Processing High-Resolution Optical Imagery

Blog Post created by hlzhang525 on Nov 30, 2014

In operation, people very commonly think that today powerful computers would help imagery specialists effectively to perform automation of color-balancing and spatial rectification for imagery processing (i.e., without intensively human involvements).


Obviously, it is wrong impression,  Including the use of GPUs-based supercomputing workstation, or multi-core /processor CPU-based server machines (over load-balancing cluster or Cloud infrastructure)...


In fact, imagery specialists are still facing challenges ‘effectively and accurately’ to process high-resolution (0.31-2.5 m) optical imagery for larger coverage (from hundreds of thousands to millions KM2), which is critical to support operations and many applications (# 1 below).


From frontline experiences, in addition to technical challenges, there is major barrier to tell people the limitations of current computer solutions with any latest imagery-processing-related algorithms, no matter how powerful those solution packagess are…

Certainly, massively 'manual' adjustment in color and spatial accuracy are highly required, to meet high-standard operation requirements like features' enhancement in mosaic, in addition to color-balanced requirement (#2 & 3 below).


Due to spectral variations, there are still some issues like seamlines in some areas (# 4 below).



1. High resolution (0.5-m) satellite imagery mosaic covering the area of over-2-million KM2,  which is well-balanced in color and spatially rectified

2. Overall, visible and near-infra remote sensing technology can be perfectly applied to the areas with limited variations to the landscape, which allows for nice color balancing in the entire region
high-level and clear features in color-balanced mosaic.PNG
3. High-level and clear features for feature extraction, mapping, and cartography in well-balanced color mosaic
4. However, some seamlines are still visible among scenes in some areas (around 3-5% of scenes) , due to 'huge' spectral differences, including atmospheric conditions