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

  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 2S 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.