Objective  Crown width is an important characteristic factor of canopy structure, which directly affects the productivity and vitality of trees. The forest canopy density is one of the important indexes to reflect forest canopy structure and density and to evaluate forest management and logging intensity. UAV has the advantages of easily getting high-resolution remote sensing images with high precision and low cost. Studying the method of extracting canopy parameters using UAV images is of great significance for improving the accuracy and efficiency of forest resource inventory and monitoring.

  Method  Taking Chinese fir plantation in Jiangle Forest Farm of Fujian Province, eastern China as the research object, using the quadrotor UAV CCD image data as the data source, based on the object-oriented classification method, the canopy parameters of the Chinese fir plantation were extracted from the UAV images. Then the canopy objects were grouped into one group according to the segmentation results of the images, and the number of raster pixels of each canopy object was counted to calculate the canopy width area and canopy density.

  Result  The object-oriented classification effectively extracted the crown of high canopy density stand. When the segmentation scale was 70, the segmentation of single tree had the best effect. Some single trees were lost during the segmentation process because of over-segmentation and under-segmentation. After completing the segmentation, optimizing the feature space of the segmented object and selecting appropriate classification features, finally the study area was divided into two types: canopy and forest gap. By counting the number of grid points of each object, the calculated stand factors included canopy density and crown area. With the measured data on the ground as reference, the crown area extraction accuracy was 0.829 1, and the forest canopy density measurement accuracy was 0.973 1.

  Conclusion  The results show that the canopy parameter extraction based on high-resolution image of UAV is also applicable in high-canopy closed forest stands, which can effectively improve the efficiency and accuracy of forest resource survey.