Determining tree species has involved a lot of studies in the past decades. This is because of the availability of certain tree species to produce a high market value type of wood or that it can be use for meditational purposes. In the old days, conventional method of survey needs to be conducted to identify tree species. This type of method is rather hard to inaccessible area and time consuming. With the advancement of the satellite imaging technology, inaccessible area can be visualize from an satellite image. Worldview-2 image data is one of the proofs of the remote sensing evolving technology that can be used in a lot and varieties of studies. The available of the near infrared band makes it very useful in vegetation analysis. Based on this study that has been conducted to eight sample leaves taken from the Bukit Nanas Forest Reserve, Kuala Lumpur, spectral signature extraction was successfully conducted on the near infrared bandwidth due to its sensitivity towards vegetation. The different of about 5% between the signature extracted from the Spectra Vista GER 1500 and Worldview-2 data indicates that this image can be use successfully for vegetation analysis. However, it is very questionable whether high resolution can successfully delineate tree crown of a tree in a natural forest due to the complexity of the characteristics of the crown which is overlapping each other. This complexity is also a challenge for this study as to relate the spectral signature extracted with the accuracy of the segmentation and classification result. Furthermore it also will give some ideas to researchers as to figure what type of improvements can be made towards this study especially for natural forest.