Histogram Matching Based Mosaicking of SPOT 6/7 Satellite Images

Abstract

This research aims to evaluate a histogram matching based custom mosaicking process on different land cover land use characteristics using multi-temporal satellite images. Three different study regions located in different parts of Turkey are selected and SPOT 6/7 satellite images belonging to these regions are used in analysis. Reference histogram data obtained from 2018 dated Sentinel 2 mosaic of Turkey. This dataset is used for histogram matching after conducting post process to enhance the visual representation of the study region. Feathering effort is operated on the target images after histogram matching step in order to eliminate dissimilarities between geometric shapes and introduce seamless transition in overlapping areas. Accuracy assessment is performed both by visual interpretation and by calculating the weighted average of differences between reference and matched histograms. Results proved that, histogram matching based method proposed in this study provides practical and efficient mosaicking approach, which can be used for producing high-resolution and large coverage satellite image dataset.