Automated detection of selective logging using SmallSat imagery

We propose an automated processing workflow to detect and classify changes in bi-temporal very high resolution (VHR) SmallSat imagery. The workflow consists of two pre-processing steps, which are an image registration method with a cross-correlation approach, and a radiometric normalization based on regression of automatically detected invariant pixels. The detection of selective logging is performed using the Iteratively Reweighted Multivariate Alteration Detection (IR-MAD), which calculates a probability of change between pair of images. The different types of changes are then discriminated using a clustering algorithm on the original spectral values as well as on the resulting values of different image transformations. The developed approach was applied on SkySat images over Rennell Island (Solomon Islands) for detecting selective logging and posterior regrowth within a conservation area. VHR SmallSat images open the possibility to successfully detect at-crown level changes such as selective logging in the tropics, where common satellites mostly take non-valid images due to cloud coverage, while the high temporal resolution of SmallSats can allow a sufficient number of valid observations. Results show high accuracy and fast processing in detecting changes related to selective logging operations, which can be relevant for processing large amounts of high-resolution data obtained on a daily basis.

DESCALS FERRANDO Adria;  SZANTOI Zoltan;  BECK Pieter;  BRINK Andreas;  STROBL Peter; 

2017-11-22

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC

JRC103858

1545-598X,   

http://ieeexplore.ieee.org/document/8107672/,    https://publications.jrc.ec.europa.eu/repository/handle/JRC103858,   

10.1109/LGRS.2017.2720841,