Image Processing And Root Diameter Estimation Of Tree Root Based On Ground Penetrating Radar

The roots are important organs of tree growth on the ground,but the research of the root system is lagging behind compared to the aboveground part of the trees.The reason is that the root grows under the ground and the observation is very inconvenient.Traditional root detection methods are not only time consuming and labor intensive but also destroying the soil environment and tree roots,so it is very important to find a more reliable and non-destructive method.In the research process this paper received the strong support of national natural science fund project(31270757/C201337)and the Guangzhou Institute of Landscape Gardening.Based on the non-destructive detection,this paper uses the ground penetrating radar to detect the root system of trees,The contents of the study are as follows:(1)Although the validity of the application of Ground Penetrating Radar to forest root detection has been confirmed,but the root structure complexity and environmental heterogeneity are strong,the application of GPR in the forest root detection is not too long,the Interpretation experience of root detection data is not yet mature.In this paper,combined with the actual array of GprMaxV2.0 forward simulation of the environment and the root system to prove the effectiveness of this simulation,to avoid the drawbacks that repeated use of Ground Penetrating Radarto tentative detection.And simulate the 0?-180?staggered root reflection,and analyze the reasons for the formation of echo.The method of geostationary radar forward modeling is helpful to the recognition of the reflection profile to the target,and provide the basis for the interpretation of the detection data.(2)The echoes of Ground Penetrating Radar are mainly processed,including the reduction of Gaussian noise,noise reduction and clutter suppression by comparing different methods,and finally selecting non-local mean filtering to perform noise reduction and de-averaging filtering.then use the different operator edge detection to extract the target information,select the best effect canny operator to do the edge extraction;and the traditional Ground Penetrating Radar image preprocessing has been improved,before the edge detection,it need to conform the threshold based on the image Segmentation,removal of the background,thus greatly reducing the amount of follow-up processing,combined with morphological and edge refinement to extract hyperbolic.(3)By using the symmetry of hyperbola,the vertex position of all hyperbolic echoes in the Ground Penetrating Radar image is extracted,and finally the root radius estimation is carried out based on the improved Hough transform equation and the optimal estimation.Fitting,comparing the difference from the fitting diagram and the result of the multiple sets of radius estimates with the same theoretical value,analyzing the range of the error and explaining the feasibility of this method.In this paper,we use the proposed non-local mean filtering and mean filtering to perform clutter suppression,and perform threshold-based image segmentation after interference suppression.The Canny operator edge detection and image morphological method are used to carry out ground-based radar image processing,The hyperbolic curve is used to extract the hyperbolic vertex position,which is more favorable to reduce the computational complexity of the Hough transform.The root radius is estimated by the Hough transform and the optimal estimation principle.The root of the project is taken as the root of the project and the root biomass estimation is made.