The Most Important Pixel!
Dr. Hernandez from Aero Hawk has developed a way to view what we would call the “Most Important Pixel”. His algorithms he developed for stitching pictures together has made it so the areas of the field that need attention don’t get overlooked because of pixel saturation from other pictures. This alone is an immense breakthrough for finding disease early in plants. In researching and using almost every stitching program out there I am starting to think that because stitching is mixing pixels from pictures next to each other we are getting a misrepresented reading of the crop. When a bunch of pictures are stitched together the actual exact reference point of a problem can be skewed. For instance, if I have between 120-1000(Depending on the camera) pictures on a 160-acre field about half of those pictures are stitched with another picture on all 4 sides. Every stitch line is a distortion area that may have a bad result and with an 80% overlap that line has become wide enough to effect pixels in the whole picture. If you are looking for disease in a field, that is leaving room for lots of errors. We believe that if there is a way to stitch but keep the integrity of the picture so we can zoom into every picture on that grid and get a reading that is not prone to stitching errors. We will eliminate problems that I see with a beautifully stitched picture that might have errors under the surface that are unseen. If we can see pixels that are a problem, let’s say it is a 5-square inch area (32^2 cm) but if the overlaying pictures are off by 2.5 inches all the telling pixels get diluted. In many cases, there will be up to 6 overlays of the same area so again we get that problem area is diluted and now we don’t see any problem because stitching caused the area to be diluted to look like everything around it.