基于改进爬峰法高分辨率遥感影像分割的树冠提取

对现有爬峰法利用高分辨率遥感影像分割进行树冠提取时存在的问题进行直方图压缩和基于类的二次合并两方面的改进,并将改进后的爬峰法在Matlab平台上模拟实现。以Quick Bird影像为基础数据源提取单木树冠,分析树冠面积提取精度,验证改进爬峰法对高分遥感影像树冠分割的可靠性。研究结果表明:实验样本精度均达到85%以上,与传统目视解译精度比较相差较少,满足应用需求。

Tree Crown Extraction Based on Segmentation of High-resolution Remote Sensing Image Improved Peak-climbing Algorithm

The peak-climbing algorithm in two aspects histogram compression and the two merging based on class when it is applied to high-resolution image segmentation to achieve the tree crown extraction improved peak-climbing algorithm was simulation with programing on Matlab. In order to verify the reliability of the peak-climbing algorithm on high-resolution image tree crown segmentation, QuickBird image to extract individual tree crown and analyze the precision of its area. The study result that the test sample accuracy could more than 85% using the improved and no much differences comparing with the visual interpretation. Thus this improved peak climbing algorithm meets the application requirements