Automatic Recognition And Positioning Of The Withered Pine Based On UAV Images

Pine wood nematode disease is a major disease badly attacking various pine trees,and has led to large ecological and economic lose in China.It is important to monitor and evaluate field damage for control of the disease.In the past,the dead pine trees were counted and recorded by means of artificial and visual method,which was time-consuming and low efficient.In this paper,the unmanned aerial vehicle(UAV)was selected to investigate the dead pine trees in the field,and based on image processing and geographic information technology,the methods of automatic identification and positioning of the dead pine trees were studied and applied.The aerial images were obtained using a vertical take-off and landing fixed wing UAV in disease area.The pix4dmapper software was used to preprocess the aerial images,and then the warp and weft information was brought into the images,and the geotif images with the latitude and longitude information were formed.Based on the special color characteristics of dead pine trees,three color space models including RGB,HSV and LAB were applied and compared to determine the location of dead pine trees in aerial images.The threshold range of R,G and B in the RGB space were set to identify the dead trees.And then the RGB images were converted into HSV images and the threshold value of H channel was set to identify dead pine trees.At the same time,the RGB images were also converted to LAB images to identify dead pine trees.The results showed that the LAB spatial model was more effective than other two models.Based on the geographic information in the aerial image,the location information of dead trees which were identified by LAB space model,the pixel coordinates of dead trees were evaluated.The positions of dead trees were represented by the pixel coordinates of this point.The location information of these pixels was acquired by using geographic information technology,and then the automatic location and counting of dead trees was realized.Field verification showed that the accuracy rate of automatic identification of dead pine trees by image processing technology reached 80%,and the latitude and longitude was about 10 m error.It was confirmed that UAV and image processing technology could be used to construct effectively model to identify and locate pine dead pine trees in the field.It has a wide application prospect.