I just completed processing some imagery taken with a MicaSense RedEdge-MX sensor onboard a FireFly6 Pro and have begun viewing the outputs. The flight was performed the first week of July around solar noon. The digital numbers seem reasonable as do the reflectance values for most of the mosaiced image. As part of the project, we have placed 6 gray scale and 3 RGB calibration targets in the flight area measuring 1.2m x 1.2m. We want to use the known reflectance from these tarps to do some post-processing. Unfortunately, with each band, the reflectance values for some of the calibration targets are essentially missing. There is data associated with the pixels as indicated from the orthomosaic albeit quite high. The digital number values are around 63000 (still below the 65636 maximum value). Any suggestions why this is happening and how I can fix this?
Brandon,
When the digital number of the target images are higher than 63000 we consider them overexposed. If possible, we usually recommend checking the calibration target shots while on the field to check for under/overexposure. We would recommend taking the image of the calibration target on top of a fairly reflective surface for the auto-exposure not to overexpose. For example, we do it on light asphalt and that typically works well (https://support.pix4d.com/hc/en-us/articles/206494883-Radiometric-calibration-target#label3)
Thank you for your response. As a standard practice, we take 2 sets of images of the MicaSense Calibration target before each flight and check the exposure on these prior to sending the UAV on its flight plan. We also take 2 additional sets of images after the flight so that we may deal with any corrections for changing atmospheric conditions during the flight at a later time. It’s essentially impractical to send the UAV up, take shots of the large calibration tarps, bring it down, look at the imagery, and then send it back up to do the flight. Doing so would require at least 2 additional batteries for each flight and when you’re flying multiple flights in a day the practicality of hauling that many batteries into the field quickly diminishes.
What I really don’t understand is for example in the ortho-mosaiced Green and NIR outputs, the digital numbers for the Green band are around 63000 and NIR around 65000 yet reflectance values are generated for the NIR band and not the Green. Any thoughts?
As a possible work around, can I pre-process the raw imagery by converting the images to reflectance and then mosaic them together with Pix4D? I really like how Pix4D does the matching and other than this issue, I’ve had nothing but good feedback from clients on the quality of the processed imagery.
Thank you.
" It’s essentially impractical to send the UAV up, take shots of the large calibration tarps, bring it down, look at the imagery, and then send it back up to do the flight. " This is very true!
What I really don’t understand is for example in the ortho-mosaiced Green and NIR outputs, the digital numbers for the Green band are around 63000 and NIR around 65000 yet reflectance values are generated for the NIR band and not the Green. Any thoughts?
Brandon, I can’t say why it is happening. I would need the dataset for that. Can you follow-up with a support ticket? We can investigate with more depth.
As a possible workaround, can I pre-process the raw imagery by converting the images to reflectance and then mosaic them together with Pix4D?
You can, but we do not encourage any pre-processing as you might lose the tags that Pix4D needs for processing (https://support.pix4d.com/hc/en-us/articles/205732309-EXIF-and-XMP-tag-information-read-by-Pix4D-Desktop). Pix4D uses all these tags for general processing and radiometric correction using the DLS sensor.
For situations like this, we recommend not using the target values or the target for calibration as then the calculations will be wrong. We ask users to process only with the DLS sensor data (camera and sun irradiance). You will get relative values, and not absolute. But if you are using the same camera, targets are not really needed. It is needed when cameras are different or you want to compare your values to some research paper or globally. Targets certainly are important and give more accurate values but if they are overexposed, nothing can be really done.
’ I really like how Pix4D does the matching and other than this issue, I’ve had nothing but good feedback from clients on the quality of the processed imagery" Thank you, we are glad you like our software 