Innovations and Best Practices in Random Forest Image Classification

Dear Colleagues,

Image classification with random forest (RF) is now widely practised within the broad discipline of remote sensing. The algorithm has been implemented in most commercial and open-source remote sensing software and platforms, and has been repeatedly shown to perform similar to or better than many competing machine learning classifiers while simultaneously providing valuable information on input variable importances.  Studies over the past five years have highlighted some valuable principles and guidelines for RF image classification, including addressing the multicollinearity of input variables, the impact of spatial autocorrelation and representativeness of training data on classification results, the proper interpretation of out-of-bag accuracy, and the importance of proper cross-validation. Despite the now ubiquitous use of the algorithm within the remote sensing community, there is a need for a wider-spread adoption of best practices in RF image classification. Opportunities also exist to adapt, test, and benchmark new RF-based image classification approaches in an effort to advance current best practices.

For this Special Issue, we are soliciting new contributions that address these needs by adopting a specific focus on the development and testing of best practices for RF implementation in remote sensing, and/or innovations and case-studies that will serve as guiding examples for researchers, practitioners, and remote sensing software developers. We particularly encourage studies that focus on methods testing such as how to optimize overall and/or class-specific accuracies through improved supervised training (e.g., for rare classes), variable reduction workflows for high-dimensional problems to reduce computational processing requirements and/or model robustness, proper determination of variable importance in the presence of correlated input variables, or implementation of unsupervised RF classification workflows for more efficient/accurate image classifications.  While the main focus of this Special Issue will be on methods testing and benchmarking, site-specific case-studies that informatively demonstrate success or failure of one or more example(s) of best practices in RF image classification are also invited.

Dr. Murray Richardson
Dr. Koreen Millard
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Remote Sensing is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2700 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• Random forest
• Machine learning
• Image classification