中文摘要：

针对高分辨率遥感影像进行树冠提取时所遇到的由各类地物之间的相关性和复杂性带来的地物提取难点,将独立分量分析算法和尺度优化法结合进行树冠提取研究.首先,通过独立分量分析算法优化高分辨率遥感影像,去除地物波谱信息之间的相关性,并将ICA变换得到的特征值作为波段加权的权重;再通过改进的最优尺度计算模型选择最优的分割尺度;最后通过对树冠提取的平均精度评价该改进的计算模型.将本研究方法与单纯尺度优化法实验对比分析,结果表明：本文方法有利于降低＂同谱异物＂和＂同物异谱＂以及树冠连冠现象,提高树冠信息提取的精度,并可有效避免人为确定分割尺度的主观性和低效性.

英文摘要：

Aiming at the difficulty of extracting features from various objects in high resolution remote sensing image, a solution that combines mean method with max-area method was proposed to extract the canopy from an unmanned aerial vehicle （UAV） image. Firstly, independent component analysis was introduced to optimize the high resolution remote sensing image, and the characteristic value obtained by the ICA was used as the weight. Then, improved optimal scaling model was used to choose the optinal image segmentation scale. Finally, the optimal scale calculation model was estimated by the average precision of extraction of canopy. Compared with other methods, the experimental results show that the method in this paper is beneficial to reduce the same spectrum with different objects, the same objects with different spectra and the crown connection phenomenon, enhancing the extraction accuracy for canopy information, and effectively avoiding the subjectivity and inefficiency of the artificial segmentation.