A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Environment, Resources and Development

Thesis (M. Sc.) - Asian Institute of Technology, 1998

This study deals with the future high spatial resolution satellite data, which are supposed to provide very accurate spatial information for various fields of interests. Number of crowns in a particular forest area is also very important for commercial purposes and environmental manner. A new algorithm uses comparison of the mean and standard deviation values of high spatial resolution image and specified identifying window for tree crown identification has been developed. The crowns covered area is the another identification method and combining the results, create a new image and clean the combined image to get the noise free result. The finalising steps upgrade the accuracy of image to get the better information. By using this algorithm, tree crowns were identified and accuracy was checked by comparing with visual identified result and overall accuracy is acceptable. The level of accuracy is depend on the resolution and 1 meter image is the best for crown size smaller than diameter in 7 meter. This algorithm can be used for tropical forest tree crown identification and the expectancy of this algorithm is to estimate the forest volume in future by further development with pattern recognition steps.