Forest Change Detection Using Remote Sensing Data

Examination of actual forest vegetation boundary has great economic importance. A positive examination result can serve as evidence of a violation of the rules for forest utilization. The use of remote sensing (RS) data can reduce the complexity of such a task and speed up the detection of the violations by using a pair of RS images collected at different dates. The article compares two algorithms of searching for changes in the forest vegetation boundary. The first is an elementwise classification algorithm using a tree histogram, whereas the second is a structural change detection algorithm based on superpixel segmentation and a modified Pytiev’s projector. The algorithms do not require atmospheric correction of images, which makes them attractive for practical use. The article also provides the optimal values of parameters of the algorithms and investigates the influence of various characteristics of RS data such as source, imagery date, band combination, and spatial resolution on the forest boundary change detection accuracy.