Detecting reed canary grass (Phalaris arundinacea L.) patches from UAV‐based digital surface model images—A case study in a timothy (Phleum pretense L.) meadow field

Abstract

AbstractAccurate determination of the weed ratio in artificial meadows is critical for efficient pasture renovation. Reed canary grass (Phalaris arundinacea L., RCG) is treated as a troublesome grass in the Hokkaido region of Japan because of its low feed quality and poor palatability in dairy farming. In the present study, we examined a method of identifying the dominant area of RCG in timothy (Phleum pretense L.) meadows by applying the Canny method to unmanned aerial vehicle (UAV)‐based digital surface model (DSM) images. Comparing the actual RCG patches observed in a field survey (50 m quadrats × 4 places) with the predicted patches, the pixel‐based recall and F value were 0.90 and 0.83, respectively. These results demonstrated that the area of RCG can be detected using a simple method without supervised data or deep learning. This study is expected to be utilized in a wide variety of applications using relative height differences.