Mosaic method about ukis-pysat

[shiwakotisurendra]

i checked and tested the the rio-tiler and rio_tiler.mosaic module can be used to create a mercator tile from multiple observations . For showing the portions of overlap in tile from multiple assets, pixel could be chosen from multiple datasets. rio-tiler.mosaic module provides pixel selection methods which define how to handle different cases for the pixel.
First, Highest, Lowest, Mean, Median,
Stdev: compute the standard deviation value of the whole stack,
LastBandHigh: Use last band (highest) as a decision factor
LastBandLow: Use last band (lowest) as a decision factor

[MWieland]

Image mosaicing is a very good point and would be highly beneficial for many applications. After a discussion with @BoehnkeC we thought about an idea that would bring this to a higher level and could provide potentially an added value compared to plain rasterio functionality could be to think of this in terms of an ImageCollection. We could implement this as a new class that collects multiple Image instances. Within this class we could implement generic placement and geometry methods and connect this with above mentioned mosaicing method. Beyond a horizontal mosaicing we could also use such a class to stack Image instances vertically to carry out timeseries analysis (or simple bandstacks). Let's further discuss this option here and think about a possible class structure before starting to implement further.

[fwfichtner]

@BoehnkeC would this help you already: https://cogeotiff.github.io/rio-tiler/mosaic/?

More information about COGs and mosaics or about the newest release of rio-tiler.

[fwfichtner]

Agree, this looks nice and should be used before we do anything here.