Support Website Contact Support Blog

Parrot Sequoia - Choose best calibration image

When we want to take a calibration image using parrot sequoia, we have tree different images. They are same but taken different mode. So generally first part of the images saturated especially in red band. We can choose second or third one  to calibrate our images. Which one is better for us ? Their CHO values also nearly same. 

Hi Emre,

When you launch the radiometric calibration with Sequoia the camera takes 3 sets of images with different exposure in order to avoid over or underexposure. Only one set of images can be included in a project. A set of images consist of one image per each band (green, red, rededge, nir). You will have to choose the best set of images (the one not over or under exposed). When the digital number of the target images are higher than 63000 we consider them overexposed.

Once you have chosen one set of images you will have to upload each image for each band under Processing Options > 3 DSM, Orthomosaic and Index > Index Calculator tab > Radiometric Processing and Calibration: Calibration: Calibrate.



Hi Ina,

Is it possible to change the threshold of 63000 for overexposed images? All my images taken with Micasense RedEdge have values up to 65520.

I also noticed that I never had this error message popping up until recently: “Unable to compute radiometric target calibration for RedEdge_5.5_1280x960 (Blue) with X:/RedEdge/Tully_310118/WT1/0010SET/000/IMG_0027_1.tif: Not enough valid pixels for band 0”. Although my previous projects also had calibration target images with spectral range of up to 65520.

Hi Yuri,


Unfortunately not, if the image of calibration target DN’s is higher than 63000 then the mapper will reject it. This is part of the algorithm and it cannot be changed from UI.




I too had a question regarding this: When you say the “best set” of images, do you mean images with the same exposure for each band where none are overexposed or is it better to select the longest exposure for each band as long as it is not overexposed? Only choosing short exposure images seems to be leading to issues in my reflectance calculation.



Hi Jason,

Could you please give me more details regarding the reflectance calculation errors. I am asking this since it should not be the case because the exposure time it is taken into consideration for the corrections, and the images of the calibration target are radiometrically corrected as well.




Hi Ina

I have been using panels of known constant reflectance to assess the quality of my products and they show relatively large deviations between bands. For the green and sometimes also the red band, the pixels belonging to the panels will be masked out in the product, I assume due to oversaturation, even though they aren’t more than 50% reflective. 

For pre/post flight calibration I have tried both highly reflective (99%) and lower reflectance (~50 and 25%) targets. 


Could you give me more details on the reflectance panel that you are using? Do you have the Albedo values for Sequoia’s wavelengths?



I am using different panels, such as a LabSphere Spectralon (99%), and calibrated grey card (50-25%). I measure their reflectance using a spectrometer and derive Albedo for sequoia wavelengths by taking the mean over the bandpass regions specified (due to limited information on sensor details).

Hi Jason,

Your albedo calculation is correct. Regarding the masked pixels for your grey panels, you guessed it right, they must have been oversaturated, which leads to be masked out in the final reflectance maps.

Sequoia, as many other multispectral cameras, has an autoexposure mode, which estimates for each image the exposure that best represents most of the scene. Using this mode, you improve the acquisition quality, by acquiring frames that are not overexposed, nor underexposed.

That being said, if your scene is composed by not highly reflective elements (crops, soil, asphalt), most of materials would have their reflectances in green and red bands between around 0% to 25%. It would explain that both highly reflective (99%) and medium reflective (50%) panels are saturated.

To respond to your first question, you should select the longest exposure for each band as long as it is not overexposed.



@ina, what happens when I include the 3 sets of Sequoia radiometric calibration images in my project? I have done this hundreds of times guessing that Pix4d will choose an acceptable calibration image for each band. Occasionally I will get an error late in the processing (maybe Step 3?) that says that the calibration image for a certain band is overexposed. At that point I go and pick another calibration image. Could you say what logic Pix4d uses when all 3 sets of calibration images are included in a project?

More details about what I do: I take all the in-flight images as well as the calibration images (3 sets) all in a single folder, present that to Pix4d Desktop, and Pix4d automatically recognizes and picks the calibration images for each band. 

Hi William,

The criteria for images to be selected as radiometric calibration panel is it should have the tag of calibration image. It runs the test on the first 3 and last 3 sets of images. If one of the candidates is successful, the result is used and the remaining candidates are not checked anymore.

Pix4D will then check for overexposure at step 3. and then throws an error or proceeds. Thus it is best to choose the best set of images manually. We might have the automated feature of detecting the best set when the processing starts in the future. 

Thanks, Momtanu for that explanation.