• P-ISSN 0974-6846 E-ISSN 0974-5645

Indian Journal of Science and Technology


Indian Journal of Science and Technology

Year: 2016, Volume: 9, Issue: Special Issue 1, Pages: 1-6

Original Article

Principal Component Analysis based Assessment of Trees Outside Forests in Satellite Images


This Paper Presents, an automated Matrix Laboratory (MATLAB) based procedure for assessing the number of tree crowns in a High Resolution Satellite Image based on Principal Component Analysis (PCA) algorithm. This automated method comes into picture to reduce the manual power, time and money. PCA involves in a mathematical approach in usage of orthogonal transforms for converting a given set of correlated observed values into a set of modified grey levels of linearly uncorrelated variables and these newly generated values are called principal components. Then a bounded set of thresholding values are used to identify the trees, Laplacian of Gaussian (LOG) operator is used to find and mark the boundaries of trees to extract the tree crowns available in the test site. Finally, count the number of tree crowns by using available morphological open and close operations with the help of connectivity methods. The results obtained are matched with manual approximated count and the accuracy of algorithm is done by testing on 20 different sets of test areas the accuracy achieved is 86.6 and it is better than NDVI and Watershed segmentations. Further improvement is possible when this method is combined with NDVI.
Keywords: Blob Detection, Counting Trees, LOG, PCA, Remote Sensing, Tree Crown Delineation 


Subscribe now for latest articles and news.