Remote sensing
Masoud Taefi Feijani; Saeed Azadnejad; Masoud Moradi
Volume 14, Issue 2 , June 2021, , Pages 27-36
Abstract
Awareness of the trend of forest canopy density classification requires an operational exact model for forest crown classification. The preliminary challenge is the separation of the forest crown from other non-warlike vegetation coverings. In the following, previous attempts to improve the performance ...
Read More
Awareness of the trend of forest canopy density classification requires an operational exact model for forest crown classification. The preliminary challenge is the separation of the forest crown from other non-warlike vegetation coverings. In the following, previous attempts to improve the performance of the FCD model, in this study, by adding the FCC index and the kernel, improved the average performance of the FCD model. The crown classification of Hyrcanian forests based on images of 1396 Landsat 8 was selected for implementation, evaluation, validation and analysis of the results. Improving the accuracy of the model is entirely sensible and even manual interpretation confirm it. The statistical analysis of the results also indicates a 10% and 24% increase in overall accuracy and kappa coefficient of the improved model compared to the initial model. Specifically, the accuracy of these two classes in the results of the improved model is about 13% and 7%, respectively.
Masoud Taefi Feijani; Abbas Alimohammadi Sarab; Mohammad Javad Valadan Zoej
Volume 12, Issue 3 , September 2019, , Pages 31-42
Abstract
Forest Canopy Density Mapper is a method based on spectral indexes integration in forest canopy density classification. In this paper, a data integration procedure is used to improve the result. In this respect, SFIM method and spectral response algorithm is utilized without a bad effect on the spectral ...
Read More
Forest Canopy Density Mapper is a method based on spectral indexes integration in forest canopy density classification. In this paper, a data integration procedure is used to improve the result. In this respect, SFIM method and spectral response algorithm is utilized without a bad effect on the spectral and radiometric properties of bands. In the following, Landsat images of Hyrcanian forests in the north of Iran were used to implement the conventional and improved methods. Also, the ground measurements including grass-land, thin forest, semi-dense forest and forest is utilized for evaluation. The result shows that the forest canopy density model is inefficient in the thin and semi-dense forests. Alternatively, the results in the dense forest and grass land is reliable. Additionally, the improvement of the proposed method in these two areas is clearly seen. It seems that a high resolution image should be used to improve the accuracy of the forest density classification in the semi-dense and thin forests.
