森林调查是森林经理学的重要组成部分，传统的人工野外观测已逐渐无法满足现代化的森林调查的需要。因此，借助主、被动多元遥感数据准确提取森林中的单木树冠并进行森林参数估测，可以很好地用来估测林分生物量和碳储量、监测树木的长势、预防树木病虫害情况、模拟能量传输等，对于现代化精准森林调查具有重要意义。本项目以凉水自然保护区为研究对象，低密度机载激光雷达点云、航空相片及高空间分辨率卫星图像WorldView-2为数据源，对郁闭度较高的原始针叶混交林进行单木树冠提取，并将提取结果应用到精准森林调查中。本项目将解决利用低密度激光点云与高空间分辨率被动遥感结合提取单木树冠问题，并提出基于Agent的区域生长法，将单边及双边竞争机制加入到单木树冠提取算法中，结合单木生长规律和遥感信息技术合理处理重叠树冠，并将提取结果应用到森林调查因子的估测中，在遥感和林学领域均有重要学术意义及应用价值。

Based on low-density airborne laser scanner (ALS) data, orthoimages and high spatial resolution satellite image (WorldView-2), this project incorporates tree competition into individual tree crown delineation for original spruce-fir-Korean pine forests in Xiaoxing'an Mountain of northeastern China, and implements automation of algorithms and accuracy assessment. The study mainly includes: (1) integrating multiple remotely sensed data into three dimensional data source that represents horizontal and vertical structures of trees; (2) applying local maximum algorithm to detect individual tree location based on variable three dimensional window; (3) proposing agent-based region growing algorithm for crown boundary delineation; (4) establishing a set of accuracy assessment system. This study would solve the problems of applying low-density ALS data and high spatial resolution passive remotely sensed data to delineate individual trees, and that of handling adjacent crowns using agent-based region growing algorithm, which incorporates one- and two-way competition mechanisms into delineation algorithm for delineation accuracy improvement. This project would firstly apply a novel individual tree crown delineation method that incorporates tree growth rules into remote sensing technologies in original northeastern forests of China. It would supply basis for modern forest inventory, and also provide academic significances and application values for both remote sensing and forestry fields.