---
title: Filtrage et classification de nuage de points sur la base d'attributs morphologiques
subtitle: ORASIS 2019 -- Classification
date: May 30, 2019
author: 
- Florent Guiotte^1,2^ 🍆 \<<florent.guiotte@uhb.fr>\>
- Thomas Corpetti^1^
- Sébastien Lefèvre^2^
institute: ^1^Univ. Rennes 2 (LETG) --- ^2^IRISA (OBELIX)
theme: metropolis
aspectratio: 169
bibliography: all.bib
header-includes: |
    \usepackage{pdfpc-commands}
    \usepackage{qrcode}
    \setbeamercolor{background canvas}{bg=white}
---

Context
=======

LiDAR Point Cloud
-----

:::: {.columns}
::: {.column width="50%"}

**LiDAR Point Cloud**

- 3D spatial data $P \in \mathbb{R}^3$
    + Voluminous (high density point cloud)
    + Continuously distributed
- LiDAR features for each point
    + Intensity
    + Number of echoes
    + ...
- Multispectral
:::
::: {.column width="50%"}

![](vid/paris.mp4)

:::
::::

::: notes
Airborne LiDAR systems are a common source of acquisition for elevation. They provide point clouds with higher and higher density.

LiDAR point clouds consist in 3D spatial data, with a set of points defined in space by x, y and z.

In addition to that, LiDAR point clouds comes with features extracted from the capture such as 
- the laser intensity at the returned point,  
- The number of echoes a pulse have returned, for some surfaces such as vegetation, the laser go through leaves of small branches and then can return several echoes, for example one point at the top of the canopy, and one point at the ground under the tree.
- and other metadata from the flight

some LiDARs are multispectral now, so we can also have several point cloud of the same scenery with intensities or number of echoes dependings of the wavelength.
:::

LiDAR Applications
---------

:::: {.columns}
::: {.column width="50%"}
**Urban Applications**

- Characterize structures
    + Buildings
    + Vegetation
    + ...
- Classification
:::
::: {.column width="50%"}
![LiDAR classification of a Paris street](img/paris.png)
:::
::::

::: notes

Some common LiDAR urban application are to characterize structures for

- heat island analysis for example

or to achieve classification 

:::

Citations
--------

- Hierarchical @bosilj_partition_2018.
- Hierarchical [@bosilj_partition_2018]
- Hierarchical [-@bosilj_partition_2018]
- Hierarchical @bosilj_partition_2018 [p. 3]

Maths
-----


$$ \mathbb{R}^3 $$

Other
----

See [context][LiDAR Point Cloud].

```haskell
qsort []     = []
qsort (x:xs) = qsort (filter (< x) xs) ++ [x] ++
               qsort (filter (>= x) xs)
```

- [ ] Not working
- [x] Working

| 200 Main St.
| Berkeley, CA 94718

Definition
--------

Term 1

:   Definition 1

Term A

:   Definition of the second term named A. Seriously.

Term 2
  ~ Definition 2a
  ~ Definition 2b

Term 2 with *inline markup*

:   Definition 2

        { some code, part of Definition 2 }

    Third paragraph of definition 2.

Broken
---

![Image manquante](img/manquante.png "Image manquante.")

::::: {#special .sidebar}
Here is a paragraph.

And another.
:::::


::: Warning ::::::
This is a warning.

::: Danger
This is a warning within a warning.
:::
::::::::::::::::::


References {.allowframebreaks}
=====

-----