Vision-based Shadow-aided Tree Crown Detection and Classification Algorithm using Imagery from an Unmanned Airborne Vehicle

摘要

We propose a novel algorithm for tree crown detection and classification in aerial images. The algorithm utilises statistical learning and computer vision techniques to identify various types of woody weeds in a natural landscape with aid of their shadow. In remote sensing context, colour spectrum and texture features are commonly used as cues for classification, the shape property of the target is not explored as often due to the image resolution constrain; however due to our relative lower altitude flights the shape of the target and the corresponding shadow could be resolved and is used to provide extra shape feature for detection. The proposed algorithm is divided into detection and classification stages, the purpose of the detection stage is to identify the interest points correspond to high likelihood of tree crown existence, and therefore reduces the search space of the classification stage from the entire image data set to significantly smaller region of interests. The detection stage is further divided into two stages, the first stage segments the image using colour and texture features, the second stage utilises template matching using shape information related to projected shadow of the woody weed, relying on information about the time of day, sun angle and UAV pose information collected by the onboard navigation system. The classification stage uses supervise learning training on the features collection from the region of interests. We present experimental results of the approach using colour vision data collected from a UAV during operations in the Julia Creek in July 2009 and August 2010.