One of our large AgTech customers is encountering limitations with the current co-registration tool in the Orders & Subscriptions API. The current method, relying on a single user-defined anchor item, might decrease the overall positional accuracy for image stacks containing multiple images instead of increasing it.
We propose implementing a new functionality that utilises an average temporal image as the anchor for co-registration, as described in greater detail in this Sentinel Hub blog post. This approach would improve the average positional accuracy of the entire image stack, leading to more precise data for AgTech applications.
This enhancement would significantly benefit our AgTech customers and streamline their image processing workflows.
Since I cannot edit my own idea, I will leave this here as a reply:
Please also consider integrating this functionality into the Planet Insights Platform as a tool for image collections. If implemented, users could apply this post-data ordering, consuming PUs each run. It may be beneficial to explore if a co-registration approach similar to ARPS/Fusion data creation could be applied here, instead. Another benefit of offering such a tool in the platform would be that users could apply it to other data sources than PlanetScope (e.g. their own data).
The link posted above to the Sentinel Hub blog post is broken. Here is the correct one: https://medium.com/sentinel-hub/how-to-co-register-temporal-stacks-of-satellite-images-5167713b3e0b