Multiband spectro-radiometric satellite image analysis with K-means cluster algorithm

Last update: Mar 16, 2022

Overview

Multi-band Spectro Radiomertric Image Analysis with K-means Cluster Algorithm

Overview

Multi-band Spectro Radiomertric images are images comprising of several channels / bands which hold information on band energy in each pixel.
The most common multi band channels are the RGB (Red Green Blue) channels of the visible light spectrum.

The images used are LANDSAT 8 satellite images and each image consist of three bands, namely: Thermal Infrared, Red and Near infrared bands corresponding to band 10, band 4 and band 5 of LANDSAT 8 satellite imagery with wavelengths of 10.895µm, 0.655µm and 0.865µm respectively.

Each pixel in each bands of each image are used to compute three features namely: NDVI (Normalized Differential Vegetative Index), PV (Portion of Vegetation) and LST (Land Surface Temperature).

The K-means cluster algorithm is initialized and the "number of clusters" hyper-parameter is set to 60. The algorithm is then trained on the extracted features and forms 60 different clusters represented by each of the 60 centroids.

These centroids are stored in the "ouput" folder and will be futher studied to learn what NDVI, PV and LST combinations a geograhical location might need to have for the occurence and spread of wild fire to be highly probable.

Features

NDVI (Normalized Differential Vegetative Index):

The Normalized Differential Vegetative Index is a metric for checking the presence and health of a vegetation in a given region.
It is basically how much RED light energy from the visible light spectrum is absorbed by the plant and how much NIR (near-infrared rays) it emmits.
Healthy vegetation absorbs red-light energy to fuel photosynthesis and create chlorophyll, and a plant with more chlorophyll will reflect more near-infrared energy than an unhealthy plant.
The NDVI ranges from -1 to 1, -1 corresponds to a very unhealthy plant and 1 corresponds to a very healthy plant.

The mathematical expression for NDVI is:
NDVI = (NIR - RED) / (NIR + RED)

PV (Portion of Vegetation):

Portion of Vegetation is the ratio of the vertical projection area of vegetation on the ground to the total vegetation area

The mathematical expression for PV is:
PV = (NDVI - NDVImin) / (NDVImin + NDVImax)
NDVImin is the minimum NDVI value a pixel holds in a single image
NDVImin is the maximum NDVI value a pixel holds in a single image

LST (Land Surface Temperature):

Land Surface Temperature is the radiative temperature / intensity of the land surface

The mathematical expression for LST is:
LST = BT / ( 1 + ( ( kn * BT / p ) * np.log(E) ) )

BT is brighness Temperature in celcius and is mathematically expressed as:
BT = (K2 / np.log( ( K1 / TOA ) + 1 )) - 273.15
where K1 and K2 are landsat 8 constants 774.8853 and 1321.0789 respectively

TOA (Top of Atmosphere) Reflectance is a unitless measurement which provides the ratio of radiation reflected to the incident solar radiation on a given surface.
It is mathematically expressed as:
TOA = ML * TIR + Al
where ML and Al are landsat 8 constants 3.42E-4 and 0.1 respectively.

p is mathematically expressed as:
p = hc/A
where h, c and a are plank's constant, speed of light and boltzmann constant respectively

E is emissivity of the land surface and is mathematically expressed as:
( Ev * PV * Rv ) + ( Es * ( 1 - PV ) * Rs ) + C
where:
Ev (Vegitation Emissivity) of location = 0.986
Es (Soil Emissivity) of location = 0.973
C (topography factor) of location = 0.0001
Rv =(0.92762 + (0.07033PV))
Rs=(0.99782 + (0.05362PV))

Dependencies

Rasterio

Numpy

Pandas

Sklearn

Pickle

Setup
clone the repository and download the 'requirement.txt' files, then open terminal in the working directory and type 'pip install -r requirements.txt' to install all the requirements for this project.

Multiband spectro-radiometric satellite image analysis with K-means cluster algorithm

Related tags

Overview

Multi-band Spectro Radiomertric Image Analysis with K-means Cluster Algorithm

Overview

Features

NDVI (Normalized Differential Vegetative Index):

PV (Portion of Vegetation):

LST (Land Surface Temperature):

Dependencies

Setup

Owner

Chibueze Henry

Official code of ICCV2021 paper "Residual Attention: A Simple but Effective Method for Multi-Label Recognition"

This program generates a random 12 digit/character password (upper and lowercase) and stores it in a file along with your username and app/website.

Prometheus Exporter for data scraped from datenplattform.darmstadt.de

PyG (PyTorch Geometric) - A library built upon PyTorch to easily write and train Graph Neural Networks (GNNs)

Uncertainty Estimation via Response Scaling for Pseudo-mask Noise Mitigation in Weakly-supervised Semantic Segmentation

Implementation of neural class expression synthesizers

Implementation of Convolutional enhanced image Transformer

Python project to take sound as input and output as RGB + Brightness values suitable for DMX

Use unsupervised and supervised learning to predict stocks

Data reduction pipeline for KOALA on the AAT.

Contrastive Language-Image Pretraining

Geneva is an artificial intelligence tool that defeats censorship by exploiting bugs in censors

TensorFlow implementation of Elastic Weight Consolidation

This implementation contains the application of GPlearn's symbolic transformer on a commodity futures sector of the financial market.

Official repository for "Orthogonal Projection Loss" (ICCV'21)

AI-generated-characters for Learning and Wellbeing

Cross-lingual Transfer for Speech Processing using Acoustic Language Similarity

A Data Annotation Tool for Semantic Segmentation, Object Detection and Lane Line Detection.(In Development Stage)

Official PyTorch implementation of the paper: Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting.

QuakeLabeler is a Python package to create and manage your seismic training data, processes, and visualization in a single place — so you can focus on building the next big thing.