AUTOMATED DELINEATION OF INDIVIDUAL TREE CROWNS THROUGH MORPHOLOGICAL ANALYSIS OF QUICKBIRD IMAGERY

The sustainable management of sparse forests requires detail information for a variety of attributes. Extracting the required information by means of traditional field based methods is time-consuming and costly, whilst application of low-medium resolution satellite imagery will not yield accurate and reliable results. It is, therefore, critically important to investigate the newadvanced spaceborne remote sensing technologies that permit to extract detail information at tree level. In this study, the capabilities of QuickBird imagery examined with the data acquired over the Oak forest site in the South West of Iran. Image fusion was used as preprocessing step. The effects of background were mitigated by using an NDVI mask. The Sobel operator was applied on the masked image to obtain a 2-D spatial gradient magnitude. Binary mathematical morphology was employed to determine the location of potential tree tops and impose them on the gradient image. Individual tree crowns were delineated usingMarker Controlled Watershed segmentation. To evaluate delineation results crown by crown accuracy assessment was performed. The results of this study showed that individual trees or clusters of trees can be automatically delineated with high accuracy in sparse forests

کلیدواژه ها:

Morphological image analysis ، QuickBird ، Individual tree crown delineation ، Marker Controlled Watershed Segmentation ، Edge detection ، Zagros forest