Over the past several years HoneyComb has worked hard to revolutionize how drone-based image processing is done. This has been an endeavor that has focused on quality, reliability, and accuracy of data, with careful development of the AgDrone System™ to match our data processing objectives.

Unlike any competitors’ offering, HoneyComb’s camera system and image processing pipeline are both custom built, in-house creations carefully designed and calibrated to work together.

The camera system consists of two sensors, one capturing ordinary color and the other three independent infrared bands. Through thoughtful calibration

[1], these sensors sport spectral fidelity rivaling that of the specialized four channel multispectral sensors commonly used in agriculture, with the added bonus of much higher image resolution. This translates to honest, clear, high resolution RGB and NDVI imagery without added expense.

The image processing pipeline first runs these images through a modern photogrammetry process [2], producing precise estimates of view parameters for every image alongside 3D terrain shape. This component is designed with aerial photography in mind (instead of general-purpose) to yield fast, highly reliable solutions while remaining unbiased.

Once this step is complete, the pipeline remaps input imagery onto the optimized terrain surface, cutting seams intelligently to minimize misalignments, maximize sharpness, and avoid solar issues. This is accompanied by a global color balancing scheme, individually correcting every pixel’s color.

The remapped, color corrected, seam-cut imagery is then finally assembled into RGB and NDVI tiles from which maps, shapefiles, and other GIS products are created.

[1] Wug Oh, Seoung et. al., Do It Yourself Hyperspectral Imaging With Everyday Digital Cameras, CVPR 2016.

[2] Szeliski, Rick, Pushing the Envelope of Modern Methods for Bundle Adjustment, CVPR 2010.

If you’d like to stay updated on the latest news from HoneyComb, sign-up for our newsletter.