Phenotypic Measurement Of Plant Leaf Based On Kinect Sensor

The spatial morphological structure of plant leaves is a carrier of the characteristics of plant life,such as plant gene expression,resource acquisition and reproductive reproduction.The phenotypic parameter is an important indicator of plant growth status,carbon storage capacity and carbon sink capacity.In a large number of phenotypic parameters,Leaf Angle indicates the spatial orientation of plant leaves;the Leaf Area directly reflects plant productivity.At present,in the measurement of leaf angle,usually with the protractor and leaf angle method;in the measurement of leaf area,usually by weighing,leaf area method,digital image processing method,and these methods will measure the plant leaf tissue Resulting in a certain degree of damage,and there is cumbersome work,expensive equipment and other shortcomings.In view of these problems,the three-dimensional reconstruction of the leaves was carried out by low cost Kinect sensor,and the leaf area and leaf angle of the plants were measured.The main contributions of the paper include:(1)In view of the current three-dimensional reconstruction of three-dimensional laser scanner and ground-based laser radar as a data acquisition device,but its expensive,this paper uses low-cost and easy to carry Kinect sensor as a collection device to obtain plant point cloud data.For the measurement of plant leaf area is usually used leaf area meter,its high precision,but when the blade area is larger than the range can not be measured,and weighing method and grid method to measure leaf area when the operation is simple but time-consuming and labor problems,this paper It is proposed to measure the leaf area of different sizes after three-dimensional reconstruction with Kinect sensor.(2)In this paper,the method of OBB bounding box is proposed to establish the minimum extroverted rectangular cube method for the three-dimensional point cloud data of plant leaves in order to measure the leaf inclination of plants by traditional method such as protractor or leaf inclinometer.Experiments show that the accuracy of the cloud after the second division rate of about 94%,greatly improving the measurement efficiency.(3)In this paper,we support the vector vector regression(SVR)algorithm,The input training set is regression,and the actual leaf area model is estimated.Finally,the validity of the regression model is verified by the test set.