基于智能手机单目视觉的多株立木高度提取方法

Height extraction method of multiple standing trees based on monocular vision of smart phones

【目的】提出一种智能手机单目视觉下的多株立木高度提取方法。【方法】该方法以智能手机为采集设备,利用Graph Cut算法对输入的立木图像进行分割定位,实现单幅图像中多株立木轮廓的自动获取;再通过智能手机相机对摄像头进行标定,从而基于几何相似法获取智能手机相机图像的深度信息。在不同角度下拍摄标靶,进行深度提取模型的精度优化,进而确定信息提取的最优方位。同时,结合高精度陀螺仪获取相机俯视角,根据提取的深度信息和相机俯视角实现非接触条件下的多株立木高度测量。【结果】使用型号为MI 2S的小米智能手机为试验设备,在本方法中的立木高度测量模型具有良好的稳定性,并且试验中最高相对误差为2.45%,树高测量精度可达97.55%。【结论】基于智能手机单目视觉下的立木高度提取方法精确度高、操作简便,能够有效满足国家森林资源二类调查中对于树高测量精度的要求。

[Objective]This paper aims to propose a method for extracting height of multiple standing trees based on monocular vision of smart phones.[Method]First of all,with smart phone as the acquisition device,this method segmented and located the input images of standing trees by Graph Cut algorithm,so as to realize the automatic outline acquisition of multiple standing trees in a single image.Then,the camera was calibrated by the camera of smart phones to acquire the depth of camera image of smart phones based on the geometry analogue method.Next,targets were taken from different angles to optimize the accuracy of depth extraction model and further determine the optimal orientation of information extraction.Meanwhile,combined with high-precision gyroscope,the depression angle of camera can be acquired,and the height measurement of multiple standing trees under non-contact condition can be realized according to the extracted depth information and depression angle of camera.[Result]In this method,the height measurement model of standing trees had good stability with smart phone MI 2 S as the experimental equipment.Moreover,the maximum relative error was 2.45%and the accuracy of tree height measurement can reach 97.55%.[Conclusion]The height extraction method of multiple standing trees based on monocular vision of smart phones has the characteristics of high accuracy and easy operation,which can effectively meet the accuracy requirements of tree height measurement in national forest resource survey II.