Why we are using Python in the EBA Project (English version)
- 20 de jul. de 2016
- 2 min de leitura
There´s a Portuguese version of this post here.
Most of the people who use LiDAR clouds in the forestry research, working with planted or native forests, uses the Fusion program, from US Forest Service, to extract metrics from the clouds.
These metrics are then correlated with field data to produce models (equations) which relate those parameters with the variable to be estimated, biomass for example.
In the EBA project, as we are interested in producing new models or even use other techniques such as BigData, genetic algorithms, etc. We will need to access the points of the cloud directly. We have also collected the full waveform data, which will certainly improve our estimates, and Fusion does not deal with this kind of data.
Therefore, it´s necessary that we build specific applications to analyze data we are collecting, and to do that we had to choose a programming language.
We chose Python for the following reasons:
It is a scripting language, which means quickly and concisely in writing programs.
It´s widely used in the scientific community, which facilitates the dissemination of knowledge, as more and more researchers and teachers know how to program in Python. Here the result of a survey that shows the growth of Python use about R, it's main "competitor": http://www.kdnuggets.com/2016/06/r-python-top-analytics-data-mining -data-science-software.html.
Python is multi-platform, running on primary existing operating systems.
An enormous amount of algorithms was ported to Python. Even canonical algorithms in C and C ++ are available in Python because someone else has been written Python interfaces making it available to us, as Python programmers
There are distributions like Anaconda that are suitable for Scientific Computing, already containing dozens of libraries that provide the necessary features. We use the Anaconda distribution in our project.
Though Python is not a language designed for multiprocessor as CUDA, having limitations for use in this way, we can develop applications that run routines in parallel, which gives us a huge gain in processing speed. Using 12 cores, I got a speed-up of about 10.
In a next post, I will explain how to install and configure the Anaconda distribution and other Python libraries that are needed to work with cloud LiDAR points.
Comentários