on patches of their lawn recently, and it's very easy to see with the naked eye. The survey picked up some of the dead strips on the southern end of the property, but missed some on the north end. I'm still withholding judgement until I can get the focus to a point that I am happy with.

  Also, met with a very large seed company yesterday. They are interested (and were actually looking at buying their own), but the main use they want a drone for is identifying "Rye Wheat" in wheat fields. I said we MIGHT be able to isolate the Rye with the cameras I have, but it would have to be done for each flight and probably involve more manhours than what they are doing now. I'm curious what capabilities you would need to be able to do this on a consistent basis? I do have some images of Rye wheat that I took from the S100 about 3 months ago, if interested. 
• 

  Vega Tech Group LLC August 13, 2015 at 7:19am

  Ned, without the UDR of an achromatic lens set at the middle of the spectral range imaged, the data.would be difficult to index, not impossible. It could take a number of imaged to get a mean value.