Summary

As we did with the points of interest, we managed the problem of excessive computational resource consumption by splitting the load into pieces. Specifically, instead of reading and processing whole images, we created Python iterators that allowed us to iterate through these images one row at a time without hitting the memory limit of the computer.

With this technique, we were able to process three Landsat 8 bands at a time to produce fancy colored images valuable for data visualization.

At this point, we are able to split our processing tasks into pieces that can be processed independently. We can do this with vectors, with database access, and now, with images as well.

With this, we completely paved the road for the next chapter, where we will be sending each of these pieces to be calculated at the same time by a different processor core performing the so-called parallel processing.