The Study And Application Of Crop Moisture Content Detection Based On Image Technology

Crop water detection,one of the "four cases" monitoring of agriculture,has been widely applied into crop drought warning,irrigation management and yield prediction etc.It has become an important research topic that detecting the crop moisture content fast and inoffensively.Generally,it uses machine vision technology of crop moisture with the crop leaf or canopy image as the research object since the rapid development of the image technology and Internet technology.In order to improve the precision and practicability of crop moisture content detection,the methods of illumination enhancement,image segmentation,and feature selection and detection model construction were studied in this paper.The experiments and applications were carried out with maize blades and wheat canopy images as investigate object.The main contents and results are as follows:(1)Research on illumination enhancement and segmentation of crop image.The combined algorithm of homomorphic filtering and Retinex was proposed to preprocess the enhancement of illumination that eliminates the adverse effects of natural light condition.The comparison of K-means clustering segmentation,two-dimensional maximum entropy segmentation and segmentation of the color characteristics were conducted.The color feature segmentation method was used to segment wheat canopy image.It maximized the retention of crop information,and removed soil and withered leaves in background effectively.(2)Research on feature extraction,filtering and model checking.In order to extracting features of color,texture and shape on the existing foundation,features in high correlation with moisture content were selected by correlation analysis and hypothesis testing etc.The principal component analysis and partial least squares regression were then adopted to establish the moisture content detection model of high precision.(3)Application of corn leaf moisture content detection.The crop leaf moisture content detection system based on smart phone-server mode was proposed and implemented.The results on corn leaf demonstrated that this method can obtain image conveniently,get the results which has the acceptable detection error in real time.Moreover,the update and extension of algorithm was transparent to the user which was very useful for farmers and agricultural technical personnel.(4)Application of winter wheat canopy moisture content detection.The pictures of canopy winter wheat photographed by digital camera were used to detect the canopy moisture content.Experimental results of two winter wheat varieties,namely “Huai-mai30”and“Yan-nong 19”,showed that the mean relative error and variance of the proposed method were 1.290% and 1.053.Respectively,there were no obvious differences between the two varieties.This method can be extended to the field of camera and agricultural unmanned aerial vehicles and other detection occasions.